id
int64
0
458k
file_name
stringlengths
4
119
file_path
stringlengths
14
227
content
stringlengths
24
9.96M
size
int64
24
9.96M
language
stringclasses
1 value
extension
stringclasses
14 values
total_lines
int64
1
219k
avg_line_length
float64
2.52
4.63M
max_line_length
int64
5
9.91M
alphanum_fraction
float64
0
1
repo_name
stringlengths
7
101
repo_stars
int64
100
139k
repo_forks
int64
0
26.4k
repo_open_issues
int64
0
2.27k
repo_license
stringclasses
12 values
repo_extraction_date
stringclasses
433 values
25,000
generate_self_signed.py
truenas_middleware/src/middlewared/middlewared/plugins/crypto_/generate_self_signed.py
import typing from cryptography import x509 from cryptography.hazmat.backends import default_backend from cryptography.hazmat.primitives import hashes, serialization from .generate_utils import generate_builder, normalize_san from .key_utils import generate_private_key from .utils import DEFAULT_LIFETIME_DAYS def generate_self_signed_certificate() -> typing.Tuple[str, str]: cert = generate_builder({ 'crypto_subject_name': { 'country_name': 'US', 'organization_name': 'iXsystems', 'common_name': 'localhost', 'email_address': 'info@ixsystems.com', 'state_or_province_name': 'Tennessee', 'locality_name': 'Maryville', }, 'lifetime': DEFAULT_LIFETIME_DAYS, 'san': normalize_san(['localhost']) }) key = generate_private_key({ 'serialize': False, 'key_length': 2048, 'type': 'RSA' }) cert = cert.public_key( key.public_key() ).add_extension( x509.ExtendedKeyUsage([x509.oid.ExtendedKeyUsageOID.SERVER_AUTH]), False ).sign( key, hashes.SHA256(), default_backend() ) return ( cert.public_bytes(serialization.Encoding.PEM).decode(), key.private_bytes( encoding=serialization.Encoding.PEM, format=serialization.PrivateFormat.PKCS8, encryption_algorithm=serialization.NoEncryption() ).decode() )
1,448
Python
.py
40
28.75
80
0.649073
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,001
attachments.py
truenas_middleware/src/middlewared/middlewared/plugins/crypto_/attachments.py
from middlewared.service import private, Service class CertificateService(Service): class Config: cli_namespace = 'system.certificate' def __init__(self, *args, **kwargs): super(CertificateService, self).__init__(*args, **kwargs) self.delegates = [] @private async def register_attachment_delegate(self, delegate): self.delegates.append(delegate) @private async def in_use_attachments(self, cert_id): return [delegate for delegate in self.delegates if await delegate.state(cert_id)] @private async def get_attachments(self, cert_id): return list(filter(bool, [await delegate.consuming_cert_human_output(cert_id) for delegate in self.delegates])) @private async def redeploy_cert_attachments(self, cert_id): for delegate in await self.in_use_attachments(cert_id): await delegate.redeploy(cert_id)
912
Python
.py
20
38.8
119
0.702489
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,002
crypto_key.py
truenas_middleware/src/middlewared/middlewared/plugins/crypto_/crypto_key.py
from middlewared.schema import accepts, Bool, Dict, Int, List, Ref, Str from middlewared.service import Service from middlewared.validators import Email from .generate_ca import generate_certificate_authority from .generate_certs import generate_certificate from .generate_self_signed import generate_self_signed_certificate from .generate_utils import normalize_san, sign_csr_with_ca from .utils import EKU_OIDS class CryptoKeyService(Service): class Config: private = True def normalize_san(self, san_list): return normalize_san(san_list) def generate_self_signed_certificate(self): return generate_self_signed_certificate() @accepts( Dict( 'certificate_cert_info', Int('key_length'), Int('serial', required=False, null=True), Int('lifetime', required=True), Str('ca_certificate', required=False, max_length=None), Str('ca_privatekey', required=False, max_length=None), Str('key_type', required=False), Str('ec_curve', required=False), Str('country', required=True), Str('state', required=True), Str('city', required=True), Str('organization', required=True), Str('organizational_unit'), Str('common', null=True), Str('email', validators=[Email()], required=True), Str('digest_algorithm', enum=['SHA1', 'SHA224', 'SHA256', 'SHA384', 'SHA512']), List('san', items=[Str('san')], required=True, empty=False), Dict( 'cert_extensions', Dict( 'BasicConstraints', Bool('ca', default=False), Bool('enabled', default=False), Int('path_length', null=True, default=None), Bool('extension_critical', default=False) ), Dict( 'AuthorityKeyIdentifier', Bool('authority_cert_issuer', default=False), Bool('enabled', default=False), Bool('extension_critical', default=False) ), Dict( 'ExtendedKeyUsage', List('usages', items=[Str('usage', enum=EKU_OIDS)]), Bool('enabled', default=False), Bool('extension_critical', default=False) ), Dict( 'KeyUsage', Bool('enabled', default=False), Bool('digital_signature', default=False), Bool('content_commitment', default=False), Bool('key_encipherment', default=False), Bool('data_encipherment', default=False), Bool('key_agreement', default=False), Bool('key_cert_sign', default=False), Bool('crl_sign', default=False), Bool('encipher_only', default=False), Bool('decipher_only', default=False), Bool('extension_critical', default=False) ), register=True ), register=True ) ) def generate_certificate(self, data): return generate_certificate(data) @accepts(Ref('certificate_cert_info')) def generate_self_signed_ca(self, data): return self.generate_certificate_authority(data) @accepts(Ref('certificate_cert_info')) def generate_certificate_authority(self, data): return generate_certificate_authority(data) @accepts( Dict( 'sign_csr', Str('ca_certificate', required=True, max_length=None), Str('ca_privatekey', required=True, max_length=None), Str('csr', required=True, max_length=None), Str('csr_privatekey', required=True, max_length=None), Int('serial', required=True), Str('digest_algorithm', default='SHA256'), Ref('cert_extensions') ) ) def sign_csr_with_ca(self, data): return sign_csr_with_ca(data)
4,180
Python
.py
96
30.604167
91
0.556946
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,003
common_validation.py
truenas_middleware/src/middlewared/middlewared/plugins/crypto_/common_validation.py
import datetime import re from middlewared.async_validators import validate_country from middlewared.utils.time_utils import utc_now from .utils import RE_CERTIFICATE async def validate_cert_name(middleware, cert_name, datastore, verrors, name): certs = await middleware.call( 'datastore.query', datastore, [('cert_name', '=', cert_name)] ) if certs: verrors.add( name, 'A certificate with this name already exists' ) if cert_name in ("external", "self-signed", "external - signature pending"): verrors.add( name, f'{cert_name} is a reserved internal keyword for Certificate Management' ) reg = re.search(r'^[a-z0-9_\-]+$', cert_name or '', re.I) if not reg: verrors.add( name, 'Use alphanumeric characters, "_" and "-".' ) async def _validate_common_attributes(middleware, data, verrors, schema_name): country = data.get('country') if country: await validate_country(middleware, country, verrors, f'{schema_name}.country') certificate = data.get('certificate') if certificate: matches = RE_CERTIFICATE.findall(certificate) if not matches or not await middleware.call('cryptokey.load_certificate', certificate): verrors.add( f'{schema_name}.certificate', 'Not a valid certificate' ) private_key = data.get('privatekey') passphrase = data.get('passphrase') if private_key: await middleware.call('cryptokey.validate_private_key', private_key, verrors, schema_name, passphrase) signedby = data.get('signedby') if signedby: valid_signing_ca = await middleware.call( 'certificateauthority.query', [ ('certificate', '!=', None), ('privatekey', '!=', None), ('certificate', '!=', ''), ('privatekey', '!=', ''), ('id', '=', signedby) ], ) if not valid_signing_ca: verrors.add( f'{schema_name}.signedby', 'Please provide a valid signing authority' ) csr = data.get('CSR') if csr: if not await middleware.call('cryptokey.load_certificate_request', csr): verrors.add( f'{schema_name}.CSR', 'Please provide a valid CSR' ) csr_id = data.get('csr_id') if csr_id and not await middleware.call('certificate.query', [['id', '=', csr_id], ['CSR', '!=', None]]): verrors.add( f'{schema_name}.csr_id', 'Please provide a valid csr_id which has a valid CSR filed' ) await middleware.call( 'cryptokey.validate_certificate_with_key', certificate, private_key, schema_name, verrors, passphrase ) key_type = data.get('key_type') if key_type: if key_type != 'EC': if not data.get('key_length'): verrors.add( f'{schema_name}.key_length', 'RSA-based keys require an entry in this field.' ) if not data.get('digest_algorithm'): verrors.add( f'{schema_name}.digest_algorithm', 'This field is required.' ) if not verrors and data.get('cert_extensions'): verrors.extend( (await middleware.call('cryptokey.validate_extensions', data['cert_extensions'], schema_name)) ) if lifetime := data.get('lifetime'): try: utc_now() + datetime.timedelta(days=lifetime) except OverflowError: verrors.add( f'{schema_name}.lifetime', 'Lifetime for the certificate is too long.' )
3,878
Python
.py
101
28.019802
110
0.565726
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,004
dhparams.py
truenas_middleware/src/middlewared/middlewared/plugins/crypto_/dhparams.py
import os import subprocess from middlewared.service import job, private, Service DHPARAM_PEM_PATH = '/data/dhparam.pem' class CertificateService(Service): class Config: cli_namespace = 'system.certificate' @private async def dhparam(self): return DHPARAM_PEM_PATH @private @job() def dhparam_setup(self, job): """Generate dhparam.pem if it doesn't exist, or has no data in it""" with open(DHPARAM_PEM_PATH, 'a+') as f: if os.fstat(f.fileno()).st_size == 0: subprocess.run( ['openssl', 'dhparam', '-out', DHPARAM_PEM_PATH, '-rand', '/dev/urandom', '2048'], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, check=True, ) os.fchmod(f.fileno(), 0o600)
817
Python
.py
21
30.142857
102
0.607098
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,005
extensions_utils.py
truenas_middleware/src/middlewared/middlewared/plugins/crypto_/extensions_utils.py
import functools import inspect import typing from cryptography import x509 @functools.cache def extensions() -> dict: # For now we only support the following extensions # We also support SubjectAlternativeName but as we include that natively if the user provides it # we don't expose it to the end user as an extension making the process for the end user easier to # create a certificate/ca as most wouldn't even want to know what extension is or does. # Apart from this we also add subjectKeyIdentifier automatically supported = ['BasicConstraints', 'AuthorityKeyIdentifier', 'ExtendedKeyUsage', 'KeyUsage'] return { attr: inspect.getfullargspec(getattr(x509.extensions, attr).__init__).args[1:] for attr in supported } def get_extension_params( extension: list, cert: typing.Union[x509.CertificateSigningRequestBuilder, x509.CertificateBuilder, None] = None, issuer: typing.Optional[x509.Certificate] = None ) -> list: params = [] if extension[0] == 'BasicConstraints': params = [extension[1].get('ca'), extension[1].get('path_length')] elif extension[0] == 'ExtendedKeyUsage': usages = [] for ext_usage in extension[1].get('usages', []): usages.append(getattr(x509.oid.ExtendedKeyUsageOID, ext_usage)) params = [usages] elif extension[0] == 'KeyUsage': params = [extension[1].get(k, False) for k in extensions()['KeyUsage']] elif extension[0] == 'AuthorityKeyIdentifier': params = [ x509.SubjectKeyIdentifier.from_public_key( issuer.public_key() if issuer else cert._public_key ).digest if cert or issuer else None, None, None ] if extension[1]['authority_cert_issuer'] and cert: params[1:] = [ [x509.DirectoryName(cert._issuer_name)], issuer.serial_number if issuer else cert._serial_number ] return params def add_extensions( cert: typing.Union[x509.CertificateSigningRequestBuilder, x509.CertificateBuilder], extensions_data: dict, key, issuer=None ) -> typing.Union[x509.CertificateSigningRequestBuilder, x509.CertificateBuilder]: # issuer must be a certificate object # By default we add the following if not isinstance(cert, x509.CertificateSigningRequestBuilder): cert = cert.public_key( key.public_key() ).add_extension( x509.SubjectKeyIdentifier.from_public_key(key.public_key()), False ) for extension in filter(lambda v: v[1]['enabled'], extensions_data.items()): klass = getattr(x509.extensions, extension[0]) cert = cert.add_extension( klass(*get_extension_params(extension, cert, issuer)), extension[1].get('extension_critical') or False ) return cert
2,871
Python
.py
62
38.83871
117
0.677754
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,006
cryptokey_crl.py
truenas_middleware/src/middlewared/middlewared/plugins/crypto_/cryptokey_crl.py
import datetime from cryptography import x509 from cryptography.x509.oid import NameOID from cryptography.hazmat.backends import default_backend from cryptography.hazmat.primitives import serialization from middlewared.service import Service from middlewared.utils.time_utils import utc_now from .load_utils import load_private_key from .key_utils import retrieve_signing_algorithm from .utils import CERT_BACKEND_MAPPINGS class CryptoKeyService(Service): class Config: private = True def generate_crl(self, ca, certs, next_update=1): # There is a tricky case here - what happens if the root CA is compromised ? # In normal world scenarios, that CA is removed from app's trust store and any # subsequent certs it had issues wouldn't be validated by the app then. Making a CRL # for a revoked root CA in normal cases doesn't make sense as the thief can sign a # counter CRL saying that everything is fine. As our environment is controlled, # i think we are safe to create a crl for root CA as well which we can publish for # services which make use of it i.e openvpn and they'll know that the certs/ca's have been # compromised. # # `ca` is root ca from where the chain `certs` starts. # `certs` is a list of all certs ca inclusive which are to be # included in the CRL ( if root ca is compromised, it will be in `certs` as well ). private_key = load_private_key(ca['privatekey']) ca_cert = x509.load_pem_x509_certificate(ca['certificate'].encode(), default_backend()) if not private_key: return None ca_data = self.middleware.call_sync('cryptokey.load_certificate', ca['certificate']) issuer = {k: ca_data.get(v) for k, v in CERT_BACKEND_MAPPINGS.items()} crl_builder = x509.CertificateRevocationListBuilder().issuer_name(x509.Name([ x509.NameAttribute(getattr(NameOID, k.upper()), v) for k, v in issuer.items() if v ])).last_update( utc_now() ).next_update( utc_now() + datetime.timedelta(next_update, 300, 0) ) for cert in certs: crl_builder = crl_builder.add_revoked_certificate( x509.RevokedCertificateBuilder().serial_number( self.middleware.call_sync('cryptokey.load_certificate', cert['certificate'])['serial'] ).revocation_date( cert['revoked_date'] ).build( default_backend() ) ) # https://www.ietf.org/rfc/rfc5280.txt # We should add AuthorityKeyIdentifier and CRLNumber at the very least crl = crl_builder.add_extension( x509.AuthorityKeyIdentifier( x509.SubjectKeyIdentifier.from_public_key( ca_cert.public_key() ).digest, [x509.DirectoryName( x509.Name([ x509.NameAttribute(getattr(NameOID, k.upper()), v) for k, v in issuer.items() if v ]) )], ca_cert.serial_number ), False ).add_extension( x509.CRLNumber(1), False ).sign( private_key=private_key, algorithm=retrieve_signing_algorithm({}, private_key), backend=default_backend() ) return crl.public_bytes(serialization.Encoding.PEM).decode()
3,482
Python
.py
69
39.695652
117
0.632539
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,007
pci.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/pci.py
import collections import os import re from pyudev import Context from middlewared.schema import accepts, Bool, Dict, Int, List, Ref, returns, Str from middlewared.service import private, Service, ValidationErrors from middlewared.utils.gpu import get_gpus from middlewared.utils.iommu import get_iommu_groups_info from middlewared.utils.pci import get_pci_device_class, SENSITIVE_PCI_DEVICE_TYPES from .utils import convert_pci_id_to_vm_pci_slot RE_DEVICE_PATH = re.compile(r'pci_(\w+)_(\w+)_(\w+)_(\w+)') class VMDeviceService(Service): class Config: namespace = 'vm.device' @accepts(roles=['VM_DEVICE_READ']) @returns(Bool()) def iommu_enabled(self): """Returns "true" if iommu is enabled, "false" otherwise""" return os.path.exists('/sys/kernel/iommu_groups') @private def get_pci_device_default_data(self): return { 'capability': { 'class': None, 'domain': None, 'bus': None, 'slot': None, 'function': None, 'product': 'Not Available', 'vendor': 'Not Available', }, 'controller_type': None, 'critical': False, 'iommu_group': {}, 'available': False, 'drivers': [], 'error': None, 'device_path': None, 'reset_mechanism_defined': False, 'description': '', } @private def get_pci_device_details(self, obj, iommu_info): data = self.get_pci_device_default_data() if not (igi := iommu_info.get(obj.sys_name)): data['error'] = 'Unable to determine iommu group' dbs, func = obj.sys_name.split('.') dom, bus, slot = dbs.split(':') device_path = os.path.join('/sys/bus/pci/devices', obj.sys_name) cap_class = f'{(obj.attributes.get("class") or b"").decode()}' or get_pci_device_class(device_path) controller_type = obj.properties.get('ID_PCI_SUBCLASS_FROM_DATABASE') or SENSITIVE_PCI_DEVICE_TYPES.get( cap_class[:6] ) drivers = [] if driver := obj.properties.get('DRIVER'): drivers.append(driver) data['capability']['class'] = cap_class or None data['capability']['domain'] = f'{int(dom, base=16)}' data['capability']['bus'] = f'{int(bus, base=16)}' data['capability']['slot'] = f'{int(slot, base=16)}' data['capability']['function'] = f'{int(func, base=16)}' data['capability']['product'] = obj.properties.get('ID_MODEL_FROM_DATABASE', 'Not Available') data['capability']['vendor'] = obj.properties.get('ID_VENDOR_FROM_DATABASE', 'Not Available') data['controller_type'] = controller_type data['critical'] = bool(not cap_class or SENSITIVE_PCI_DEVICE_TYPES.get(cap_class[:6])) data['iommu_group'] = igi data['available'] = all(i == 'vfio-pci' for i in drivers) and not data['critical'] data['drivers'] = drivers data['device_path'] = os.path.join('/sys/bus/pci/devices', obj.sys_name) data['reset_mechanism_defined'] = os.path.exists(os.path.join(data['device_path'], 'reset')) prefix = obj.sys_name + (f' {controller_type!r}' if controller_type else '') vendor = data['capability']['vendor'].strip() suffix = data['capability']['product'].strip() if vendor and suffix: data['description'] = f'{prefix}: {suffix} by {vendor!r}' else: data['description'] = prefix return data @private def get_all_pci_devices_details(self): result = dict() iommu_info = get_iommu_groups_info() for i in Context().list_devices(subsystem='pci'): key = f"pci_{i.sys_name.replace(':', '_').replace('.', '_')}" result[key] = self.get_pci_device_details(i, iommu_info) return result @private def get_single_pci_device_details(self, pcidev): result = dict() iommu_info = get_iommu_groups_info() for i in filter(lambda x: x.sys_name == pcidev, Context().list_devices(subsystem='pci')): key = f"pci_{i.sys_name.replace(':', '_').replace('.', '_')}" result[key] = self.get_pci_device_details(i, iommu_info) return result @accepts(Str('device'), roles=['VM_DEVICE_READ']) @returns(Dict( 'passthrough_device', Dict( 'capability', Str('class', null=True, required=True), Str('domain', null=True, required=True), Str('bus', null=True, required=True), Str('slot', null=True, required=True), Str('function', null=True, required=True), Str('product', null=True, required=True), Str('vendor', null=True, required=True), required=True, ), Str('controller_type', null=True, required=True), Dict( 'iommu_group', Int('number', required=True), List('addresses', items=[Dict( 'address', Str('domain', required=True), Str('bus', required=True), Str('slot', required=True), Str('function', required=True), )]), required=True, ), Bool('available', required=True), List('drivers', items=[Str('driver', required=False)], required=True), Str('error', null=True, required=True), Str('device_path', null=True, required=True), Bool('reset_mechanism_defined', required=True), Str('description', empty=True, required=True), register=True, )) def passthrough_device(self, device): """Retrieve details about `device` PCI device""" self.middleware.call_sync('vm.check_setup_libvirt') if device_details := self.get_single_pci_device_details(RE_DEVICE_PATH.sub(r'\1:\2:\3.\4', device)): return device_details[device] else: return { **self.get_pci_device_default_data(), 'error': 'Device not found', } @accepts(roles=['VM_DEVICE_READ']) @returns(List(items=[Ref('passthrough_device')], register=True)) def passthrough_device_choices(self): """Available choices for PCI passthru devices""" return self.get_all_pci_devices_details() @accepts() @returns(Ref('passthrough_device_choices')) def pptdev_choices(self): """Available choices for PCI passthru device""" return self.get_all_pci_devices_details() @accepts(Str('gpu_pci_id', empty=False)) @returns(List(items=[Str('pci_ids')])) def get_pci_ids_for_gpu_isolation(self, gpu_pci_id): """ Get PCI IDs of devices which are required to be isolated for `gpu_pci_id` GPU isolation. Basically when a GPU passthrough is desired for a VM, we need to isolate all the devices which are in the same IOMMU group as the GPU. This is required because if we don't do this, the VM will not be able to start because the devices in the same IOMMU group as the GPU will be in use by the host and will not be available for the VM to use. This endpoints retrieves all the PCI devices which are in the same IOMMU group as the GPU and returns their PCI IDs so UI can use those and create PCI devices for them and isolate them. """ gpu = next((gpu for gpu in get_gpus() if gpu['addr']['pci_slot'] == gpu_pci_id), None) verrors = ValidationErrors() if not gpu: verrors.add('gpu_pci_id', f'GPU with {gpu_pci_id!r} PCI ID not found') verrors.check() iommu_groups = get_iommu_groups_info() iommu_groups_mapping_with_group_no = collections.defaultdict(set) for pci_slot, pci_details in iommu_groups.items(): iommu_groups_mapping_with_group_no[pci_details['number']].add(convert_pci_id_to_vm_pci_slot(pci_slot)) pci_ids = set() for device in gpu['devices']: if not (device_info := iommu_groups.get(device['pci_slot'])): continue pci_ids.update(iommu_groups_mapping_with_group_no[device_info['number']]) return list(pci_ids)
8,306
Python
.py
176
37.528409
119
0.598074
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,008
vm_memory_info.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/vm_memory_info.py
import psutil from middlewared.schema import accepts, Bool, Dict, Int, returns, Str from middlewared.service import CallError, Service from middlewared.validators import MACAddr from .devices import NIC from .utils import ACTIVE_STATES class VMService(Service): @accepts() @returns(Dict( 'vmemory_in_use', Int('RNP', required=True, description='Running but not provisioned'), Int('PRD', required=True, description='Provisioned but not running'), Int('RPRD', required=True, description='Running and provisioned'), )) async def get_vmemory_in_use(self): """ The total amount of virtual memory in MB used by guests Returns a dict with the following information: RNP - Running but not provisioned PRD - Provisioned but not running RPRD - Running and provisioned """ memory_allocation = {'RNP': 0, 'PRD': 0, 'RPRD': 0} guests = await self.middleware.call('datastore.query', 'vm.vm') for guest in guests: status = await self.middleware.call('vm.status', guest['id']) if status['state'] in ACTIVE_STATES: memory_allocation['RPRD' if guest['autostart'] else 'RNP'] += guest['memory'] * 1024 * 1024 elif guest['autostart']: memory_allocation['PRD'] += guest['memory'] * 1024 * 1024 return memory_allocation @accepts(Bool('overcommit', default=False), roles=['VM_READ']) @returns(Int('available_memory')) async def get_available_memory(self, overcommit): """ Get the current maximum amount of available memory to be allocated for VMs. In case of `overcommit` being `true`, calculations are done in the following manner: 1. If a VM has requested 10G but is only consuming 5G, only 5G will be counted 2. System will consider shrinkable ZFS ARC as free memory ( shrinkable ZFS ARC is current ZFS ARC minus ZFS ARC minimum ) In case of `overcommit` being `false`, calculations are done in the following manner: 1. Complete VM requested memory will be taken into account regardless of how much actual physical memory the VM is consuming 2. System will not consider shrinkable ZFS ARC as free memory Memory is of course a very "volatile" resource, values may change abruptly between a second but I deem it good enough to give the user a clue about how much memory is available at the current moment and if a VM should be allowed to be launched. """ # Use 90% of available memory to play safe free = int(psutil.virtual_memory().available * 0.9) # Difference between current ARC total size and the minimum allowed arc_total = await self.middleware.call('sysctl.get_arcstats_size') arc_min = await self.middleware.call('sysctl.get_arc_min') arc_shrink = max(0, arc_total - arc_min) total_free = free + arc_shrink vms_memory_used = 0 if overcommit is False: # If overcommit is not wanted its verified how much physical memory # the vm process is currently using and add the maximum memory its # supposed to have. for vm in await self.middleware.call('vm.query', [['status.state', 'in', ACTIVE_STATES]]): try: current_vm_mem = await self.middleware.call('vm.get_memory_usage_internal', vm) except Exception: self.logger.error('Unable to retrieve %r vm memory usage', vm['name'], exc_info=True) continue else: vm_max_mem = vm['memory'] * 1024 * 1024 # We handle edge case with vm_max_mem < current_vm_mem if vm_max_mem > current_vm_mem: vms_memory_used += vm_max_mem - current_vm_mem return max(0, total_free - vms_memory_used) @accepts(Int('vm_id'), roles=['VM_READ']) @returns(Dict( Int('minimum_memory_requested', description='Minimum memory requested by the VM'), Int('total_memory_requested', description='Maximum / total memory requested by the VM'), Bool('overcommit_required', description='Overcommit of memory is required to start VM'), Bool( 'memory_req_fulfilled_after_overcommit', description='Memory requirements of VM are fulfilled if over-committing memory is specified' ), Int('arc_to_shrink', description='Size of ARC to shrink in bytes', null=True), Int('current_arc_max', description='Current size of max ARC in bytes'), Int('arc_min', description='Minimum size of ARC in bytes'), Int('arc_max_after_shrink', description='Size of max ARC in bytes after shrinking'), Int( 'actual_vm_requested_memory', description='VM memory in bytes to consider when making calculations for available/required memory.' ' If VM ballooning is specified for the VM, the minimum VM memory specified by user will' ' be taken into account otherwise total VM memory requested will be taken into account.' ), )) async def get_vm_memory_info(self, vm_id): """ Returns memory information for `vm_id` VM if it is going to be started. All memory attributes are expressed in bytes. """ vm = await self.middleware.call('vm.get_instance', vm_id) if vm['status']['state'] in ACTIVE_STATES: # TODO: Let's add this later as we have a use case in the UI - could be useful to # show separate info of each VM in the UI moving on raise CallError(f'Unable to retrieve {vm["name"]!r} VM information as it is already running.') arc_max = await self.middleware.call('sysctl.get_arc_max') arc_min = await self.middleware.call('sysctl.get_arc_min') shrinkable_arc_max = max(0, arc_max - arc_min) available_memory = await self.get_available_memory(False) available_memory_with_overcommit = await self.get_available_memory(True) vm_max_memory = vm['memory'] * 1024 * 1024 vm_min_memory = vm['min_memory'] * 1024 * 1024 if vm['min_memory'] else None vm_requested_memory = vm_min_memory or vm_max_memory overcommit_required = vm_requested_memory > available_memory arc_to_shrink = 0 if overcommit_required: arc_to_shrink = min(shrinkable_arc_max, vm_requested_memory - available_memory) return { 'minimum_memory_requested': vm_min_memory, 'total_memory_requested': vm_max_memory, 'overcommit_required': overcommit_required, 'arc_to_shrink': arc_to_shrink, 'memory_req_fulfilled_after_overcommit': vm_requested_memory < available_memory_with_overcommit, 'current_arc_max': arc_max, 'arc_min': arc_min, 'arc_max_after_shrink': arc_max - arc_to_shrink, 'actual_vm_requested_memory': vm_requested_memory, } @accepts() @returns(Str('mac', validators=[MACAddr(separator=':')]),) def random_mac(self): """ Create a random mac address. Returns: str: with six groups of two hexadecimal digits """ return NIC.random_mac()
7,437
Python
.py
134
44.977612
113
0.635902
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,009
vm_devices.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/vm_devices.py
import os import re import middlewared.sqlalchemy as sa from middlewared.plugins.vm.devices.storage_devices import IOTYPE_CHOICES from middlewared.plugins.zfs_.utils import zvol_name_to_path, zvol_path_to_name from middlewared.schema import accepts, Bool, Dict, Int, OROperator, Patch, returns, Str from middlewared.service import CallError, CRUDService, private from middlewared.utils import run from middlewared.async_validators import check_path_resides_within_volume from .devices import DEVICES from .utils import ACTIVE_STATES RE_PPTDEV_NAME = re.compile(r'([0-9]+/){2}[0-9]+') class VMDeviceModel(sa.Model): __tablename__ = 'vm_device' id = sa.Column(sa.Integer(), primary_key=True) dtype = sa.Column(sa.String(50)) attributes = sa.Column(sa.JSON()) vm_id = sa.Column(sa.ForeignKey('vm_vm.id'), index=True) order = sa.Column(sa.Integer(), nullable=True) class VMDeviceService(CRUDService): ENTRY = Patch( 'vmdevice_create', 'vm_device_entry', ('add', Int('id')), ('rm', {'name': 'attributes'}), ('add', OROperator(*[device.schema for device in DEVICES.values()], name='attributes')), ) class Config: namespace = 'vm.device' datastore = 'vm.device' datastore_extend = 'vm.device.extend_device' cli_namespace = 'service.vm.device' role_prefix = 'VM_DEVICE' @accepts() @returns(Dict(additional_attrs=True, example={'vms/test 1': '/dev/zvol/vms/test+1'})) async def disk_choices(self): """ Returns disk choices for device type "DISK". """ out = {} zvols = await self.middleware.call( 'zfs.dataset.unlocked_zvols_fast', [ ['OR', [['attachment', '=', None], ['attachment.method', '=', 'vm.devices.query']]], ['ro', '=', False], ], {}, ['ATTACHMENT', 'RO'] ) for zvol in zvols: out[zvol['path']] = zvol['name'] return out @accepts() @returns(Dict( *[Str(k, enum=[k]) for k in IOTYPE_CHOICES] )) async def iotype_choices(self): """ IO-type choices for storage devices. """ return {k: k for k in IOTYPE_CHOICES} @private async def extend_device(self, device): if device['vm']: device['vm'] = device['vm']['id'] if not device['order']: if device['dtype'] == 'CDROM': device['order'] = 1000 elif device['dtype'] in ('DISK', 'RAW'): device['order'] = 1001 else: device['order'] = 1002 return device @accepts(roles=['VM_DEVICE_READ']) @returns(Dict(additional_attrs=True)) def nic_attach_choices(self): """ Available choices for NIC Attach attribute. """ return self.middleware.call_sync('interface.choices', {'exclude': ['epair', 'tap', 'vnet']}) @accepts(roles=['VM_DEVICE_READ']) @returns(Dict(additional_attrs=True)) async def bind_choices(self): """ Available choices for Bind attribute. """ return { d['address']: d['address'] for d in await self.middleware.call( 'interface.ip_in_use', {'static': True, 'any': True, 'loopback': True} ) } @private async def update_device(self, data, old=None): if data['dtype'] == 'DISK': create_zvol = data['attributes'].pop('create_zvol', False) if create_zvol: ds_options = { 'name': data['attributes'].pop('zvol_name'), 'type': 'VOLUME', 'volsize': data['attributes'].pop('zvol_volsize'), } self.logger.debug(f'Creating ZVOL {ds_options["name"]} with volsize {ds_options["volsize"]}') zvol_blocksize = await self.middleware.call( 'pool.dataset.recommended_zvol_blocksize', ds_options['name'].split('/', 1)[0] ) ds_options['volblocksize'] = zvol_blocksize await self.middleware.call('pool.dataset.create', ds_options) elif data['dtype'] == 'RAW' and ( not data['attributes'].pop('exists', True) or ( old and old['attributes']['size'] != data['attributes']['size'] ) ): path = data['attributes']['path'] cp = await run(['truncate', '-s', str(data['attributes']['size']), path], check=False) if cp.returncode: raise CallError(f'Failed to create or update raw file {path}: {cp.stderr}') return data @accepts( Dict( 'vmdevice_create', Str('dtype', enum=['NIC', 'DISK', 'CDROM', 'PCI', 'DISPLAY', 'RAW', 'USB'], required=True), Int('vm', required=True), Dict('attributes', additional_attrs=True, default=None), Int('order', default=None, null=True), register=True, ), ) async def do_create(self, data): """ Create a new device for the VM of id `vm`. If `dtype` is the `RAW` type and a new raw file is to be created, `attributes.exists` will be passed as false. This means the API handles creating the raw file and raises the appropriate exception if file creation fails. If `dtype` is of `DISK` type and a new Zvol is to be created, `attributes.create_zvol` will be passed as true with valid `attributes.zvol_name` and `attributes.zvol_volsize` values. """ data = await self.validate_device(data, update=False) data = await self.update_device(data) id_ = await self.middleware.call( 'datastore.insert', self._config.datastore, data ) await self.__reorder_devices(id_, data['vm'], data['order']) return await self.get_instance(id_) async def do_update(self, id_, data): """ Update a VM device of `id`. Pass `attributes.size` to resize a `dtype` `RAW` device. The raw file will be resized. """ device = await self.get_instance(id_) new = device.copy() new.update(data) new = await self.validate_device(new, device) new = await self.update_device(new, device) await self.middleware.call('datastore.update', self._config.datastore, id_, new) await self.__reorder_devices(id_, device['vm'], new['order']) return await self.get_instance(id_) @private async def delete_resource(self, options, device): if options['zvol']: if device['dtype'] != 'DISK': raise CallError('The device is not a disk and has no zvol to destroy.') if not device['attributes'].get('path', '').startswith('/dev/zvol'): raise CallError('Unable to destroy zvol as disk device has misconfigured path') zvol_id = zvol_path_to_name(device['attributes']['path']) if await self.middleware.call('pool.dataset.query', [['id', '=', zvol_id]]): # FIXME: We should use pool.dataset.delete but right now FS attachments will consider # the current device as a valid reference. Also should we stopping the vm only when deleting an # attachment ? await self.middleware.call('zfs.dataset.delete', zvol_id) if options['raw_file']: if device['dtype'] != 'RAW': raise CallError('Device is not of RAW type.') try: os.unlink(device['attributes']['path']) except OSError: raise CallError(f'Failed to destroy {device["attributes"]["path"]}') @accepts( Int('id'), Dict( 'vm_device_delete', Bool('zvol', default=False), Bool('raw_file', default=False), Bool('force', default=False), ) ) async def do_delete(self, id_, options): """ Delete a VM device of `id`. """ device = await self.get_instance(id_) status = await self.middleware.call('vm.status', device['vm']) if status['state'] in ACTIVE_STATES: raise CallError('Please stop/resume associated VM before deleting VM device.') try: await self.delete_resource(options, device) except CallError: if not options['force']: raise return await self.middleware.call('datastore.delete', self._config.datastore, id_) async def __reorder_devices(self, id_, vm_id, order): if order is None: return filters = [('vm', '=', vm_id), ('id', '!=', id_)] if await self.middleware.call('vm.device.query', filters + [('order', '=', order)]): used_order = [order] for device in await self.middleware.call('vm.device.query', filters, {'order_by': ['order']}): if not device['order']: continue if device['order'] not in used_order: used_order.append(device['order']) continue device['order'] = min(used_order) + 1 while device['order'] in used_order: device['order'] += 1 used_order.append(device['order']) await self.middleware.call('datastore.update', self._config.datastore, device['id'], device) @private async def disk_uniqueness_integrity_check(self, device, vm): # This ensures that the disk is not already present for `vm` def translate_device(dev): # A disk should have a path configured at all times, when that is not the case, that means `dtype` is DISK # and end user wants to create a new zvol in this case. return dev['attributes'].get('path') or zvol_name_to_path(dev['attributes']['zvol_name']) disks = [ d for d in vm['devices'] if d['dtype'] in ('DISK', 'RAW', 'CDROM') and translate_device(d) == translate_device(device) ] if not disks: # We don't have that disk path in vm devices, we are good to go return True elif len(disks) > 1: # VM is mis-configured return False elif not device.get('id') and disks: # A new device is being created, however it already exists in vm. This can also happen when VM instance # is being created, in that case it's okay. Key here is that we won't have the id field present return not bool(disks[0].get('id')) elif device.get('id'): # The device is being updated, if the device is same as we have in db, we are okay return device['id'] == disks[0].get('id') else: return False @private async def validate_path_field(self, verrors, schema, path): await check_path_resides_within_volume(verrors, self.middleware, schema, path) @private async def validate_device(self, device, old=None, update=True): vm_instance = await self.middleware.call('vm.get_instance', device['vm']) device_obj = DEVICES[device['dtype']](device, self.middleware) await self.middleware.run_in_thread(device_obj.validate, device, old, vm_instance, update) return device @private async def validate_display_devices(self, verrors, vm_instance): devs = await self.get_display_devices(vm_instance) if len(devs['spice']) > 1: verrors.add('attributes.type', 'Only one SPICE Display device is supported') @private async def get_display_devices(self, vm_instance): devs = {'spice': []} for dev in filter(lambda d: d['dtype'] == 'DISPLAY', vm_instance['devices']): if dev['attributes']['type'] == 'SPICE': devs['spice'].append(dev) return devs
11,970
Python
.py
259
35.915058
118
0.584034
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,010
vm_lifecycle.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/vm_lifecycle.py
from middlewared.schema import accepts, Bool, Dict, Int, returns from middlewared.service import CallError, item_method, job, private, Service from .vm_supervisor import VMSupervisorMixin class VMService(Service, VMSupervisorMixin): @private async def lifecycle_action_check(self): if not await self.middleware.call('vm.license_active'): raise CallError('Requested action cannot be performed as system is not licensed to use VMs') @item_method @accepts( Int('id'), Dict('options', Bool('overcommit', default=False)), roles=['VM_WRITE'] ) @returns() async def start(self, id_, options): """ Start a VM. options.overcommit defaults to false, meaning VMs are not allowed to start if there is not enough available memory to hold all configured VMs. If true, VM starts even if there is not enough memory for all configured VMs. Error codes: ENOMEM(12): not enough free memory to run the VM without overcommit """ await self.lifecycle_action_check() await self.middleware.run_in_thread(self._check_setup_connection) vm = await self.middleware.call('vm.get_instance', id_) vm_state = vm['status']['state'] if vm_state == 'RUNNING': raise CallError(f'{vm["name"]!r} is already running') if vm_state == 'SUSPENDED': raise CallError(f'{vm["name"]!r} VM is suspended and can only be resumed/powered off') if vm['bootloader'] not in await self.middleware.call('vm.bootloader_options'): raise CallError(f'"{vm["bootloader"]}" is not supported on this platform.') if await self.middleware.call('system.is_ha_capable'): for device in vm['devices']: if device['dtype'] in ('PCI', 'USB'): raise CallError( 'Please remove PCI/USB devices from VM before starting it in HA capable machines as ' 'they are not supported.' ) # Perhaps we should have a default config option for VMs? await self.middleware.call('vm.init_guest_vmemory', vm, options['overcommit']) try: await self.middleware.run_in_thread(self._start, vm['name']) except Exception: if (await self.middleware.call('vm.get_instance', id_))['status']['state'] != 'RUNNING': await self.middleware.call('vm.teardown_guest_vmemory', id_) raise await self.middleware.call('service.reload', 'http') @item_method @accepts( Int('id'), Dict( 'options', Bool('force', default=False), Bool('force_after_timeout', default=False), ), roles=['VM_WRITE'] ) @returns() @job(lock=lambda args: f'stop_vm_{args[0]}') def stop(self, job, id_, options): """ Stops a VM. For unresponsive guests who have exceeded the `shutdown_timeout` defined by the user and have become unresponsive, they required to be powered down using `vm.poweroff`. `vm.stop` is only going to send a shutdown signal to the guest and wait the desired `shutdown_timeout` value before tearing down guest vmemory. `force_after_timeout` when supplied, it will initiate poweroff for the VM forcing it to exit if it has not already stopped within the specified `shutdown_timeout`. """ self._check_setup_connection() vm_data = self.middleware.call_sync('vm.get_instance', id_) if options['force']: self._poweroff(vm_data['name']) else: self._stop(vm_data['name'], vm_data['shutdown_timeout']) if options['force_after_timeout'] and self.middleware.call_sync('vm.status', id_)['state'] == 'RUNNING': self._poweroff(vm_data['name']) @item_method @accepts(Int('id'), roles=['VM_WRITE']) @returns() def poweroff(self, id_): """ Poweroff a VM. """ self._check_setup_connection() vm_data = self.middleware.call_sync('vm.get_instance', id_) self._poweroff(vm_data['name']) @item_method @accepts(Int('id'), roles=['VM_WRITE']) @returns() @job(lock=lambda args: f'restart_vm_{args[0]}') def restart(self, job, id_): """ Restart a VM. """ self._check_setup_connection() vm = self.middleware.call_sync('vm.get_instance', id_) stop_job = self.middleware.call_sync('vm.stop', id_, {'force_after_timeout': True}) stop_job.wait_sync() if stop_job.error: raise CallError(f'Failed to stop {vm["name"]!r} vm: {stop_job.error}') self.middleware.call_sync('vm.start', id_, {'overcommit': True}) @item_method @accepts(Int('id'), roles=['VM_WRITE']) @returns() def suspend(self, id_): """ Suspend `id` VM. """ self._check_setup_connection() vm = self.middleware.call_sync('vm.get_instance', id_) self._suspend(vm['name']) @item_method @accepts(Int('id'), roles=['VM_WRITE']) @returns() def resume(self, id_): """ Resume suspended `id` VM. """ self._check_setup_connection() vm = self.middleware.call_sync('vm.get_instance', id_) self._resume(vm['name']) @private def suspend_vms(self, vm_ids): vms = {vm['id']: vm for vm in self.middleware.call_sync('vm.query')} for vm_id in filter( lambda vm_id: vms.get(vm_id).get('status', {}).get('state') == 'RUNNING', map(int, vm_ids) ): try: self.suspend(vm_id) except Exception: self.logger.error('Failed to suspend %r vm', vms[vm_id]['name'], exc_info=True) @private def resume_suspended_vms(self, vm_ids): vms = {vm['id']: vm for vm in self.middleware.call_sync('vm.query')} for vm_id in filter( lambda vm_id: vms.get(vm_id).get('status', {}).get('state') == 'SUSPENDED', map(int, vm_ids) ): try: self.resume(vm_id) except Exception: self.logger.error('Failed to resume %r vm', vms[vm_id]['name'], exc_info=True) async def _event_vms(middleware, event_type, args): vm = await middleware.call('vm.query', [['id', '=', args['id']]]) if not vm or vm[0]['status']['state'] != 'STOPPED' or args.get('state') != 'SHUTOFF': return middleware.create_task(middleware.call('vm.teardown_guest_vmemory', args['id'])) async def setup(middleware): middleware.event_subscribe('vm.query', _event_vms)
6,748
Python
.py
153
34.882353
117
0.597927
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,011
vm_display_info.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/vm_display_info.py
from socket import AF_INET6 from middlewared.schema import accepts, Dict, Int, List, returns, Str from middlewared.service import pass_app, private, Service from .devices import DISPLAY from .utils import NGINX_PREFIX class VMService(Service): @accepts(roles=['VM_READ']) @returns(Dict( 'available_display_port', Int('port', required=True, description='Available server port'), Int('web', required=True, description='Web port to be used based on available `port`'), )) async def port_wizard(self): """ It returns the next available Display Server Port and Web Port. Returns a dict with two keys `port` and `web`. """ all_ports = await self.middleware.call('port.get_all_used_ports') def get_next_port(): for i in filter(lambda i: i not in all_ports, range(5900, 65535)): yield i gen = get_next_port() return {'port': next(gen), 'web': next(gen)} @private async def all_used_display_device_ports(self, additional_filters=None): all_ports = [6000] additional_filters = additional_filters or [] for device in await self.middleware.call('vm.device.query', [['dtype', '=', 'DISPLAY']] + additional_filters): all_ports.extend([device['attributes']['port'], device['attributes']['web_port']]) return all_ports @accepts() @returns(Dict( *[Str(r, enum=[r]) for r in DISPLAY.RESOLUTION_ENUM] )) async def resolution_choices(self): """ Retrieve supported resolution choices for VM Display devices. """ return {r: r for r in DISPLAY.RESOLUTION_ENUM} @accepts(Int('id'), roles=['VM_READ']) @returns(List( 'vmdevice', items=[ Dict( 'vmdevice', Int('id'), Str('dtype'), DISPLAY.schema, Int('order'), Int('vm'), ), ] )) async def get_display_devices(self, id_): """ Get the display devices from a given guest. If a display device has password configured, `attributes.password_configured` will be set to `true`. """ devices = [] for device in await self.middleware.call('vm.device.query', [['vm', '=', id_], ['dtype', '=', 'DISPLAY']]): device['attributes']['password_configured'] = bool(device['attributes'].get('password')) devices.append(device) return devices @accepts( Int('id'), Str('host', default=''), Dict( 'options', Str('protocol', default='HTTP', enum=['HTTP', 'HTTPS']), ), roles=['VM_READ'] ) @returns(Dict( 'display_devices_uri', Str('error', null=True), Str('uri', null=True), )) @pass_app() async def get_display_web_uri(self, app, id_, host, options): """ Retrieve Display URI for a given VM or appropriate error if there is no display device available or if it is not configured to use web interface """ uri_data = {'error': None, 'uri': None} protocol = options['protocol'].lower() if not host: try: if app.origin.is_tcp_ip_family and (_h := app.origin.loc_addr): host = _h if app.origin.family == AF_INET6: host = f'[{_h}]' except AttributeError: pass if display_devices := await self.get_display_devices(id_): for device in map(lambda d: DISPLAY(d, middleware=self.middleware), display_devices): if device.data['attributes'].get('web'): uri_data['uri'] = device.web_uri(host, protocol=protocol) else: uri_data['error'] = 'Web display is not configured' else: uri_data['error'] = 'Display device is not configured for this VM' return uri_data @private async def get_display_devices_ui_info(self): devices = [] for device in await self.middleware.call('vm.device.query', [['dtype', '=', 'DISPLAY']]): devices.append(DISPLAY(device, middleware=self.middleware).get_webui_info()) return devices @private async def get_vm_display_nginx_route(self): return NGINX_PREFIX
4,425
Python
.py
110
30.536364
118
0.575017
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,012
memory.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/memory.py
import errno from middlewared.schema import accepts, Int, returns from middlewared.service import CallError, private, Service from .utils import ACTIVE_STATES from .vm_supervisor import VMSupervisorMixin class VMService(Service, VMSupervisorMixin): async def _set_guest_vmemory(self, vm_id, overcommit): vm = await self.middleware.call('vm.get_instance', vm_id) memory_details = await self.middleware.call('vm.get_vm_memory_info', vm_id) if not memory_details['overcommit_required']: # There really isn't anything to be done if over-committing is not required return if not overcommit: raise CallError(f'Cannot guarantee memory for guest {vm["name"]}', errno.ENOMEM) if memory_details['current_arc_max'] != memory_details['arc_max_after_shrink']: self.logger.debug( 'Setting ARC from %s to %s', memory_details['current_arc_max'], memory_details['arc_max_after_shrink'] ) await self.middleware.call('sysctl.set_arc_max', memory_details['arc_max_after_shrink']) @private async def init_guest_vmemory(self, vm, overcommit): guest_status = await self.middleware.call('vm.status', vm['id']) if guest_status.get('state') not in ACTIVE_STATES: await self._set_guest_vmemory(vm['id'], overcommit) else: raise CallError('VM process is running, we won\'t allocate memory') @private async def teardown_guest_vmemory(self, vm_id): vm = await self.middleware.call('vm.get_instance', vm_id) if vm['status']['state'] != 'STOPPED': return guest_memory = vm['memory'] * 1024 * 1024 arc_max = await self.middleware.call('sysctl.get_arc_max') arc_min = await self.middleware.call('sysctl.get_arc_min') new_arc_max = min( await self.middleware.call('sysctl.get_default_arc_max'), arc_max + guest_memory ) if arc_max != new_arc_max: if new_arc_max > arc_min: self.logger.debug(f'Giving back guest memory to ARC: {new_arc_max}') await self.middleware.call('sysctl.set_arc_max', new_arc_max) else: self.logger.warn( f'Not giving back memory to ARC because new arc_max ({new_arc_max}) <= arc_min ({arc_min})' ) @accepts(Int('vm_id'), roles=['VM_READ']) @returns(Int('memory_usage', description='Memory usage of a VM in bytes')) def get_memory_usage(self, vm_id): return self.get_memory_usage_internal(self.middleware.call_sync('vm.get_instance', vm_id)) @private def get_memory_usage_internal(self, vm): return self._memory_info(vm['name'])
2,769
Python
.py
53
42.716981
118
0.636834
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,013
usb.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/usb.py
import re from xml.etree import ElementTree as etree from middlewared.schema import accepts, Bool, Dict, List, Ref, Str, returns from middlewared.service import CallError, private, Service from middlewared.utils import run from .devices.usb import USB_CONTROLLER_CHOICES from .utils import get_virsh_command_args RE_VALID_USB_DEVICE = re.compile(r'^usb_\d+_\d+(?:_\d)*$') class VMDeviceService(Service): class Config: namespace = 'vm.device' @accepts() @returns(Dict(*[Str(k, enum=[k]) for k in USB_CONTROLLER_CHOICES])) async def usb_controller_choices(self): """ Retrieve USB controller type choices """ return {k: k for k in USB_CONTROLLER_CHOICES} @private def retrieve_usb_device_information(self, xml_str): xml = etree.fromstring(xml_str.strip()) capability = next((e for e in list(xml) if e.tag == 'capability'), None) if capability is None: return capability required_keys = set(self.get_capability_keys()) capability_info = {} for element in filter(lambda e: e.tag in required_keys and e.text is not None, capability): capability_info[element.tag] = element.text if element.tag in ('product', 'vendor') and element.get('id'): capability_info[f'{element.tag}_id'] = element.get('id') return None if set(capability_info) != required_keys else capability_info @private def get_capability_keys(self): return { 'product': None, 'vendor': None, 'product_id': None, 'vendor_id': None, 'bus': None, 'device': None, } @accepts(Str('device', empty=False), roles=['VM_DEVICE_READ']) @returns(Dict( Dict( 'capability', Str('product', required=True, null=True), Str('product_id', required=True, null=True), Str('vendor', required=True, null=True), Str('vendor_id', required=True, null=True), Str('bus', required=True, null=True), Str('device', required=True, null=True), ), Bool('available', required=True), Str('error', required=True, null=True), register=True, )) async def usb_passthrough_device(self, device): """ Retrieve details about `device` USB device. """ await self.middleware.call('vm.check_setup_libvirt') data = await self.get_basic_usb_passthrough_device_data() cp = await run(get_virsh_command_args() + ['nodedev-dumpxml', device], check=False) if cp.returncode: data['error'] = cp.stderr.decode() return data capability_info = await self.middleware.call( 'vm.device.retrieve_usb_device_information', cp.stdout.decode() ) if not capability_info: data['error'] = 'Unable to determine capabilities of USB device' else: data['capability'] = capability_info return { **data, 'available': not data['error'], } @private async def get_basic_usb_passthrough_device_data(self): return { 'capability': self.get_capability_keys(), 'available': False, 'error': None, } @accepts(roles=['VM_DEVICE_READ']) @returns(List(items=[Ref('usb_passthrough_device')])) async def usb_passthrough_choices(self): """ Available choices for USB passthrough devices. """ await self.middleware.call('vm.check_setup_libvirt') cp = await run(get_virsh_command_args() + ['nodedev-list', 'usb_device'], check=False) if cp.returncode: raise CallError(f'Unable to retrieve USB devices: {cp.stderr.decode()}') devices = [k for k in map(str.strip, cp.stdout.decode().split('\n')) if RE_VALID_USB_DEVICE.findall(k)] mapping = {} for device in devices: details = await self.usb_passthrough_device(device) if details['error']: continue mapping[device] = details return mapping @private async def get_usb_port_from_usb_details(self, usb_data): if any(not usb_data.get(k) for k in ('product_id', 'vendor_id')): raise CallError('Product / Vendor ID must be specified for USBs') for device, device_details in (await self.usb_passthrough_choices()).items(): capability = device_details['capability'] if all(usb_data[k] == capability[k] for k in ('product_id', 'vendor_id')): return device
4,665
Python
.py
110
33.136364
111
0.606
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,014
clone.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/clone.py
import errno import re import uuid from middlewared.plugins.zfs_.utils import zvol_name_to_path, zvol_path_to_name from middlewared.schema import accepts, Bool, Int, returns, Str from middlewared.service import CallError, item_method, Service, private from middlewared.service_exception import ValidationErrors ZVOL_CLONE_SUFFIX = '_clone' ZVOL_CLONE_RE = re.compile(rf'^(.*){ZVOL_CLONE_SUFFIX}\d+$') class VMService(Service): async def __next_clone_name(self, name): vm_names = [ i['name'] for i in await self.middleware.call('vm.query', [ ('name', '~', rf'{name}{ZVOL_CLONE_SUFFIX}\d+') ]) ] clone_index = 0 while True: clone_name = f'{name}{ZVOL_CLONE_SUFFIX}{clone_index}' if clone_name not in vm_names: break clone_index += 1 return clone_name async def __clone_zvol(self, name, zvol, created_snaps, created_clones): if not await self.middleware.call('zfs.dataset.query', [('id', '=', zvol)]): raise CallError(f'zvol {zvol} does not exist.', errno.ENOENT) snapshot_name = name i = 0 while True: zvol_snapshot = f'{zvol}@{snapshot_name}' if await self.middleware.call('zfs.snapshot.query', [('id', '=', zvol_snapshot)]): if ZVOL_CLONE_RE.search(snapshot_name): snapshot_name = ZVOL_CLONE_RE.sub(rf'\1{ZVOL_CLONE_SUFFIX}{i}', snapshot_name) else: snapshot_name = f'{name}{ZVOL_CLONE_SUFFIX}{i}' i += 1 continue break await self.middleware.call('zfs.snapshot.create', {'dataset': zvol, 'name': snapshot_name}) created_snaps.append(zvol_snapshot) clone_suffix = name i = 0 while True: clone_dst = f'{zvol}_{clone_suffix}' if await self.middleware.call('zfs.dataset.query', [('id', '=', clone_dst)]): if ZVOL_CLONE_RE.search(clone_suffix): clone_suffix = ZVOL_CLONE_RE.sub(rf'\1{ZVOL_CLONE_SUFFIX}{i}', clone_suffix) else: clone_suffix = f'{name}{ZVOL_CLONE_SUFFIX}{i}' i += 1 continue break await self.middleware.call('zfs.snapshot.clone', {'snapshot': zvol_snapshot, 'dataset_dst': clone_dst}) created_clones.append(clone_dst) return clone_dst @item_method @accepts( Int('id'), Str('name', default=None), roles=['VM_WRITE'] ) @returns(Bool()) async def clone(self, id_, name): """ Clone the VM `id`. `name` is an optional parameter for the cloned VM. If not provided it will append the next number available to the VM name. """ vm = await self.middleware.call('vm.get_instance', id_) await self.validate(vm) origin_name = vm['name'] for key in ('id', 'status', 'display_available'): vm.pop(key, None) devices = vm.pop('devices') vm['name'] = await self.__next_clone_name(vm['name']) vm['uuid'] = str(uuid.uuid4()) # We want to use a newer uuid here as it is supposed to be unique per VM if name is not None: vm['name'] = name # In case we need to rollback created_snaps = [] created_clones = [] try: new_vm = await self.middleware.call('vm.do_create', vm) for item in devices: item.pop('id', None) item['vm'] = new_vm['id'] if item['dtype'] == 'NIC': if 'mac' in item['attributes']: del item['attributes']['mac'] if item['dtype'] == 'DISPLAY': if 'port' in item['attributes']: dev_dict = await self.middleware.call('vm.port_wizard') item['attributes']['port'] = dev_dict['port'] item['attributes']['web_port'] = dev_dict['web'] if item['dtype'] == 'DISK': zvol = zvol_path_to_name(item['attributes']['path']) item['attributes']['path'] = zvol_name_to_path( await self.__clone_zvol(vm['name'], zvol, created_snaps, created_clones) ) if item['dtype'] == 'RAW': self.logger.warning('RAW disks must be copied manually. Skipping...') continue await self.middleware.call('vm.device.create', item) except Exception as e: for i in reversed(created_clones): try: await self.middleware.call('zfs.dataset.delete', i) except Exception: self.logger.warning('Rollback of VM clone left dangling zvol: %s', i) for i in reversed(created_snaps): try: dataset, snap = i.split('@') await self.middleware.call('zfs.snapshot.remove', { 'dataset': dataset, 'name': snap, 'defer_delete': True, }) except Exception: self.logger.warn('Rollback of VM clone left dangling snapshot: %s', i) raise e self.logger.info('VM cloned from {0} to {1}'.format(origin_name, vm['name'])) return True @private async def validate(self, vm): verrors = ValidationErrors() for index, device in enumerate(vm['devices']): if device['dtype'] == 'DISPLAY' and not device['attributes'].get('password'): verrors.add( f'vm.devices.{index}.attributes.password', 'Password must be configured for display device in order to clone the VM.' ) verrors.check()
6,001
Python
.py
133
32.120301
112
0.533196
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,015
lifecycle.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/lifecycle.py
import asyncio import contextlib from middlewared.schema import accepts, Bool, Dict from middlewared.service import CallError, private, Service from middlewared.utils.asyncio_ import asyncio_map from .utils import ACTIVE_STATES from .vm_supervisor import VMSupervisorMixin SHUTDOWN_LOCK = asyncio.Lock() class VMService(Service, VMSupervisorMixin): @private async def wait_for_libvirtd(self, timeout): async def libvirtd_started(middleware): await middleware.call('service.start', 'libvirtd') while not await middleware.call('service.started', 'libvirtd'): await asyncio.sleep(2) try: self._system_supports_virtualization() if not await self.middleware.call('service.started', 'libvirtd'): await asyncio.wait_for(self.middleware.create_task(libvirtd_started(self.middleware)), timeout=timeout) # We want to do this before initializing libvirt connection await self.middleware.run_in_thread(self._open) await self.middleware.run_in_thread(self._check_connection_alive) await self.middleware.call('vm.setup_libvirt_events') except (asyncio.TimeoutError, CallError): self.middleware.logger.error('Failed to setup libvirt', exc_info=True) @private def setup_libvirt_connection(self, timeout=30): self.middleware.call_sync('vm.wait_for_libvirtd', timeout) @private async def check_setup_libvirt(self): if not await self.middleware.call('service.started', 'libvirtd'): await self.middleware.call('vm.setup_libvirt_connection') @private def initialize_vms(self, timeout=30): vms = self.middleware.call_sync('vm.query') if vms and self._is_kvm_supported(): self.setup_libvirt_connection(timeout) else: return if self._is_connection_alive(): for vm_data in vms: try: self._add_with_vm_data(vm_data) except Exception as e: # Whatever happens, we don't want middlewared not booting self.middleware.logger.error( 'Unable to setup %r VM object: %s', vm_data['name'], str(e), exc_info=True ) self.middleware.call_sync('service.reload', 'http') else: self.middleware.logger.error('Failed to establish libvirt connection') @private async def start_on_boot(self): for vm in await self.middleware.call('vm.query', [('autostart', '=', True)], {'force_sql_filters': True}): try: await self.middleware.call('vm.start', vm['id']) except Exception as e: self.middleware.logger.error(f'Failed to start VM {vm["name"]}: {e}') @private @accepts( Dict( 'deinitialize_vms_options', Bool('reload_ui', default=True), Bool('stop_libvirt', default=True), ) ) async def deinitialize_vms(self, options): await self.middleware.call('vm.close_libvirt_connection') if options['reload_ui']: await self.middleware.call('service.reload', 'http') if options['stop_libvirt']: await self.middleware.call('service.stop', 'libvirtd') @private def close_libvirt_connection(self): if self.LIBVIRT_CONNECTION: with contextlib.suppress(CallError): self._close() @private def setup_details(self): return { 'connected': self._is_connection_alive(), 'connection_initialised': bool(self.LIBVIRT_CONNECTION), 'domains': list(self.vms.keys()), 'libvirt_domains': self._list_domains() if self.LIBVIRT_CONNECTION else None, } @private async def terminate(self): async with SHUTDOWN_LOCK: await self.middleware.call('vm.close_libvirt_connection') @private async def terminate_timeout(self): return max(map(lambda v: v['shutdown_timeout'], await self.middleware.call('vm.query')), default=10) async def __event_system_ready(middleware, event_type, args): """ Method called when system is ready, supposed to start VMs flagged that way. """ await middleware.call('vm.initialize_vms') # we ignore the 'ready' event on an HA system since the failover event plugin # is responsible for starting this service, however, the VMs still need to be # initialized (which is what the above callers are doing) if await middleware.call('failover.licensed'): return middleware.create_task(middleware.call('vm.start_on_boot')) async def __event_system_shutdown(middleware, event_type, args): async def poweroff_stop_vm(vm): if vm['status']['state'] == 'RUNNING': stop_job = await middleware.call('vm.stop', vm['id'], {'force_after_timeout': True}) await stop_job.wait() if stop_job.error: middleware.logger.error('Stopping %r VM failed: %r', vm['name'], stop_job.error) else: try: await middleware.call('vm.poweroff', vm['id']) except Exception: middleware.logger.error('Powering off %r VM failed', vm['name'], exc_info=True) vms = await middleware.call('vm.query', [('status.state', 'in', ACTIVE_STATES)]) if vms: async with SHUTDOWN_LOCK: # FIXME: Why a global lock?? Not needed.... await asyncio_map(poweroff_stop_vm, vms, 16) middleware.logger.debug('VM(s) stopped successfully') # We do this in vm.terminate as well, reasoning for repeating this here is that we don't want to # stop libvirt on middlewared restarts, we only want that to happen if a shutdown has been initiated # and we have cleanly exited await middleware.call('vm.deinitialize_vms', {'reload_ui': False}) async def setup(middleware): # it's _very_ important that we run this before we do # any type of VM initialization. We have to capture the # zfs c_max value before we start manipulating these # sysctls during vm start/stop await middleware.call('sysctl.store_default_arc_max') if await middleware.call('system.ready'): middleware.create_task(middleware.call('vm.initialize_vms', 5)) # We use a short timeout here deliberately middleware.event_subscribe('system.ready', __event_system_ready) middleware.event_subscribe('system.shutdown', __event_system_shutdown)
6,606
Python
.py
135
39.488889
119
0.646438
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,016
utils.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/utils.py
from xml.etree import ElementTree as etree ACTIVE_STATES = ['RUNNING', 'SUSPENDED'] SYSTEM_NVRAM_FOLDER_PATH = '/data/subsystems/vm/nvram' LIBVIRT_URI = 'qemu+unix:///system?socket=/run/truenas_libvirt/libvirt-sock' LIBVIRT_USER = 'libvirt-qemu' NGINX_PREFIX = '/vm/display' def create_element(*args, **kwargs): attribute_dict = kwargs.pop('attribute_dict', {}) element = etree.Element(*args, **kwargs) element.text = attribute_dict.get('text') element.tail = attribute_dict.get('tail') for child in attribute_dict.get('children', []): element.append(child) return element def get_virsh_command_args(): return ['virsh', '-c', LIBVIRT_URI] def convert_pci_id_to_vm_pci_slot(pci_id: str) -> str: return f'pci_{pci_id.replace(".", "_").replace(":", "_")}' def get_vm_nvram_file_name(vm_data: dict) -> str: return f'{vm_data["id"]}_{vm_data["name"]}_VARS.fd' def get_default_status() -> dict: return { 'state': 'ERROR', 'pid': None, 'domain_state': 'ERROR', }
1,044
Python
.py
26
35.769231
76
0.656064
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,017
numeric_set.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/numeric_set.py
class NumericSet: def __call__(self, value): parse_numeric_set(value) def parse_numeric_set(value): if value == '': return [] cpus = {} parts = value.split(',') for part in parts: part = part.split('-') if len(part) == 1: cpu = int(part[0]) cpus[cpu] = None elif len(part) == 2: start = int(part[0]) end = int(part[1]) if start >= end: raise ValueError(f'End of range has to greater that start: {start}-{end}') for cpu in range(start, end + 1): cpus[cpu] = None else: raise ValueError(f'Range has to be in format start-end: {part}') return list(cpus)
743
Python
.py
23
22.956522
90
0.509777
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,018
events.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/events.py
import libvirt import threading from middlewared.service import private, Service from .connection import LibvirtConnectionMixin class VMService(Service, LibvirtConnectionMixin): @private def setup_libvirt_events(self): self._check_setup_connection() def callback(conn, dom, event, detail, opaque): """ 0: 'DEFINED', 1: 'UNDEFINED', 2: 'STARTED', 3: 'SUSPENDED', 4: 'RESUMED', 5: 'STOPPED', 6: 'SHUTDOWN', 7: 'PMSUSPENDED' Above is event mapping for internal reference """ vm_id = dom.name().split('_')[0] vm = None if vm_id.isdigit(): if vms := self.middleware.call_sync('vm.query', [['id', '=', int(vm_id)]], {'force_sql_filters': True}): if dom.name() == f'{vms[0]["id"]}_{vms[0]["name"]}': vm = vms[0] vm.pop('devices', None) if vm is None: emit_type = 'REMOVED' elif event == 0: emit_type = 'ADDED' else: emit_type = 'CHANGED' vm_state_mapping = { 0: 'NOSTATE', 1: 'RUNNING', 2: 'BLOCKED', 3: 'SUSPENDED', # Actual libvirt event here is PAUSED 4: 'SHUTDOWN', 5: 'SHUTOFF', 6: 'CRASHED', 7: 'PMSUSPENDED', } try: if event == 1: if emit_type == 'REMOVED': state = 'NOSTATE' else: # We undefine/define domain numerous times based on if vm has any new changes # registered, this is going to reflect that state = 'UPDATING CONFIGURATION' else: state = vm_state_mapping.get(dom.state()[0], 'UNKNOWN') except libvirt.libvirtError: state = 'UNKNOWN' # We do not send an event on removed because that would already be done by vm.delete if vm is not None: vm['status']['state'] = state self.middleware.send_event( 'vm.query', emit_type, id=int(vm_id), fields=vm, state=vm_state_mapping.get(event, 'UNKNOWN') ) def event_loop_execution(): while self.LIBVIRT_CONNECTION and self.LIBVIRT_CONNECTION._o and self.LIBVIRT_CONNECTION.isAlive(): libvirt.virEventRunDefaultImpl() event_thread = threading.Thread(target=event_loop_execution, name='libvirt_event_loop') event_thread.setDaemon(True) event_thread.start() self.LIBVIRT_CONNECTION.domainEventRegister(callback, None) self.LIBVIRT_CONNECTION.setKeepAlive(5, 3)
2,926
Python
.py
69
28.086957
120
0.507379
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,019
capabilities.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/capabilities.py
from collections import defaultdict from xml.etree import ElementTree as etree from middlewared.schema import accepts, Dict, returns from middlewared.service import private, Service from .connection import LibvirtConnectionMixin class VMService(Service, LibvirtConnectionMixin): CAPABILITIES = None @private def update_capabilities_cache(self): self._check_setup_connection() xml = etree.fromstring(self.LIBVIRT_CONNECTION.getCapabilities()) supported_archs = defaultdict(list) for guest in xml.findall('guest'): arch = guest.find('arch') if not arch or not arch.get('name'): continue arch_name = arch.get('name') for machine_type in filter(lambda m: m.text, arch.findall('machine')): supported_archs[arch_name].append(machine_type.text) self.CAPABILITIES = supported_archs @accepts(roles=['VM_READ']) @returns(Dict( additional_attrs=True, example={ 'x86_64': ['pc-i440fx-5.2', 'pc-q35-5.2', 'pc-i440fx-2.7'], 'i686': ['pc-i440fx-3.0', 'xenfv'], } )) async def guest_architecture_and_machine_choices(self): """ Retrieve choices for supported guest architecture types and machine choices. Keys in the response would be supported guest architecture(s) on the host and their respective values would be supported machine type(s) for the specific architecture on the host. """ if not self.CAPABILITIES: await self.middleware.call('vm.update_capabilities_cache') return self.CAPABILITIES
1,656
Python
.py
37
36.351351
115
0.668117
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,020
vms.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/vms.py
import asyncio import errno import functools import os import re import shlex import shutil import uuid import middlewared.sqlalchemy as sa from middlewared.plugins.zfs_.utils import zvol_path_to_name from middlewared.schema import accepts, Bool, Dict, Int, List, Patch, returns, Str, ValidationErrors from middlewared.service import CallError, CRUDService, item_method, job, private from middlewared.validators import Range, UUID from middlewared.plugins.vm.numeric_set import parse_numeric_set, NumericSet from .utils import ACTIVE_STATES, get_default_status, get_vm_nvram_file_name, SYSTEM_NVRAM_FOLDER_PATH from .vm_supervisor import VMSupervisorMixin BOOT_LOADER_OPTIONS = { 'UEFI': 'UEFI', 'UEFI_CSM': 'Legacy BIOS', } LIBVIRT_LOCK = asyncio.Lock() RE_NAME = re.compile(r'^[a-zA-Z_0-9]+$') class VMModel(sa.Model): __tablename__ = 'vm_vm' id = sa.Column(sa.Integer(), primary_key=True) name = sa.Column(sa.String(150)) description = sa.Column(sa.String(250)) vcpus = sa.Column(sa.Integer(), default=1) memory = sa.Column(sa.Integer()) min_memory = sa.Column(sa.Integer(), nullable=True) autostart = sa.Column(sa.Boolean(), default=False) time = sa.Column(sa.String(5), default='LOCAL') bootloader = sa.Column(sa.String(50), default='UEFI') cores = sa.Column(sa.Integer(), default=1) threads = sa.Column(sa.Integer(), default=1) hyperv_enlightenments = sa.Column(sa.Boolean(), default=False) shutdown_timeout = sa.Column(sa.Integer(), default=90) cpu_mode = sa.Column(sa.Text()) cpu_model = sa.Column(sa.Text(), nullable=True) cpuset = sa.Column(sa.Text(), default=None, nullable=True) nodeset = sa.Column(sa.Text(), default=None, nullable=True) pin_vcpus = sa.Column(sa.Boolean(), default=False) hide_from_msr = sa.Column(sa.Boolean(), default=False) suspend_on_snapshot = sa.Column(sa.Boolean(), default=False) ensure_display_device = sa.Column(sa.Boolean(), default=True) arch_type = sa.Column(sa.String(255), default=None, nullable=True) machine_type = sa.Column(sa.String(255), default=None, nullable=True) uuid = sa.Column(sa.String(255)) command_line_args = sa.Column(sa.Text(), default='', nullable=False) bootloader_ovmf = sa.Column(sa.String(1024), default='OVMF_CODE.fd') trusted_platform_module = sa.Column(sa.Boolean(), default=False) enable_cpu_topology_extension = sa.Column(sa.Boolean(), default=False) @functools.cache def ovmf_options(): return [path for path in os.listdir('/usr/share/OVMF') if re.findall(r'^OVMF_CODE.*.fd', path)] class VMService(CRUDService, VMSupervisorMixin): class Config: namespace = 'vm' datastore = 'vm.vm' datastore_extend = 'vm.extend_vm' datastore_extend_context = 'vm.extend_context' cli_namespace = 'service.vm' role_prefix = 'VM' ENTRY = Patch( 'vm_create', 'vm_entry', ('add', List('devices')), ('add', Dict( 'status', Str('state', required=True), Int('pid', null=True, required=True), Str('domain_state', required=True), )), ('add', Bool('display_available')), ('add', Int('id')), ) @accepts(roles=['VM_READ']) @returns(Dict(additional_attrs=True)) def bootloader_ovmf_choices(self): """ Retrieve bootloader ovmf choices """ return {path: path for path in ovmf_options()} @private def extend_context(self, rows, extra): status = {} shutting_down = self.middleware.call_sync('system.state') == 'SHUTTING_DOWN' kvm_supported = self._is_kvm_supported() if shutting_down is False and rows and kvm_supported: self._safely_check_setup_connection(5) libvirt_running = shutting_down is False and self._is_connection_alive() for row in rows: status[row['id']] = self.status_impl(row) if libvirt_running else get_default_status() return { 'status': status, } @accepts(roles=['VM_READ']) @returns(Dict( *[Str(k, enum=[v]) for k, v in BOOT_LOADER_OPTIONS.items()], )) async def bootloader_options(self): """ Supported motherboard firmware options. """ return BOOT_LOADER_OPTIONS @private async def extend_vm(self, vm, context): vm['devices'] = await self.middleware.call( 'vm.device.query', [('vm', '=', vm['id'])], {'force_sql_filters': True}, ) vm['display_available'] = any(device['dtype'] == 'DISPLAY' for device in vm['devices']) vm['status'] = context['status'][vm['id']] return vm @accepts(Dict( 'vm_create', Str('command_line_args', default=''), Str('cpu_mode', default='CUSTOM', enum=[ 'CUSTOM', 'HOST-MODEL', 'HOST-PASSTHROUGH']), Str('cpu_model', default=None, null=True), Str('name', required=True), Str('description'), Int('vcpus', default=1), Int('cores', default=1), Int('threads', default=1), Str('cpuset', default=None, null=True, validators=[NumericSet()]), Str('nodeset', default=None, null=True, validators=[NumericSet()]), Bool('enable_cpu_topology_extension', default=False), Bool('pin_vcpus', default=False), Bool('suspend_on_snapshot', default=False), Bool('trusted_platform_module', default=False), Int('memory', required=True, validators=[Range(min_=20)]), Int('min_memory', null=True, validators=[Range(min_=20)], default=None), Bool('hyperv_enlightenments', default=False), Str('bootloader', enum=list(BOOT_LOADER_OPTIONS.keys()), default='UEFI'), Str('bootloader_ovmf', default='OVMF_CODE.fd'), Bool('autostart', default=True), Bool('hide_from_msr', default=False), Bool('ensure_display_device', default=True), Str('time', enum=['LOCAL', 'UTC'], default='LOCAL'), Int('shutdown_timeout', default=90, validators=[Range(min_=5, max_=300)]), Str('arch_type', null=True, default=None), Str('machine_type', null=True, default=None), Str('uuid', null=True, default=None, validators=[UUID()]), register=True, )) async def do_create(self, data): """ Create a Virtual Machine (VM). Maximum of 16 guest virtual CPUs are allowed. By default, every virtual CPU is configured as a separate package. Multiple cores can be configured per CPU by specifying `cores` attributes. `vcpus` specifies total number of CPU sockets. `cores` specifies number of cores per socket. `threads` specifies number of threads per core. `ensure_display_device` when set ( the default ) will ensure that the guest always has access to a video device. For headless installations like ubuntu server this is required for the guest to operate properly. However for cases where consumer would like to use GPU passthrough and does not want a display device added should set this to `false`. `arch_type` refers to architecture type and can be specified for the guest. By default the value is `null` and system in this case will choose a reasonable default based on host. `machine_type` refers to machine type of the guest based on the architecture type selected with `arch_type`. By default the value is `null` and system in this case will choose a reasonable default based on `arch_type` configuration. `shutdown_timeout` indicates the time in seconds the system waits for the VM to cleanly shutdown. During system shutdown, if the VM hasn't exited after a hardware shutdown signal has been sent by the system within `shutdown_timeout` seconds, system initiates poweroff for the VM to stop it. `hide_from_msr` is a boolean which when set will hide the KVM hypervisor from standard MSR based discovery and is useful to enable when doing GPU passthrough. `hyperv_enlightenments` can be used to enable subset of predefined Hyper-V enlightenments for Windows guests. These enlightenments improve performance and enable otherwise missing features. `suspend_on_snapshot` is a boolean attribute which when enabled will automatically pause/suspend VMs when a snapshot is being taken for periodic snapshot tasks. For manual snapshots, if user has specified vms to be paused, they will be in that case. """ async with LIBVIRT_LOCK: await self.middleware.run_in_thread(self._check_setup_connection) verrors = ValidationErrors() await self.common_validation(verrors, 'vm_create', data) verrors.check() vm_id = await self.middleware.call('datastore.insert', 'vm.vm', data) await self.middleware.run_in_thread(self._add, vm_id) await self.middleware.call('etc.generate', 'libvirt_guests') return await self.get_instance(vm_id) @private async def common_validation(self, verrors, schema_name, data, old=None): if data['bootloader_ovmf'] not in await self.middleware.call('vm.bootloader_ovmf_choices'): verrors.add( f'{schema_name}.bootloader_ovmf', 'Invalid bootloader ovmf choice specified' ) if not data.get('uuid'): data['uuid'] = str(uuid.uuid4()) if not await self.middleware.call('vm.license_active'): verrors.add( f'{schema_name}.name', 'System is not licensed to use VMs' ) if data['min_memory'] and data['min_memory'] > data['memory']: verrors.add( f'{schema_name}.min_memory', 'Minimum memory should not be greater than defined/maximum memory' ) try: shlex.split(data['command_line_args']) except ValueError as e: verrors.add( f'{schema_name}.command_line_args', f'Parse error: {e.args[0]}' ) vcpus = data['vcpus'] * data['cores'] * data['threads'] if vcpus: flags = await self.middleware.call('vm.flags') max_vcpus = await self.middleware.call('vm.maximum_supported_vcpus') if vcpus > max_vcpus: verrors.add( f'{schema_name}.vcpus', f'Maximum {max_vcpus} vcpus are supported.' f'Please ensure the product of "{schema_name}.vcpus", "{schema_name}.cores" and ' f'"{schema_name}.threads" is less then {max_vcpus}.' ) elif flags['intel_vmx']: if vcpus > 1 and flags['unrestricted_guest'] is False: verrors.add( f'{schema_name}.vcpus', 'Only one Virtual CPU is allowed in this system.') elif flags['amd_rvi']: if vcpus > 1 and flags['amd_asids'] is False: verrors.add( f'{schema_name}.vcpus', 'Only one virtual CPU is allowed in this system.' ) elif not await self.middleware.call('vm.supports_virtualization'): verrors.add( schema_name, 'This system does not support virtualization.' ) if data.get('arch_type') or data.get('machine_type'): choices = await self.middleware.call('vm.guest_architecture_and_machine_choices') if data.get('arch_type') and data['arch_type'] not in choices: verrors.add(f'{schema_name}.arch_type', 'Specified architecture type is not supported on this system') if data.get('machine_type'): if not data.get('arch_type'): verrors.add( f'{schema_name}.arch_type', f'Must be specified when "{schema_name}.machine_type" is set' ) elif data['arch_type'] in choices and data['machine_type'] not in choices[data['arch_type']]: verrors.add( f'{schema_name}.machine_type', f'Specified machine type is not supported for {choices[data["arch_type"]]!r} architecture type' ) if data.get('cpu_mode') != 'CUSTOM' and data.get('cpu_model'): verrors.add( f'{schema_name}.cpu_model', 'This attribute should not be specified when "cpu_mode" is not "CUSTOM".' ) elif data.get('cpu_model') and data['cpu_model'] not in await self.middleware.call('vm.cpu_model_choices'): verrors.add(f'{schema_name}.cpu_model', 'Please select a valid CPU model.') if 'name' in data: filters = [('name', '=', data['name'])] if old: filters.append(('id', '!=', old['id'])) if await self.middleware.call('vm.query', filters): verrors.add( f'{schema_name}.name', 'This name already exists.', errno.EEXIST ) elif not RE_NAME.search(data['name']): verrors.add( f'{schema_name}.name', 'Only alphanumeric characters are allowed.' ) if data['pin_vcpus']: if not data['cpuset']: verrors.add( f'{schema_name}.cpuset', f'Must be specified when "{schema_name}.pin_vcpus" is set.' ) elif len(parse_numeric_set(data['cpuset'])) != vcpus: verrors.add( f'{schema_name}.pin_vcpus', f'Number of cpus in "{schema_name}.cpuset" must be equal to total number vpcus if pinning is enabled.' ) # TODO: Let's please implement PCI express hierarchy as the limit on devices in KVM is quite high # with reports of users having thousands of disks # Let's validate that the VM has the correct no of slots available to accommodate currently configured devices @accepts( Int('id', required=True), Patch( 'vm_entry', 'vm_update', ('rm', {'name': 'devices'}), ('rm', {'name': 'display_available'}), ('rm', {'name': 'status'}), ('attr', {'update': True}), ) ) async def do_update(self, id_, data): """ Update all information of a specific VM. `devices` is a list of virtualized hardware to attach to the virtual machine. If `devices` is not present, no change is made to devices. If either the device list order or data stored by the device changes when the attribute is passed, these actions are taken: 1) If there is no device in the `devices` list which was previously attached to the VM, that device is removed from the virtual machine. 2) Devices are updated in the `devices` list when they contain a valid `id` attribute that corresponds to an existing device. 3) Devices that do not have an `id` attribute are created and attached to `id` VM. """ old = await self.get_instance(id_) new = old.copy() new.update(data) if new['name'] != old['name']: await self.middleware.run_in_thread(self._check_setup_connection) if old['status']['state'] in ACTIVE_STATES: raise CallError('VM name can only be changed when VM is inactive') if old['name'] not in self.vms: raise CallError(f'Unable to locate domain for {old["name"]}') verrors = ValidationErrors() await self.common_validation(verrors, 'vm_update', new, old=old) verrors.check() for key in ('devices', 'status', 'display_available'): new.pop(key, None) await self.middleware.call('datastore.update', 'vm.vm', id_, new) vm_data = await self.get_instance(id_) if new['name'] != old['name']: await self.middleware.run_in_thread(self._rename_domain, old, vm_data) try: new_path = os.path.join(SYSTEM_NVRAM_FOLDER_PATH, get_vm_nvram_file_name(new)) await self.middleware.run_in_thread( os.rename, os.path.join(SYSTEM_NVRAM_FOLDER_PATH, get_vm_nvram_file_name(old)), new_path ) except FileNotFoundError: if old['bootloader'] == new['bootloader'] == 'UEFI': # So we only want to raise an error if bootloader is UEFI because for BIOS # nvram file will not exist and it is fine. If bootloader is changed from # BIOS to UEFI, even then we will not have it and it is fine so we don't want # to raise an error in that case. raise CallError( f'VM name has been updated but nvram file for {old["name"]} does not exist ' f'which can result in {new["name"]} VM not booting properly.' ) if old['shutdown_timeout'] != new['shutdown_timeout']: await self.middleware.call('etc.generate', 'libvirt_guests') return await self.get_instance(id_) @accepts( Int('id'), Dict( 'vm_delete', Bool('zvols', default=False), Bool('force', default=False), ), ) async def do_delete(self, id_, data): """ Delete a VM. """ async with LIBVIRT_LOCK: vm = await self.get_instance(id_) # Deletion should be allowed even if host does not support virtualization if self._is_kvm_supported(): await self.middleware.run_in_thread(self._check_setup_connection) status = await self.middleware.call('vm.status', id_) else: status = vm['status'] force_delete = data.get('force') if status['state'] in ACTIVE_STATES: await self.middleware.call('vm.poweroff', id_) # We would like to wait at least 7 seconds to have the vm # complete it's post vm actions which might require interaction with it's domain await asyncio.sleep(7) elif status.get('state') == 'ERROR' and not force_delete: raise CallError('Unable to retrieve VM status. Failed to destroy VM') if data['zvols']: devices = await self.middleware.call('vm.device.query', [ ('vm', '=', id_), ('dtype', '=', 'DISK') ]) for zvol in devices: if not zvol['attributes']['path'].startswith('/dev/zvol/'): continue disk_name = zvol_path_to_name(zvol['attributes']['path']) try: await self.middleware.call('zfs.dataset.delete', disk_name, {'recursive': True}) except Exception: if not force_delete: raise else: self.logger.error( 'Failed to delete %r volume when removing %r VM', disk_name, vm['name'], exc_info=True ) try: await self.middleware.run_in_thread(self._undefine_domain, vm['name']) except Exception: if not force_delete: raise else: self.logger.error('Failed to un-define %r VM\'s domain', vm['name'], exc_info=True) # We remove vm devices first for device in vm['devices']: await self.middleware.call('vm.device.delete', device['id'], {'force': data['force']}) result = await self.middleware.call('datastore.delete', 'vm.vm', id_) if not await self.middleware.call('vm.query'): await self.middleware.call('vm.deinitialize_vms') self._clear() else: await self.middleware.call('etc.generate', 'libvirt_guests') return result @item_method @accepts(Int('id'), roles=['VM_READ']) @returns(Dict( 'vm_status', Str('state', required=True), Int('pid', null=True, required=True), Str('domain_state', required=True), )) def status(self, id_): """ Get the status of `id` VM. Returns a dict: - state, RUNNING / STOPPED / SUSPENDED - pid, process id if RUNNING """ vm = self.middleware.call_sync('datastore.query', 'vm.vm', [['id', '=', id_]], {'get': True}) self._check_setup_connection() return self.status_impl(vm) @private def status_impl(self, vm): if self._has_domain(vm['name']): try: # Whatever happens, query shouldn't fail return self._status(vm['name']) except Exception: self.logger.debug('Failed to retrieve VM status for %r', vm['name'], exc_info=True) return get_default_status() @accepts(Int('id'), roles=['VM_READ']) @returns(Str(null=True)) def log_file_path(self, vm_id): """ Retrieve log file path of `id` VM. It will return path of the log file if it exists and `null` otherwise. """ vm = self.middleware.call_sync('vm.get_instance', vm_id) path = f'/var/log/libvirt/qemu/{vm["id"]}_{vm["name"]}.log' return path if os.path.exists(path) else None @accepts(Int('id'), roles=['VM_READ']) @returns() @job(pipes=['output']) def log_file_download(self, job, vm_id): """ Retrieve log file contents of `id` VM. It will download empty file if log file does not exist. """ if path := self.log_file_path(vm_id): with open(path, 'rb') as f: shutil.copyfileobj(f, job.pipes.output.w)
22,323
Python
.py
454
37.817181
122
0.589064
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,021
vm_supervisor.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/vm_supervisor.py
import contextlib from middlewared.service import CallError from .connection import LibvirtConnectionMixin from .supervisor import VMSupervisor from .utils import ACTIVE_STATES class VMSupervisorMixin(LibvirtConnectionMixin): vms = {} def _add(self, vm_id): vm = self.middleware.call_sync('vm.get_instance', vm_id) self._add_with_vm_data(vm) def _add_with_vm_data(self, vm): self.vms[vm['name']] = VMSupervisor(vm, self.middleware) def _has_domain(self, vm_name): return vm_name in self.vms and self.vms[vm_name].domain def _rename_domain(self, old, new): vm = self.vms.pop(old['name']) vm.update_domain(new) self.vms[new['name']] = vm def _clear(self): VMSupervisorMixin.vms = {} def _vm_from_name(self, vm_name): return self.middleware.call_sync('vm.query', [['name', '=', vm_name]], {'get': True, 'force_sql_filters': True}) def _undefine_domain(self, vm_name): domain = self.vms.pop(vm_name, None) if domain and domain.domain: domain.undefine_domain() else: VMSupervisor(self._vm_from_name(vm_name), self.middleware).undefine_domain() def _check_add_domain(self, vm_name): if not self._has_domain(vm_name): try: self._add(self._vm_from_name(vm_name)['id']) except Exception as e: raise CallError(f'Unable to define domain for {vm_name}: {e}') if not self._has_domain(vm_name): raise CallError(f'Libvirt domain for {vm_name} does not exist') def _check_domain_status(self, vm_name, desired_status='RUNNING'): if not self._has_domain(vm_name): raise CallError(f'Libvirt Domain for {vm_name} does not exist') desired_status = desired_status if isinstance(desired_status, list) else [desired_status] configured_status = 'ERROR' with contextlib.suppress(Exception): configured_status = self._status(vm_name)['state'] if configured_status == 'ERROR': raise CallError(f'Unable to determine {vm_name!r} VM state') if configured_status not in desired_status: raise CallError(f'VM state is currently not {" / ".join(desired_status)!r}') def _start(self, vm_name): self._check_add_domain(vm_name) self.vms[vm_name].start(vm_data=self._vm_from_name(vm_name)) def _poweroff(self, vm_name): self._check_domain_status(vm_name, ACTIVE_STATES) self.vms[vm_name].poweroff() def _stop(self, vm_name, shutdown_timeout): self._check_domain_status(vm_name) self.vms[vm_name].stop(shutdown_timeout) def _suspend(self, vm_name): self._check_domain_status(vm_name) self.vms[vm_name].suspend() def _resume(self, vm_name): self._check_domain_status(vm_name, 'SUSPENDED') self.vms[vm_name].resume() def _status(self, vm_name): self._check_setup_connection() return self.vms[vm_name].status() def _memory_info(self, vm_name): self._check_setup_connection() self._check_domain_status(vm_name, ACTIVE_STATES) return self.vms[vm_name].memory_usage()
3,234
Python
.py
69
38.449275
120
0.640242
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,022
connection.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/connection.py
import contextlib import libvirt import os from middlewared.service import CallError from .utils import LIBVIRT_URI class LibvirtConnectionMixin: LIBVIRT_CONNECTION = None KVM_SUPPORTED = None def _open(self): try: # We want to do this before initializing libvirt connection libvirt.virEventRegisterDefaultImpl() LibvirtConnectionMixin.LIBVIRT_CONNECTION = libvirt.open(LIBVIRT_URI) except libvirt.libvirtError as e: raise CallError(f'Failed to open libvirt connection: {e}') def _close(self): try: self.LIBVIRT_CONNECTION.close() except libvirt.libvirtError as e: raise CallError(f'Failed to close libvirt connection: {e}') else: LibvirtConnectionMixin.LIBVIRT_CONNECTION = None def _is_kvm_supported(self): # We check if /dev/kvm exists to ensure that kvm can be consumed on this machine. # Libvirt will still start even if kvm cannot be used on the machine which would falsely # give the impression that virtualization can be used. We have checks in place to check if system # supports virtualization but if we incorporate that check in all of the vm exposed methods which # consume libvirt, it would be an expensive call as we figure that out by making a subprocess call if self.KVM_SUPPORTED is None: self.KVM_SUPPORTED = os.path.exists('/dev/kvm') return self.KVM_SUPPORTED def _is_libvirt_connection_alive(self): with contextlib.suppress(libvirt.libvirtError): # We see isAlive call failed for a user in NAS-109072, it would be better # if we handle this to ensure that system recognises libvirt connection # is no longer active and a new one should be initiated. return ( self.LIBVIRT_CONNECTION and self.LIBVIRT_CONNECTION.isAlive() and isinstance(self.LIBVIRT_CONNECTION.listAllDomains(), list) ) return False def _list_domains(self): with contextlib.suppress(libvirt.libvirtError): return {domain.name(): domain.state() for domain in self.LIBVIRT_CONNECTION.listAllDomains()} def _is_connection_alive(self): return self._is_kvm_supported() and self._is_libvirt_connection_alive() def _system_supports_virtualization(self): if not self._is_kvm_supported(): raise CallError('This system does not support virtualization.') def _check_connection_alive(self): self._system_supports_virtualization() if not self._is_libvirt_connection_alive(): raise CallError('Failed to connect to libvirt') def _safely_check_setup_connection(self, timeout: int = 10): if not self._is_connection_alive(): self.middleware.call_sync('vm.setup_libvirt_connection', timeout) def _check_setup_connection(self): self._safely_check_setup_connection() self._check_connection_alive()
3,040
Python
.py
59
42.474576
106
0.685098
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,023
disk_utils.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/disk_utils.py
import errno import os import re import shlex import subprocess from middlewared.plugins.zfs_.utils import zvol_name_to_path from middlewared.schema import accepts, Bool, Dict, returns, Str from middlewared.service import CallError, Service, job # Valid Disk Formats we can export VALID_DISK_FORMATS = ['qcow2', 'qed', 'raw', 'vdi', 'vpc', 'vmdk' ] class VMService(Service): @accepts(Dict( 'vm_info', Str('diskimg', required=True), Str('zvol', required=True) )) @returns(Bool()) @job(lock_queue_size=1, lock=lambda args: f"zvol_disk_image_{args[-1]['zvol']}") def import_disk_image(self, job, data): """ Imports a specified disk image. Utilized qemu-img with the auto-detect functionality to auto-convert any supported disk image format to RAW -> ZVOL As of this implementation it supports: - QCOW2 - QED - RAW - VDI - VPC - VMDK `diskimg` is an required parameter for the incoming disk image `zvol` is the required target for the imported disk image """ if not self.middleware.call_sync('zfs.dataset.query', [('id', '=', data['zvol'])]): raise CallError(f"zvol {data['zvol']} does not exist.", errno.ENOENT) if os.path.exists(data['diskimg']) is False: raise CallError('Disk Image does not exist.', errno.ENOENT) if os.path.exists(zvol_name_to_path(data['zvol'])) is False: raise CallError('Zvol device does not exist.', errno.ENOENT) zvol_device_path = str(zvol_name_to_path(data['zvol'])) # Use quotes safely and assemble the command imgsafe = shlex.quote(data['diskimg']) devsafe = shlex.quote(zvol_device_path) command = f"qemu-img convert -p -O raw {imgsafe} {devsafe}" self.logger.warning('Running Disk Import using: "' + command + '"') cp = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, bufsize=1, universal_newlines=True) re_progress = re.compile(r'(\d+\.\d+)') stderr = '' for line in iter(cp.stdout.readline, ""): progress = re_progress.search(line.lstrip()) if progress: try: progress = round(float(progress.group(1))) job.set_progress(progress, "Disk Import Progress") except ValueError: self.logger.warning('Invalid progress in: "' + progress.group(1) + '"') else: stderr += line self.logger.warning('No progress reported from qemu-img: "' + line.lstrip() + '"') cp.wait() if cp.returncode: raise CallError(f'Failed to import disk: {stderr}') return True @accepts(Dict( 'vm_info', Str('format', required=True), Str('directory', required=True), Str('zvol', required=True) )) @returns(Bool()) @job(lock_queue_size=1, lock=lambda args: f"zvol_disk_image_{args[-1]['zvol']}") def export_disk_image(self, job, data): """ Exports a zvol to a formatted VM disk image. Utilized qemu-img with the conversion functionality to export a zvol to any supported disk image format, from RAW -> ${OTHER}. The resulting file will be set to inherit the permissions of the target directory. As of this implementation it supports the following {format} options : - QCOW2 - QED - RAW - VDI - VPC - VMDK `format` is an required parameter for the exported disk image `directory` is an required parameter for the export disk image `zvol` is the source for the disk image """ if not self.middleware.call_sync('zfs.dataset.query', [('id', '=', data['zvol'])]): raise CallError(f"zvol {data['zvol']} does not exist.", errno.ENOENT) if os.path.isdir(data['directory']) is False: raise CallError(f"Export directory {data['directory']} does not exist.", errno.ENOENT) if os.path.exists(zvol_name_to_path(data['zvol'])) is False: raise CallError('Zvol device does not exist.', errno.ENOENT) # Check that a supported format was specified format = data['format'].lower() if format not in VALID_DISK_FORMATS: raise CallError('Invalid disk format specified.', errno.ENOENT) # Grab the owner / group of the parent directory parent_stat = os.stat(data['directory']) owner = parent_stat.st_uid group = parent_stat.st_gid # Get the raw zvol device path zvol_device_path = str(zvol_name_to_path(data['zvol'])) # Set the target file location zvolbasename = os.path.basename(data['zvol']) targetfile = f"{data['directory']}/vmdisk-{zvolbasename}.{format}" # Use quotes safely and assemble the command filesafe = shlex.quote(targetfile) devsafe = shlex.quote(zvol_device_path) command = f"qemu-img convert -p -f raw -O {data['format']} {devsafe} {filesafe}" self.logger.warning('Running Disk export using: "' + command + '"') cp = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, bufsize=1, universal_newlines=True) re_progress = re.compile(r'(\d+\.\d+)') stderr = '' for line in iter(cp.stdout.readline, ""): progress = re_progress.search(line.lstrip()) if progress: try: progress = round(float(progress.group(1))) job.set_progress(progress, "Disk Export Progress") except ValueError: self.logger.warning('Invalid progress in: "' + progress.group(1) + '"') else: stderr += line self.logger.warning('No progress reported from qemu-img: "' + line.lstrip() + '"') cp.wait() if cp.returncode: raise CallError(f'Failed to export disk: {stderr}') # Set the owner / group of the target file to inherit that of the saved parent directory os.chown(targetfile, owner, group) return True
6,295
Python
.py
132
37.856061
136
0.612492
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,024
attachments.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/attachments.py
import collections import os.path from middlewared.common.attachment import FSAttachmentDelegate from middlewared.common.ports import PortDelegate from middlewared.plugins.zfs_.utils import zvol_path_to_name from middlewared.service import private, Service from .utils import ACTIVE_STATES async def determine_recursive_search(recursive, device, child_datasets): # TODO: Add unit tests for this please if recursive: return True elif device['dtype'] == 'DISK': return False # What we want to do here is make sure that any raw files or cdrom files are not living in the child # dataset and not affected by the parent snapshot as they live on a different filesystem path = device['attributes']['path'].removeprefix('/mnt/') for split_count in range(path.count('/')): potential_ds = path.rsplit('/', split_count)[0] if potential_ds in child_datasets: return False else: return True class VMService(Service): @private async def periodic_snapshot_task_begin(self, task_id): task = await self.middleware.call('pool.snapshottask.query', [['id', '=', task_id]], {'get': True}) return await self.query_snapshot_begin(task['dataset'], task['recursive']) @private async def query_snapshot_begin(self, dataset, recursive): vms = collections.defaultdict(list) datasets = { d['id']: d for d in await self.middleware.call( 'pool.dataset.query', [['id', '^', f'{dataset}/']], {'extra': {'properties': []}} ) } to_ignore_vms = await self.get_vms_to_ignore_for_querying_attachments(True, [['suspend_on_snapshot', '=', False]]) for device in await self.middleware.call( 'vm.device.query', [ ['dtype', 'in', ('DISK', 'RAW', 'CDROM')], ['vm', 'nin', to_ignore_vms], ] ): path = device['attributes'].get('path') if not path: continue elif path.startswith('/dev/zvol'): path = os.path.join('/mnt', zvol_path_to_name(path)) dataset_path = os.path.join('/mnt', dataset) if await determine_recursive_search(recursive, device, datasets): if await self.middleware.call('filesystem.is_child', path, dataset_path): vms[device['vm']].append(device) elif dataset_path == path: vms[device['vm']].append(device) return vms @private async def get_vms_to_ignore_for_querying_attachments(self, enabled, extra_filters=None): extra_filters = extra_filters or [] return { vm['id']: vm for vm in await self.middleware.call( 'vm.query', [('status.state', 'nin' if enabled else 'in', ACTIVE_STATES)] + extra_filters ) } class VMFSAttachmentDelegate(FSAttachmentDelegate): name = 'vm' title = 'VM' async def query(self, path, enabled, options=None): vms_attached = [] ignored_vms = await self.middleware.call('vm.get_vms_to_ignore_for_querying_attachments', enabled) for device in await self.middleware.call('datastore.query', 'vm.device'): if (device['dtype'] not in ('DISK', 'RAW', 'CDROM')) or device['vm']['id'] in ignored_vms: continue disk = device['attributes'].get('path') if not disk: continue if disk.startswith('/dev/zvol'): disk = os.path.join('/mnt', zvol_path_to_name(disk)) if await self.middleware.call('filesystem.is_child', disk, path): vm = { 'id': device['vm'].get('id'), 'name': device['vm'].get('name'), } if vm not in vms_attached: vms_attached.append(vm) return vms_attached async def delete(self, attachments): for attachment in attachments: try: await self.middleware.call('vm.stop', attachment['id']) except Exception: self.middleware.logger.warning('Unable to vm.stop %r', attachment['id']) async def toggle(self, attachments, enabled): for attachment in attachments: action = 'vm.start' if enabled else 'vm.stop' try: await self.middleware.call(action, attachment['id']) except Exception: self.middleware.logger.warning('Unable to %s %r', action, attachment['id']) async def stop(self, attachments): await self.toggle(attachments, False) async def start(self, attachments): await self.toggle(attachments, True) class VMPortDelegate(PortDelegate): name = 'vm devices' namespace = 'vm.device' title = 'VM Device Service' async def get_ports(self): ports = [] vms = {vm['id']: vm['name'] for vm in await self.middleware.call('vm.query')} for device in await self.middleware.call('vm.device.query', [['dtype', '=', 'DISPLAY']]): ports.append({ 'description': f'{vms[device["vm"]]!r} VM', 'ports': [ (device['attributes']['bind'], device['attributes']['port']), (device['attributes']['bind'], device['attributes']['web_port']), ] }) return ports async def setup(middleware): middleware.create_task( middleware.call('pool.dataset.register_attachment_delegate', VMFSAttachmentDelegate(middleware)) ) await middleware.call('port.register_attachment_delegate', VMPortDelegate(middleware))
5,702
Python
.py
122
36.409836
122
0.59964
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,025
info.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/info.py
import os import re import subprocess from xml.etree import ElementTree as etree from middlewared.schema import Bool, Dict, Int, returns, Str from middlewared.service import accepts, private, Service from middlewared.utils import run from .connection import LibvirtConnectionMixin from .utils import get_virsh_command_args RE_AMD_NASID = re.compile(r'NASID:.*\((.*)\)') RE_VENDOR_AMD = re.compile(r'AuthenticAMD') RE_VENDOR_INTEL = re.compile(r'GenuineIntel') class VMService(Service, LibvirtConnectionMixin): CPU_MODEL_CHOICES = {} @accepts(roles=['VM_READ']) @returns(Bool()) def supports_virtualization(self): """ Returns "true" if system supports virtualization, "false" otherwise """ return self._is_kvm_supported() @private async def license_active(self): """ If this is HA capable hardware and has NOT been licensed to run VMs then this will return False. Otherwise this will return true. """ can_run_vms = True if await self.middleware.call('system.is_ha_capable'): license_ = await self.middleware.call('system.license') can_run_vms = license_ is not None and 'VM' in license_['features'] return can_run_vms @accepts(roles=['VM_READ']) @returns(Dict( Bool('supported', required=True), Str('error', null=True, required=True), )) def virtualization_details(self): """ Retrieve details if virtualization is supported on the system and in case why it's not supported if it isn't. """ return { 'supported': self._is_kvm_supported(), 'error': None if self._is_kvm_supported() else 'Your CPU does not support KVM extensions', } @accepts(roles=['VM_READ']) @returns(Int()) async def maximum_supported_vcpus(self): """ Returns maximum supported VCPU's """ return 255 @accepts(roles=['VM_READ']) @returns(Dict( 'cpu_flags', Bool('intel_vmx', required=True), Bool('unrestricted_guest', required=True), Bool('amd_rvi', required=True), Bool('amd_asids', required=True), )) async def flags(self): """ Returns a dictionary with CPU flags for the hypervisor. """ flags = { 'intel_vmx': False, 'unrestricted_guest': False, 'amd_rvi': False, 'amd_asids': False, } supports_vm = await self.middleware.call('vm.supports_virtualization') if not supports_vm: return flags cp = await run(['lscpu'], check=False) if cp.returncode: self.middleware.logger.error('Failed to retrieve CPU details: %s', cp.stderr.decode()) return flags if RE_VENDOR_INTEL.findall(cp.stdout.decode()): flags['intel_vmx'] = True unrestricted_guest_path = '/sys/module/kvm_intel/parameters/unrestricted_guest' if os.path.exists(unrestricted_guest_path): with open(unrestricted_guest_path, 'r') as f: flags['unrestricted_guest'] = f.read().strip().lower() == 'y' elif RE_VENDOR_AMD.findall(cp.stdout.decode()): flags['amd_rvi'] = True cp = await run(['cpuid', '-l', '0x8000000A'], check=False) if cp.returncode: self.middleware.logger.error('Failed to execute "cpuid -l 0x8000000A": %s', cp.stderr.decode()) else: flags['amd_asids'] = all(v != '0' for v in (RE_AMD_NASID.findall(cp.stdout.decode()) or ['0']) if v) return flags @accepts(Int('id')) @returns(Str('console_device')) async def get_console(self, id_): """ Get the console device from a given guest. """ vm = await self.middleware.call('vm.get_instance', id_) return f'{vm["id"]}_{vm["name"]}' @accepts(roles=['VM_READ']) @returns(Dict( additional_attrs=True, example={ '486': '486', 'pentium': 'pentium', } )) def cpu_model_choices(self): """ Retrieve CPU Model choices which can be used with a VM guest to emulate the CPU in the guest. """ self.middleware.call_sync('vm.check_setup_libvirt') base_path = '/usr/share/libvirt/cpu_map' if self.CPU_MODEL_CHOICES or not os.path.exists(base_path): return self.CPU_MODEL_CHOICES mapping = {} with open(os.path.join(base_path, 'index.xml'), 'r') as f: index_xml = etree.fromstring(f.read().strip()) for arch in filter(lambda a: a.tag == 'arch' and a.get('name'), list(index_xml)): cp = subprocess.Popen( get_virsh_command_args() + ['cpu-models', arch.get('name') if arch.get('name') != 'x86' else 'x86_64'], stdout=subprocess.PIPE, stderr=subprocess.PIPE ) stdout = cp.communicate()[0] if cp.returncode: continue mapping.update({m: m for m in filter(bool, stdout.decode().strip().split('\n'))}) self.CPU_MODEL_CHOICES.update(mapping) return self.CPU_MODEL_CHOICES
5,250
Python
.py
129
31.782946
119
0.59902
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,026
domain_xml.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/supervisor/domain_xml.py
import os import shlex from middlewared.plugins.vm.devices import CDROM, DISK, PCI, RAW, DISPLAY, USB from middlewared.plugins.vm.numeric_set import parse_numeric_set from middlewared.plugins.vm.utils import SYSTEM_NVRAM_FOLDER_PATH, get_vm_nvram_file_name from middlewared.utils import Nid from .utils import create_element def domain_children(vm_data, context): children = [ create_element('name', attribute_dict={'text': f'{vm_data["id"]}_{vm_data["name"]}'}), create_element('uuid', attribute_dict={'text': vm_data['uuid']}), create_element('title', attribute_dict={'text': vm_data['name']}), create_element('description', attribute_dict={'text': vm_data['description']}), # OS/boot related xml - returns an iterable os_xml(vm_data), # CPU related xml *cpu_xml(vm_data, context), # Memory related xml *memory_xml(vm_data), # Add features features_xml(vm_data), # Clock offset clock_xml(vm_data), # Command line args commandline_xml(vm_data), # Devices devices_xml(vm_data, context), ] # Wire memory if PCI passthru device is configured # Implicit configuration for now. # # To avoid surprising side effects from implicit configuration, wiring of memory # should preferably be an explicit vm configuration option and trigger error # message if not selected when PCI passthru is configured. # if any(isinstance(device, PCI) for device in context['devices']): children.append( create_element( 'memoryBacking', attribute_dict={ 'children': [ create_element('locked'), ] } ) ) return children def clock_xml(vm_data): timers = [] if vm_data['hyperv_enlightenments']: timers = [create_element('timer', name='hypervclock', present='yes')] return create_element( 'clock', attribute_dict={'children': timers}, offset='localtime' if vm_data['time'] == 'LOCAL' else 'utc' ) def commandline_xml(vm_data): return create_element( 'commandline', xmlns='http://libvirt.org/schemas/domain/qemu/1.0', attribute_dict={ 'children': [create_element('arg', value=arg) for arg in shlex.split(vm_data['command_line_args'])] } ) def cpu_xml(vm_data, context): features = [] if vm_data['cpu_mode'] == 'HOST-PASSTHROUGH': features.append(create_element('cache', mode='passthrough')) if vm_data['enable_cpu_topology_extension']: features.append(create_element('feature', policy='require', name='topoext')) cpu_nodes = [ create_element( 'cpu', attribute_dict={ 'children': [ create_element( 'topology', sockets=str(vm_data['vcpus']), cores=str(vm_data['cores']), threads=str(vm_data['threads']) ), ] + ([ create_element( 'model', fallback='forbid', attribute_dict={'text': vm_data['cpu_model']} ) # Right now this is best effort for the domain to start with specified CPU Model and not fallback # However if some features are missing in the host, qemu will right now still start the domain # and mark them as missing. We should perhaps make this configurable in the future to control # if domain should/should not be started ] if vm_data['cpu_mode'] == 'CUSTOM' and vm_data['cpu_model'] and context['cpu_model_choices'].get( vm_data['cpu_model'] ) else []) + features, }, mode=vm_data['cpu_mode'].lower(), ), # VCPU related xml create_element( 'vcpu', attribute_dict={ 'text': str(vm_data['vcpus'] * vm_data['cores'] * vm_data['threads']), }, **({'cpuset': vm_data['cpuset']} if vm_data['cpuset'] else {}), ) ] if vm_data['pin_vcpus'] and vm_data['cpuset']: cpu_nodes.append(create_element('cputune', attribute_dict={ 'children': [ create_element('vcpupin', vcpu=str(i), cpuset=str(cpu)) for i, cpu in enumerate(parse_numeric_set(vm_data['cpuset'])) ] })) if vm_data['nodeset']: cpu_nodes.append(create_element( 'numatune', attribute_dict={ 'children': [ create_element('memory', nodeset=vm_data['nodeset']), ] }, )) return cpu_nodes def devices_xml(vm_data, context): boot_no = Nid(1) scsi_device_no = Nid(1) usb_controller_no = Nid(1) # nec-xhci is added by default for each domain by libvirt so we update our mapping accordingly usb_controllers = {'nec-xhci': 0} virtual_device_no = Nid(1) devices = [] for device in context['devices']: if isinstance(device, (DISK, CDROM, RAW)): if device.data['attributes'].get('type') == 'VIRTIO': disk_no = virtual_device_no() else: disk_no = scsi_device_no() device_xml = device.xml(disk_number=disk_no, boot_number=boot_no()) elif isinstance(device, USB): device_xml = [] if device.controller_type not in usb_controllers: usb_controllers[device.controller_type] = usb_controller_no() device_xml.append(create_element( 'controller', type='usb', index=str(usb_controllers[device.controller_type]), model=device.controller_type) ) usb_device_xml = device.xml(controller_mapping=usb_controllers) if isinstance(usb_device_xml, (tuple, list)): device_xml.extend(usb_device_xml) else: device_xml.append(usb_device_xml) else: device_xml = device.xml() devices.extend(device_xml if isinstance(device_xml, (tuple, list)) else [device_xml]) spice_server_available = display_device_available = False for device in filter(lambda d: isinstance(d, DISPLAY), context['devices']): display_device_available = True if device.is_spice_type(): spice_server_available = True break if vm_data['ensure_display_device'] and not display_device_available: # We should add a video device if there is no display device configured because most by # default if not all headless servers like ubuntu etc require it to boot devices.append(create_element('video')) if spice_server_available: # We always add spicevmc channel device when a spice display device is available to allow users # to install guest agents for improved vm experience devices.append(create_element( 'channel', type='spicevmc', attribute_dict={ 'children': [create_element('target', type='virtio', name='com.redhat.spice.0')] } )) if vm_data['trusted_platform_module']: devices.append(create_element( 'tpm', model='tpm-crb', attribute_dict={ 'children': [create_element('backend', type='emulator', version='2.0')] }, )) devices.append(create_element('channel', type='unix', attribute_dict={ 'children': [create_element('target', type='virtio', name='org.qemu.guest_agent.0')] })) devices.append(create_element('serial', type='pty')) if vm_data['min_memory']: # memballoon device needs to be added if memory ballooning is enabled devices.append(create_element('memballoon', model='virtio', autodeflate='on')) return create_element('devices', attribute_dict={'children': devices}) def features_xml(vm_data): features = [] if vm_data['hide_from_msr']: features.append( create_element('kvm', attribute_dict={'children': [create_element('hidden', state='on')]}) ) if vm_data['hyperv_enlightenments']: features.append(get_hyperv_xml()) return create_element( 'features', attribute_dict={ 'children': [ create_element('acpi'), create_element('apic'), create_element('msrs', unknown='ignore'), ] + features, } ) # Documentation for each enlightenment can be found from: # https://github.com/qemu/qemu/blob/master/docs/system/i386/hyperv.rst def get_hyperv_xml(): return create_element( 'hyperv', attribute_dict={ 'children': [ create_element('relaxed', state='on'), create_element('vapic', state='on'), create_element('spinlocks', state='on', retries='8191'), create_element('reset', state='on'), create_element('frequencies', state='on'), # All enlightenments under vpindex depend on it. create_element('vpindex', state='on'), create_element('synic', state='on'), create_element('ipi', state='on'), create_element('tlbflush', state='on'), create_element('stimer', state='on') ], } ) def memory_xml(vm_data): memory_xml_nodes = [create_element('memory', unit='M', attribute_dict={'text': str(vm_data['memory'])})] # Memory Ballooning - this will be memory which will always be allocated to the VM # If not specified, this defaults to `memory` if vm_data['min_memory']: memory_xml_nodes.append( create_element('currentMemory', unit='M', attribute_dict={'text': str(vm_data['min_memory'])}) ) return memory_xml_nodes def os_xml(vm_data): children = [create_element( 'type', attribute_dict={'text': 'hvm'}, **{ k[:-5]: vm_data[k] for k in filter(lambda t: vm_data[t], ('arch_type', 'machine_type')) } )] if vm_data['bootloader'] == 'UEFI': children.extend([ create_element( 'loader', attribute_dict={'text': f'/usr/share/OVMF/{vm_data["bootloader_ovmf"]}'}, readonly='yes', type='pflash', ), create_element('nvram', attribute_dict={ 'text': os.path.join(SYSTEM_NVRAM_FOLDER_PATH, get_vm_nvram_file_name(vm_data)), }) ]) return create_element('os', attribute_dict={'children': children})
10,684
Python
.py
238
34.44958
117
0.585246
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,027
utils.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/supervisor/utils.py
import enum import libvirt from middlewared.plugins.vm.utils import create_element # noqa class DomainState(enum.Enum): NOSTATE = libvirt.VIR_DOMAIN_NOSTATE RUNNING = libvirt.VIR_DOMAIN_RUNNING BLOCKED = libvirt.VIR_DOMAIN_BLOCKED PAUSED = libvirt.VIR_DOMAIN_PAUSED SHUTDOWN = libvirt.VIR_DOMAIN_SHUTDOWN SHUTOFF = libvirt.VIR_DOMAIN_SHUTOFF CRASHED = libvirt.VIR_DOMAIN_CRASHED PMSUSPENDED = libvirt.VIR_DOMAIN_PMSUSPENDED
460
Python
.py
12
34.416667
63
0.779775
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,028
supervisor.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/supervisor/supervisor.py
import contextlib import itertools import libvirt import os import sys import threading import time from xml.etree import ElementTree as etree from middlewared.service import CallError from middlewared.plugins.vm.connection import LibvirtConnectionMixin from middlewared.plugins.vm.devices import CDROM, DISK, NIC, PCI, RAW, DISPLAY, USB # noqa from middlewared.plugins.vm.utils import ACTIVE_STATES from .domain_xml import domain_children from .utils import create_element, DomainState class VMSupervisor(LibvirtConnectionMixin): def __init__(self, vm_data, middleware=None): self.vm_data = vm_data self.middleware = middleware self.devices = [] self._check_setup_connection() self.libvirt_domain_name = f'{self.vm_data["id"]}_{self.vm_data["name"]}' self._domain = self.stop_devices_thread = None self.update_domain() @property def domain(self): return self.domain_health_check() def domain_health_check(self): try: self._domain.state() except (AttributeError, libvirt.libvirtError): self.update_domain(update_devices=False) return self._domain def update_domain(self, vm_data=None, update_devices=True): # This can be called to update domain to reflect any changes introduced to the VM if update_devices: self.update_vm_data(vm_data) try: self._domain = self.LIBVIRT_CONNECTION.lookupByName(self.libvirt_domain_name) except libvirt.libvirtError: self._domain = None else: if not self._domain.isActive(): # We have a domain defined and it is not running self.undefine_domain(for_update=True) if not self._domain: # This ensures that when a domain has been renamed, we undefine the previous domain name - if object # persists in this case of VMSupervisor - else it's the users responsibility to take care of this case new_name = f'{self.vm_data["id"]}_{self.vm_data["name"]}' if new_name != self.libvirt_domain_name: old_nvram_filename = f'/var/lib/libvirt/qemu/nvram/{self.libvirt_domain_name}_VARS.fd' with contextlib.suppress(FileNotFoundError): os.rename(old_nvram_filename, f'/var/lib/libvirt/qemu/nvram/{new_name}_VARS.fd') self.libvirt_domain_name = new_name self.__define_domain() def status(self): domain = self.domain domain_state = DomainState(domain.state()[0]) pid_path = os.path.join('/var/run/libvirt', 'qemu', f'{self.libvirt_domain_name}.pid') if domain.isActive(): state = 'SUSPENDED' if domain_state == DomainState.PAUSED else 'RUNNING' else: state = 'STOPPED' data = { 'state': state, 'pid': None, 'domain_state': domain_state.name, } if domain_state in (DomainState.PAUSED, DomainState.RUNNING): with contextlib.suppress(FileNotFoundError): # Do not make a stat call to check if file exists or not with open(pid_path, 'r') as f: data['pid'] = int(f.read()) return data def memory_usage(self): # We return this in bytes return self.domain.memoryStats().get('actual', 0) * 1024 def __define_domain(self): if self._domain: raise CallError(f'{self.libvirt_domain_name} domain has already been defined') vm_xml = etree.tostring(self.construct_xml()).decode() if not self.LIBVIRT_CONNECTION.defineXML(vm_xml): raise CallError(f'Unable to define persistent domain for {self.libvirt_domain_name}') self._domain = self.LIBVIRT_CONNECTION.lookupByName(self.libvirt_domain_name) def undefine_domain(self, for_update=False): if self._domain.isActive(): raise CallError(f'Domain {self.libvirt_domain_name} is active. Please stop it first') flags = 0 if for_update: flags |= libvirt.VIR_DOMAIN_UNDEFINE_KEEP_NVRAM else: flags |= libvirt.VIR_DOMAIN_UNDEFINE_NVRAM self._domain.undefineFlags(flags) self._domain = None def __getattribute__(self, item): retrieved_item = object.__getattribute__(self, item) if callable(retrieved_item) and item in ('start', 'stop', 'poweroff', 'undefine_domain', 'status'): self.domain_health_check() if not getattr(self, '_domain', None): raise RuntimeError('Domain attribute not defined, please re-instantiate the VM class') return retrieved_item def update_vm_data(self, vm_data=None): self.vm_data = vm_data or self.vm_data self.devices = [ getattr(sys.modules[__name__], device['dtype'])(device, self.middleware) for device in sorted(self.vm_data['devices'], key=lambda x: (x['order'], x['id'])) ] def unavailable_devices(self): return [d for d in self.devices if not d.is_available()] def vm_devices_context(self): return { 'vms': self.middleware.call_sync('vm.query'), 'vm_devices': self.middleware.call_sync('vm.device.query'), } def start(self, vm_data=None): if self.domain.isActive(): raise CallError(f'{self.libvirt_domain_name} domain is already active') self.update_vm_data(vm_data) errors = [] context = self.vm_devices_context() for device in self.devices: try: device.pre_start_vm_device_setup(context) except Exception as e: errors.append(str(e)) if errors: errors = '\n'.join(errors) raise CallError(f'Failed setting up devices before VM start:\n{errors}') unavailable_devices = self.unavailable_devices() if unavailable_devices: raise CallError( f'VM will not start as {", ".join([str(d) for d in unavailable_devices])} device(s) are not available.' ) successful = [] errors = [] for device in self.devices: try: device.pre_start_vm() except Exception as e: errors.append(f'Failed to setup {device.data["dtype"]} device: {e}') for d in itertools.chain([device], successful): try: d.pre_start_vm_rollback() except Exception as d_error: errors.append( f'Failed to rollback pre start changes for {d.data["dtype"]} device: {d_error}' ) break else: successful.append(device) if errors: raise CallError('\n'.join(errors)) try: self.update_domain(vm_data, update_devices=False) if self.domain.create() < 0: raise CallError(f'Failed to boot {self.vm_data["name"]} domain') except (libvirt.libvirtError, CallError) as e: errors = [str(e)] for device in self.devices: try: device.pre_start_vm_rollback() except Exception as d_error: errors.append(f'Failed to rollback pre start changes for {device.data["dtype"]} device: {d_error}') raise CallError('\n'.join(errors)) # We initialize this when we are certain that the VM has indeed booted self.stop_devices_thread = threading.Thread( name=f'post_stop_devices_{self.libvirt_domain_name}', target=self.run_post_stop_actions ) self.stop_devices_thread.start() errors = [] for device in self.devices: try: device.post_start_vm() except Exception as e: errors.append(f'Failed to execute post start actions for {device.data["dtype"]} device: {e}') else: if errors: raise CallError('\n'.join(errors)) def _before_stopping_checks(self): if not self.domain.isActive(): raise CallError(f'{self.libvirt_domain_name} domain is not active') def run_post_stop_actions(self): while self.status()['state'] in ACTIVE_STATES: time.sleep(5) errors = [] context = self.vm_devices_context() for device in self.devices: try: device.post_stop_vm(context) except Exception as e: errors.append(f'Failed to execute post stop actions for {device.data["dtype"]} device: {e}') else: if errors: raise CallError('\n'.join(errors)) def stop(self, shutdown_timeout=None): self._before_stopping_checks() self.domain.shutdown() shutdown_timeout = shutdown_timeout or self.vm_data['shutdown_timeout'] # We wait for timeout seconds before initiating post stop activities for the vm # This is done because the shutdown call above is non-blocking while shutdown_timeout > 0 and self.status()['state'] == 'RUNNING': shutdown_timeout -= 5 time.sleep(5) def poweroff(self): self._before_stopping_checks() self.domain.destroy() def suspend(self): self._before_stopping_checks() self.domain.suspend() def _before_resuming_checks(self): if self.status()['state'] != 'SUSPENDED': raise CallError(f'{self.libvirt_domain_name!r} domain is not paused') def resume(self): self._before_resuming_checks() self.domain.resume() def get_domain_children(self): context = { 'cpu_model_choices': self.middleware.call_sync('vm.cpu_model_choices'), 'devices': self.devices, } return domain_children(self.vm_data, context) def construct_xml(self): return create_element( 'domain', type='kvm', id=str(self.vm_data['id']), attribute_dict={'children': self.get_domain_children()} )
10,250
Python
.py
225
34.626667
119
0.603789
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,029
device.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/devices/device.py
from abc import ABC from middlewared.validators import validate_schema class Device(ABC): schema = NotImplemented def __init__(self, data, middleware=None): self.data = data self.middleware = middleware def xml(self, *args, **kwargs): raise NotImplementedError def is_available(self): raise NotImplementedError def pre_start_vm(self, *args, **kwargs): pass def pre_start_vm_rollback(self, *args, **kwargs): pass def post_start_vm(self, *args, **kwargs): pass def post_stop_vm(self, *args, **kwargs): pass def __str__(self): return f'{self.__class__.__name__} Device: {self.identity()}' def identity(self): raise NotImplementedError def pre_start_vm_device_setup(self, *args, **kwargs): pass def validate(self, device, old=None, vm_instance=None, update=True): verrors = validate_schema(list(self.schema.attrs.values()), device['attributes']) verrors.check() self._validate(device, verrors, old, vm_instance, update) verrors.check() def _validate(self, device, verrors, old=None, vm_instance=None, update=True): raise NotImplementedError
1,233
Python
.py
32
31.40625
89
0.652321
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,030
pci.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/devices/pci.py
import subprocess from middlewared.service import CallError from middlewared.schema import Dict, Str from middlewared.utils import filter_list from .device import Device from .utils import ACTIVE_STATES, create_element, LIBVIRT_URI class PCIBase(Device): def is_available(self): return self.get_details()['available'] def in_use_by_vm(self, vms, vm_devices): return any(vm['status']['state'] in ACTIVE_STATES for vm in self.get_vms_using_device(vms, vm_devices)) def get_vms_using_device(self, vms, vm_devices): devs = filter_list(vm_devices, self.vm_device_filters()) return filter_list(vms, [['id', 'in', [dev['vm'] for dev in devs]]]) def vm_device_filters(self): raise NotImplementedError() def pre_start_vm_device_setup(self, context): if self.in_use_by_vm(context['vms'], context['vm_devices']): raise CallError(f'{self.data["dtype"]} device is already being used by another active VM') class PCI(PCIBase): schema = Dict( 'attributes', Str('pptdev', required=True, empty=False), ) def vm_device_filters(self): return [['attributes.pptdev', '=', self.passthru_device()], ['dtype', '=', 'PCI']] def detach_device(self): cp = subprocess.Popen( ['virsh', '-c', LIBVIRT_URI, 'nodedev-detach', self.passthru_device()], stderr=subprocess.PIPE, stdout=subprocess.DEVNULL ) stderr = cp.communicate()[1] if cp.returncode: raise CallError(f'Unable to detach {self.passthru_device()} PCI device: {stderr.decode()}') def reattach_device(self): cp = subprocess.Popen( ['virsh', '-c', LIBVIRT_URI, 'nodedev-reattach', self.passthru_device()], stderr=subprocess.PIPE, stdout=subprocess.DEVNULL ) stderr = cp.communicate()[1] if cp.returncode: raise CallError(f'Unable to re-attach {self.passthru_device()} PCI device: {stderr.decode()}') def pre_start_vm_device_setup(self, *args, **kwargs): super().pre_start_vm_device_setup(*args, **kwargs) device = self.get_details() if not device['error'] and not device['available']: self.detach_device() def identity(self): return str(self.passthru_device()) def passthru_device(self): return str(self.data['attributes']['pptdev']) def post_stop_vm(self, context): # safe to re-attach if not self.get_details()['error'] and not self.in_use_by_vm(context['vms'], context['vm_devices']): try: self.reattach_device() except CallError: self.middleware.logger.error('Failed to re-attach %s device', self.passthru_device(), exc_info=True) def get_details(self): return self.middleware.call_sync('vm.device.passthrough_device', self.passthru_device()) def xml(self, *args, **kwargs): address_info = { k: hex(int(v)) for k, v in self.get_details()['capability'].items() if k in ('domain', 'bus', 'slot', 'function') } return create_element( 'hostdev', mode='subsystem', type='pci', managed='yes', attribute_dict={ 'children': [ create_element('source', attribute_dict={'children': [create_element('address', **address_info)]}), ] } ) def _validate(self, device, verrors, old=None, vm_instance=None, update=True): pptdev = device['attributes'].get('pptdev') device_details = self.middleware.call_sync('vm.device.passthrough_device', pptdev) if device_details['error']: verrors.add( 'attribute.pptdev', f'Not a valid choice. The PCI device is not available for passthru: {device_details["error"]}' ) elif device_details['critical']: verrors.add( 'attribute.pptdev', f'{device_details["controller_type"]!r} based PCI devices are critical for system function ' 'and cannot be used for PCI passthrough' ) if self.middleware.call_sync('system.is_ha_capable'): verrors.add('attribute.pptdev', 'HA capable systems do not support PCI passthrough') if not self.middleware.call_sync('vm.device.iommu_enabled'): verrors.add('attribute.pptdev', 'IOMMU support is required.') if old and vm_instance and vm_instance['status']['state'] in ACTIVE_STATES and old[ 'attributes' ].get('pptdev') != pptdev: verrors.add( 'attribute.pptdev', 'Changing PCI device is not allowed while the VM is active.' )
4,767
Python
.py
97
39.247423
119
0.61555
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,031
cdrom.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/devices/cdrom.py
import os from middlewared.plugins.boot import BOOT_POOL_NAME from middlewared.schema import Dict, File from middlewared.service import CallError from middlewared.utils.zfs import query_imported_fast_impl from middlewared.validators import check_path_resides_within_volume_sync, Match from .device import Device from .utils import create_element, disk_from_number, LIBVIRT_USER class CDROM(Device): schema = Dict( 'attributes', File( 'path', required=True, validators=[ Match( r'^/mnt/[^{}]*$', explanation='Path must not contain "{", "}" characters, and it should start with "/mnt/"' ), ], empty=False ), ) def identity(self): return self.data['attributes']['path'] def is_available(self): return os.path.exists(self.identity()) def xml(self, *args, **kwargs): disk_number = kwargs.pop('disk_number') return create_element( 'disk', type='file', device='cdrom', attribute_dict={ 'children': [ create_element('driver', name='qemu', type='raw'), create_element('source', file=self.data['attributes']['path']), create_element('target', dev=f'sd{disk_from_number(disk_number)}', bus='sata'), create_element('boot', order=str(kwargs.pop('boot_number'))), ] } ) def _validate(self, device, verrors, old=None, vm_instance=None, update=True): path = device['attributes']['path'] check_path_resides_within_volume_sync( verrors, 'attributes.path', path, [ i['name'] for i in query_imported_fast_impl().values() if i['name'] != BOOT_POOL_NAME ] ) if not self.middleware.call_sync('vm.device.disk_uniqueness_integrity_check', device, vm_instance): verrors.add( 'attributes.path', f'{vm_instance["name"]} has "{self.identity()}" already configured' ) if not verrors: # We would like to check now if libvirt will actually be able to read the iso file # How this works is that if libvirt user is not able to read the file, libvirt automatically changes # ownership of the iso file to the libvirt user so that it is able to read however there are cases where # even this can fail with perms like 000 or maybe parent path(s) not allowing access. # To mitigate this, we can do the following: # 1) See if owner of the file is libvirt user # 2) If it's not libvirt user: # a) Check if libvirt user can access the file # b) Change ownership of the file to libvirt user as libvirt would eventually do # 3) Check if libvirt user can access the file libvirt_user = self.middleware.call_sync('user.get_user_obj', {"username": LIBVIRT_USER}) current_owner = os.stat(path) is_valid = False if current_owner.st_uid != libvirt_user['pw_uid']: if self.middleware.call_sync('filesystem.can_access_as_user', LIBVIRT_USER, path, {'read': True}): is_valid = True else: os.chown(path, libvirt_user['pw_uid'], libvirt_user['pw_gid']) if not is_valid: try: self.middleware.call_sync( 'filesystem.check_path_execute', path, 'USER', libvirt_user['pw_uid'], False ) except CallError as e: verrors.add('attributes.path', e.errmsg) if not self.middleware.call_sync( 'filesystem.can_access_as_user', LIBVIRT_USER, path, {'read': True} ): verrors.add( 'attributes.path', f'{LIBVIRT_USER!r} user cannot read from {path!r} path. Please ensure correct ' 'permissions are specified.' ) # Now that we know libvirt user would not be able to read the file in any case, # let's rollback the chown change we did os.chown(path, current_owner.st_uid, current_owner.st_gid)
4,365
Python
.py
85
38.035294
116
0.571562
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,032
usb.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/devices/usb.py
from middlewared.schema import Dict, Str from middlewared.validators import Match from .pci import PCIBase from .utils import create_element USB_CONTROLLER_CHOICES = [ 'piix3-uhci', 'piix4-uhci', 'ehci', 'ich9-ehci1', 'vt82c686b-uhci', 'pci-ohci', 'nec-xhci', 'qemu-xhci', ] class USB(PCIBase): schema = Dict( 'attributes', Dict( 'usb', Str( 'vendor_id', empty=False, required=True, validators=[Match(r'^0x.*')], description='Vendor id must start with "0x" prefix e.g 0x0451' ), Str( 'product_id', empty=False, required=True, validators=[Match(r'^0x.*')], description='Product id must start with "0x" prefix e.g 0x16a8' ), default=None, null=True, ), Str('controller_type', empty=False, default='nec-xhci', enum=USB_CONTROLLER_CHOICES), Str('device', empty=False, null=True, default=None), ) @property def usb_device(self): return self.data['attributes']['device'] @property def controller_type(self): return self.data['attributes']['controller_type'] @property def usb_details(self): return self.data['attributes']['usb'] def identity(self): return self.usb_device or f'{self.usb_details["product_id"]}--{self.usb_details["vendor_id"]}' def vm_device_filters(self): if self.usb_device: return [['attributes.device', '=', self.usb_device], ['dtype', '=', 'USB']] else: return [['attributes.usb', '=', self.usb_details], ['dtype', '=', 'USB']] def get_details(self): usb_device = self.usb_device if not usb_device and self.usb_details: usb_device = self.middleware.call_sync('vm.device.get_usb_port_from_usb_details', self.usb_details) if usb_device: return self.middleware.call_sync('vm.device.usb_passthrough_device', usb_device) else: return { **self.middleware.call_sync('vm.device.get_basic_usb_passthrough_device_data'), 'error': 'Could not find matching device as no usb device has been specified', } def xml(self, *args, **kwargs): controller_mapping = kwargs.pop('controller_mapping') details = self.get_details()['capability'] if self.is_available(): return create_element( 'hostdev', mode='subsystem', type='usb', managed='yes', attribute_dict={ 'children': [ create_element('source', attribute_dict={'children': [ create_element('vendor', id=details['vendor_id']), create_element('product', id=details['product_id']), create_element('address', bus=details['bus'], device=details['device']), ]}), create_element('address', type='usb', bus=str(controller_mapping[self.controller_type])), ] } ) else: return [] def _validate(self, device, verrors, old=None, vm_instance=None, update=True): if device['attributes']['device'] and device['attributes']['usb']: verrors.add( 'attributes.usb', 'Either device must be specified or USB details but not both' ) elif not device['attributes']['device'] and not device['attributes']['usb']: verrors.add( 'attributes.device', 'Either device or attributes.usb must be specified' ) if self.middleware.call_sync('system.is_ha_capable'): verrors.add('attributes.usb', 'HA capable systems do not support USB passthrough.') if verrors: return if device['attributes']['device']: self._validate_usb_port(device, verrors) else: self._validate_usb_details(device, verrors) def _validate_usb_details(self, device, verrors): usb_details = device['attributes']['usb'] if not self.middleware.call_sync('vm.device.get_usb_port_from_usb_details', usb_details): verrors.add( 'attributes.usb', 'Unable to locate USB, please confirm its present in a USB port' ) def _validate_usb_port(self, device, verrors): usb_device = device['attributes']['device'] device_details = self.middleware.call_sync('vm.device.usb_passthrough_device', usb_device) if device_details.get('error'): verrors.add( 'attribute.device', f'Not a valid choice. The device is not available for USB passthrough: {device_details["error"]}' )
4,850
Python
.py
106
34.235849
113
0.57545
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,033
display.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/devices/display.py
import psutil import subprocess from urllib.parse import urlencode, quote_plus from middlewared.schema import Bool, Dict, Int, Password, Str, ValidationErrors from middlewared.validators import Range from .device import Device from .utils import create_element, NGINX_PREFIX class DISPLAY(Device): RESOLUTION_ENUM = [ '1920x1200', '1920x1080', '1600x1200', '1600x900', '1400x1050', '1280x1024', '1280x720', '1024x768', '800x600', '640x480', ] schema = Dict( 'attributes', Str('resolution', enum=RESOLUTION_ENUM, default='1024x768'), Int('port', default=None, null=True, validators=[Range(min_=5900, max_=65535)]), Int('web_port', default=None, null=True, validators=[Range(min_=5900, max_=65535)]), Str('bind', default='127.0.0.1'), Bool('wait', default=False), Password('password', required=True, null=False, empty=False), Bool('web', default=True), Str('type', default='SPICE', enum=['SPICE']), ) def __init__(self, *args, **kwargs): super().__init__(*args, **kwargs) self.web_process = None def identity(self): data = self.data['attributes'] return f'{data["bind"]}:{data["port"]}' def is_spice_type(self): return self.data['attributes']['type'] == 'SPICE' def web_uri(self, host, protocol='http'): path = self.get_webui_info()['path'][1:] params = {'path': path, 'autoconnect': 1} get_params = f'?{urlencode(params, quote_via=quote_plus)}' return f'{protocol}://{host}/{path}spice_auto.html{get_params}' def is_available(self): bind_ip_available = self.data['attributes']['bind'] in self.middleware.call_sync('vm.device.bind_choices') return bind_ip_available and not self.validate_port_attrs(self.data) def resolution(self): return self.data['attributes']['resolution'] def xml(self, *args, **kwargs): # FIXME: Resolution is not respected when we have more then 1 display device as we are not able to bind # video element to a graphic element attrs = self.data['attributes'] return create_element( 'graphics', type='spice', port=str(self.data['attributes']['port']), attribute_dict={ 'children': [ create_element('listen', type='address', address=self.data['attributes']['bind']), ] }, **({} if not attrs['password'] else {'passwd': attrs['password']}) ), create_element( 'controller', type='usb', model='nec-xhci' ), create_element('input', type='tablet', bus='usb'), create_element('video', attribute_dict={ 'children': [ create_element('model', type='qxl', attribute_dict={ 'children': [create_element( 'resolution', x=self.resolution().split('x')[0], y=self.resolution().split('x')[-1] )] }) ] }) def get_start_attrs(self): port = self.data['attributes']['port'] bind = self.data['attributes']['bind'] web_port = self.data['attributes']['web_port'] return { 'web_bind': f':{web_port}' if bind == '0.0.0.0' else f'{bind}:{web_port}', 'server_addr': f'{bind}:{port}' } def post_start_vm(self, *args, **kwargs): start_args = self.get_start_attrs() self.web_process = subprocess.Popen( [ 'websockify', '--web', '/usr/share/spice-html5/', '--wrap-mode=ignore', start_args['web_bind'], start_args['server_addr'] ], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL, ) def post_stop_vm(self, *args, **kwargs): if self.web_process and psutil.pid_exists(self.web_process.pid): self.middleware.call_sync('service.terminate_process', self.web_process.pid) self.web_process = None def get_webui_info(self): return { 'id': self.data['id'], 'path': f'{NGINX_PREFIX}/{self.data["id"]}/', 'redirect_uri': f'{self.data["attributes"]["bind"]}:' f'{self.data["attributes"]["web_port"]}', } def _validate(self, device, verrors, old=None, vm_instance=None, update=True): if vm_instance: if not update: vm_instance['devices'].append(device) self.middleware.call_sync('vm.device.validate_display_devices', verrors, vm_instance) verrors = self.validate_port_attrs(device, verrors) if device['attributes']['bind'] not in self.middleware.call_sync('vm.device.bind_choices'): verrors.add('attributes.bind', 'Requested bind address is not valid') def validate_port_attrs(self, device, verrors=None): verrors = ValidationErrors() if verrors is None else verrors display_devices_ports = self.middleware.call_sync( 'vm.all_used_display_device_ports', [['id', '!=', device.get('id')]] ) new_ports = list((self.middleware.call_sync('vm.port_wizard')).values()) dev_attrs = device['attributes'] for port in filter(lambda p: p in new_ports, (dev_attrs.get('port'), dev_attrs.get('web_port'))): new_ports.remove(port) for key in ('port', 'web_port'): if device['attributes'].get(key): if dev_attrs[key] in display_devices_ports: verrors.add( f'attributes.{key}', f'Specified display port({dev_attrs[key]}) is already in use by another Display device' ) else: verrors.extend(self.middleware.call_sync( 'port.validate_port', f'attributes.{key}', dev_attrs[key], dev_attrs['bind'], 'vm.device' )) else: device['attributes'][key] = new_ports.pop(0) return verrors
6,037
Python
.py
122
38.647541
114
0.580037
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,034
utils.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/devices/utils.py
import string from middlewared.plugins.vm.utils import ACTIVE_STATES, create_element, LIBVIRT_URI, LIBVIRT_USER, NGINX_PREFIX # noqa def disk_from_number(number): def i_divmod(n): a, b = divmod(n, 26) if b == 0: return a - 1, b + 26 return a, b chars = [] while number > 0: number, d = i_divmod(number) chars.append(string.ascii_lowercase[d - 1]) return ''.join(reversed(chars))
451
Python
.py
13
28.153846
118
0.619816
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,035
__init__.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/devices/__init__.py
from .cdrom import CDROM from .nic import NIC from .pci import PCI from .storage_devices import DISK, RAW from .display import DISPLAY from .usb import USB __all__ = ['CDROM', 'DEVICES', 'DISK', 'NIC', 'PCI', 'RAW', 'DISPLAY', 'USB'] DEVICES = { device_class.__name__: device_class for device_class in ( CDROM, DISK, NIC, PCI, RAW, DISPLAY, USB ) }
368
Python
.py
12
28.083333
77
0.674221
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,036
nic.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/devices/nic.py
import random from middlewared.plugins.interface.netif import netif from middlewared.schema import Bool, Dict, Str from middlewared.service import CallError from middlewared.validators import MACAddr from .device import Device from .utils import create_element class NIC(Device): schema = Dict( 'attributes', Bool('trust_guest_rx_filters', default=False), Str('type', enum=['E1000', 'VIRTIO'], default='E1000'), Str('nic_attach', default=None, null=True), Str('mac', default=None, null=True, validators=[MACAddr(separator=':')]), ) def __init__(self, *args, **kwargs): super().__init__(*args, **kwargs) self.bridge = self.bridge_created = self.nic_attach = None def identity(self): nic_attach = self.data['attributes'].get('nic_attach') if not nic_attach: nic_attach = netif.RoutingTable().default_route_ipv4.interface return nic_attach def is_available(self): return self.identity() in netif.list_interfaces() @staticmethod def random_mac(): mac_address = [ 0x00, 0xa0, 0x98, random.randint(0x00, 0x7f), random.randint(0x00, 0xff), random.randint(0x00, 0xff) ] return ':'.join(['%02x' % x for x in mac_address]) def setup_nic_attach(self): nic_attach = self.data['attributes'].get('nic_attach') interfaces = netif.list_interfaces() if nic_attach and nic_attach not in interfaces: raise CallError(f'{nic_attach} not found.') else: if not nic_attach: try: nic_attach = netif.RoutingTable().default_route_ipv4.interface nic = netif.get_interface(nic_attach) except Exception as e: raise CallError(f'Unable to retrieve default interface: {e}') else: nic = netif.get_interface(nic_attach) if netif.InterfaceFlags.UP not in nic.flags: nic.up() self.nic_attach = nic.name def xml_children(self): return [ create_element('model', type='virtio' if self.data['attributes']['type'] == 'VIRTIO' else 'e1000'), create_element( 'mac', address=self.data['attributes']['mac'] if self.data['attributes'].get('mac') else self.random_mac() ), ] def xml(self, *args, **kwargs): self.setup_nic_attach() if self.nic_attach.startswith('br'): return create_element( 'interface', type='bridge', attribute_dict={ 'children': [ create_element('source', bridge=self.nic_attach) ] + self.xml_children() } ) else: trust_guest_rx_filters = 'yes' if self.data['attributes']['trust_guest_rx_filters'] else 'no' return create_element( 'interface', type='direct', trustGuestRxFilters=trust_guest_rx_filters, attribute_dict={ 'children': [ create_element('source', dev=self.nic_attach, mode='bridge') ] + self.xml_children() } ) def _validate(self, device, verrors, old=None, vm_instance=None, update=True): nic = device['attributes'].get('nic_attach') if nic: nic_choices = self.middleware.call_sync('vm.device.nic_attach_choices') if nic not in nic_choices: verrors.add('attributes.nic_attach', 'Not a valid choice.') elif nic.startswith('br') and device['attributes']['trust_guest_rx_filters']: verrors.add( 'attributes.trust_guest_rx_filters', 'This can only be set when "nic_attach" is not a bridge device' ) if device['attributes']['trust_guest_rx_filters'] and device['attributes']['type'] == 'E1000': verrors.add( 'attributes.trust_guest_rx_filters', 'This can only be set when "type" of NIC device is "VIRTIO"' )
4,158
Python
.py
91
34.021978
112
0.574778
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,037
storage_devices.py
truenas_middleware/src/middlewared/middlewared/plugins/vm/devices/storage_devices.py
import errno import os from middlewared.plugins.zfs_.utils import zvol_name_to_path, zvol_path_to_name from middlewared.plugins.zfs_.validation_utils import check_zvol_in_boot_pool_using_path from middlewared.schema import Bool, Dict, Int, Str from middlewared.validators import Match from .device import Device from .utils import create_element, disk_from_number IOTYPE_CHOICES = ['NATIVE', 'THREADS', 'IO_URING'] class StorageDevice(Device): TYPE = NotImplemented def identity(self): return self.data['attributes']['path'] def is_available(self): return os.path.exists(self.identity()) def xml(self, *args, **kwargs): disk_number = kwargs.pop('disk_number') virtio = self.data['attributes']['type'] == 'VIRTIO' logical_sectorsize = self.data['attributes']['logical_sectorsize'] physical_sectorsize = self.data['attributes']['physical_sectorsize'] iotype = self.data['attributes']['iotype'] return create_element( 'disk', type=self.TYPE, device='disk', attribute_dict={ 'children': [ create_element('driver', name='qemu', type='raw', cache='none', io=iotype.lower(), discard='unmap'), self.create_source_element(), create_element( 'target', bus='sata' if not virtio else 'virtio', dev=f'{"vd" if virtio else "sd"}{disk_from_number(disk_number)}' ), create_element('boot', order=str(kwargs.pop('boot_number'))), *([] if not logical_sectorsize else [create_element( 'blockio', logical_block_size=str(logical_sectorsize), **({} if not physical_sectorsize else { 'physical_block_size': str(physical_sectorsize) }) )]), ] } ) def create_source_element(self): raise NotImplementedError def _validate(self, device, verrors, old=None, vm_instance=None, update=True): if not self.middleware.call_sync('vm.device.disk_uniqueness_integrity_check', device, vm_instance): verrors.add( 'attributes.path', f'{vm_instance["name"]} has "{self.identity()}" already configured' ) if device['attributes'].get('physical_sectorsize') and not device['attributes'].get('logical_sectorsize'): verrors.add( 'attributes.logical_sectorsize', 'This field must be provided when physical_sectorsize is specified.' ) class RAW(StorageDevice): TYPE = 'file' schema = Dict( 'attributes', Str('path', required=True, validators=[Match( r'^[^{}]*$', explanation='Path should not contain "{", "}" characters' )], empty=False), Str('type', enum=['AHCI', 'VIRTIO'], default='AHCI'), Bool('exists'), Bool('boot', default=False), Int('size', default=None, null=True), Int('logical_sectorsize', enum=[None, 512, 4096], default=None, null=True), Int('physical_sectorsize', enum=[None, 512, 4096], default=None, null=True), Str('iotype', enum=IOTYPE_CHOICES, default='THREADS'), ) def create_source_element(self): return create_element('source', file=self.data['attributes']['path']) def _validate(self, device, verrors, old=None, vm_instance=None, update=True): path = device['attributes']['path'] exists = device['attributes'].get('exists', True) if exists and not os.path.exists(path): verrors.add('attributes.path', 'Path must exist when "exists" is set.') elif not exists: if os.path.exists(path): verrors.add('attributes.path', 'Path must not exist when "exists" is unset.') elif not device['attributes'].get('size'): verrors.add('attributes.size', 'Please provide a valid size for the raw file.') if ( old and old['attributes'].get('size') != device['attributes'].get('size') and not device['attributes'].get('size') ): verrors.add('attributes.size', 'Please provide a valid size for the raw file.') self.middleware.call_sync('vm.device.validate_path_field', verrors, 'attributes.path', path) super()._validate(device, verrors, old, vm_instance, update) class DISK(StorageDevice): TYPE = 'block' schema = Dict( 'attributes', Str('path'), Str('type', enum=['AHCI', 'VIRTIO'], default='AHCI'), Bool('create_zvol'), Str('zvol_name'), Int('zvol_volsize'), Int('logical_sectorsize', enum=[None, 512, 4096], default=None, null=True), Int('physical_sectorsize', enum=[None, 512, 4096], default=None, null=True), Str('iotype', enum=IOTYPE_CHOICES, default='THREADS'), ) def create_source_element(self): return create_element('source', dev=self.data['attributes']['path']) def _validate(self, device, verrors, old=None, vm_instance=None, update=True): create_zvol = device['attributes'].get('create_zvol') path = device['attributes'].get('path') if create_zvol: for attr in ('zvol_name', 'zvol_volsize'): if not device['attributes'].get(attr): verrors.add(f'attributes.{attr}', 'This field is required.') if device['attributes'].get('path'): verrors.add('attributes.path', 'Must not be specified when creating zvol') verrors.check() # Add normalized path for the zvol device['attributes']['path'] = zvol_name_to_path(device['attributes']['zvol_name']) if zvol := self.middleware.call_sync( 'pool.dataset.query', [['id', '=', device['attributes']['zvol_name']]] ): verrors.add('attributes.zvol_name', f'{zvol[0]["id"]!r} already exists.') parentzvol = device['attributes']['zvol_name'].rsplit('/', 1)[0] if parentzvol and not self.middleware.call_sync('pool.dataset.query', [('id', '=', parentzvol)]): verrors.add( 'attributes.zvol_name', f'Parent dataset {parentzvol} does not exist.', errno.ENOENT ) else: for attr in filter(lambda k: device['attributes'].get(k), ('zvol_name', 'zvol_volsize')): verrors.add(f'attributes.{attr}', 'This field should not be specified when "create_zvol" is unset.') if not path: verrors.add('attributes.path', 'Disk path is required.') elif not path.startswith('/dev/zvol/'): verrors.add('attributes.path', 'Disk path must start with "/dev/zvol/"') elif check_zvol_in_boot_pool_using_path(path): verrors.add('attributes.path', 'Disk residing in boot pool cannot be consumed and is not supported') else: zvol = self.middleware.call_sync( 'zfs.dataset.query', [['id', '=', zvol_path_to_name(path)]], {'extra': {'properties': []}} ) if not zvol: verrors.add('attributes.path', 'Zvol referenced by path does not exist', errno.ENOENT) elif zvol[0]['type'] != 'VOLUME': verrors.add('attributes.path', 'Path specified does not reference to a VOLUME') super()._validate(device, verrors, old, vm_instance, update)
7,619
Python
.py
141
42.198582
120
0.588955
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,038
boot_loader.py
truenas_middleware/src/middlewared/middlewared/plugins/boot_/boot_loader.py
import os import shutil import tempfile from middlewared.service import Service, private from middlewared.utils import run class BootService(Service): @private async def install_loader(self, dev): legacy_schema = await self.middleware.call('boot.legacy_schema', dev) if legacy_schema == 'EFI_ONLY': efi_partition_number = 1 else: efi_partition_number = 2 await run('grub-install', '--target=i386-pc', f'/dev/{dev}') if legacy_schema == 'BIOS_ONLY': return partition = await self.middleware.call('disk.get_partition_for_disk', dev, efi_partition_number) await run('mkdosfs', '-F', '32', '-s', '1', '-n', 'EFI', f'/dev/{partition}') with tempfile.TemporaryDirectory() as tmpdirname: efi_dir = os.path.join(tmpdirname, 'efi') os.makedirs(efi_dir) await run('mount', '-t', 'vfat', f'/dev/{partition}', efi_dir) grub_cmd = [ 'grub-install', '--target=x86_64-efi', f'--efi-directory={efi_dir}', '--bootloader-id=debian', '--recheck', '--no-floppy', ] if not os.path.exists('/sys/firmware/efi'): grub_cmd.append('--no-nvram') await run(*grub_cmd) mounted_efi_dir = os.path.join(efi_dir, 'EFI') os.makedirs(os.path.join(mounted_efi_dir, 'boot'), exist_ok=True) shutil.copy( os.path.join(mounted_efi_dir, 'debian/grubx64.efi'), os.path.join(mounted_efi_dir, 'boot/bootx64.efi') ) await run('umount', efi_dir)
1,643
Python
.py
36
35.111111
104
0.574375
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,039
format.py
truenas_middleware/src/middlewared/middlewared/plugins/boot_/format.py
from middlewared.schema import accepts, Dict, Int, Str from middlewared.service import CallError, private, Service from middlewared.utils import run class BootService(Service): @accepts( Str('dev'), Dict( 'options', Int('size'), Str('legacy_schema', enum=[None, 'BIOS_ONLY', 'EFI_ONLY'], null=True, default=None), ) ) @private async def format(self, dev, options): """ Format a given disk `dev` using the appropriate partition layout """ job = await self.middleware.call('disk.wipe', dev, 'QUICK') await job.wait() if job.error: raise CallError(job.error) disk_details = await self.middleware.call('device.get_disk', dev) if not disk_details: raise CallError(f'Details for {dev} not found.') commands = [] partitions = [] if options['legacy_schema'] == 'BIOS_ONLY': partitions.extend([ ('BIOS boot partition', 524288), ]) elif options['legacy_schema'] == 'EFI_ONLY': partitions.extend([ ('EFI System', 272629760), ]) else: partitions.extend([ ('BIOS boot partition', 1048576), # We allot 1MiB to bios boot partition ('EFI System', 536870912) # We allot 512MiB for EFI partition ]) if options.get('size'): partitions.append(('Solaris /usr & Mac ZFS', options['size'])) # 73 sectors are reserved by Linux for GPT tables and # our 4096 bytes alignment offset for the boot disk partitions.append(( 'GPT partition table', 73 * disk_details['sectorsize'] )) total_partition_size = sum(map(lambda y: y[1], partitions)) if disk_details['size'] < total_partition_size: partitions = [ '%s: %s blocks' % (p[0], '{:,}'.format(p[1] // disk_details['sectorsize'])) for p in partitions ] partitions.append( 'total of %s blocks' % '{:,}'.format(total_partition_size // disk_details['sectorsize']) ) disk_blocks = '{:,}'.format(disk_details["blocks"]) raise CallError( f'The new device ({dev}, {disk_details["size"] / (1024 ** 3)} GB, {disk_blocks} blocks) ' f'does not have enough space to to hold the required new partitions ({", ".join(partitions)}). ' 'New mirrored devices might require more space than existing devices due to changes in the ' 'booting procedure.' ) zfs_part_size = f'+{options["size"] // 1024}K' if options.get('size') else 0 if options['legacy_schema']: if options['legacy_schema'] == 'BIOS_ONLY': commands.extend(( ['sgdisk', f'-a{4096 // disk_details["sectorsize"]}', '-n1:0:+512K', '-t1:EF02', f'/dev/{dev}'], )) elif options['legacy_schema'] == 'EFI_ONLY': commands.extend(( ['sgdisk', f'-a{4096 // disk_details["sectorsize"]}', '-n1:0:+260M', '-t1:EF00', f'/dev/{dev}'], )) # Creating standard-size partitions first leads to better alignment and more compact disk usage # and can help to fit larger data partition. commands.extend([ ['sgdisk', f'-n2:0:{zfs_part_size}', '-t2:BF01', f'/dev/{dev}'], ]) else: commands.extend(( ['sgdisk', f'-a{4096 // disk_details["sectorsize"]}', '-n1:0:+1024K', '-t1:EF02', f'/dev/{dev}'], ['sgdisk', '-n2:0:+524288K', '-t2:EF00', f'/dev/{dev}'], )) # Creating standard-size partitions first leads to better alignment and more compact disk usage # and can help to fit larger data partition. commands.extend([ ['sgdisk', f'-n3:0:{zfs_part_size}', '-t3:BF01', f'/dev/{dev}'] ]) for command in commands: p = await run(*command, check=False) if p.returncode != 0: raise CallError( '{} failed:\n{}{}'.format(' '.join(command), p.stdout.decode('utf-8'), p.stderr.decode('utf-8')) ) await self.middleware.call('device.settle_udev_events')
4,419
Python
.py
93
35.344086
116
0.538355
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,040
device_info.py
truenas_middleware/src/middlewared/middlewared/plugins/device_/device_info.py
import re import pyudev import libsgio from middlewared.plugins.disk_.disk_info import get_partition_size_info from middlewared.schema import Dict, returns from middlewared.service import Service, accepts, private from middlewared.utils.disks import DISKS_TO_IGNORE, get_disk_names, get_disk_serial_from_block_device, safe_retrieval from middlewared.utils.functools_ import cache from middlewared.utils.gpu import get_gpus from middlewared.utils.serial import serial_port_choices RE_NVME_PRIV = re.compile(r'nvme[0-9]+c') ISCSI_DEV_PATH = re.compile( r'/devices/platform/host[0-9]+/session[0-9]+/target[0-9]+:[0-9]+:[0-9]+/[0-9]+:[0-9]+:[0-9]+:[0-9]+/block/.*' ) def is_iscsi_device(dev): """Return True if the specified pyudev device is iSCSI based.""" # The implementation may change at a later date return ISCSI_DEV_PATH.match(dev.device_path) is not None class DeviceService(Service): DISK_ROTATION_ERROR_LOG_CACHE = set() @private @cache def host_type(self): return self.middleware.call_sync('system.dmidecode_info')['system-product-name'] @private def get_serials(self): return serial_port_choices() @private def get_disk_serial(self, dev): return get_disk_serial_from_block_device(dev) @private def get_disk_names(self): """ This endpoint serves almost exclusively to be called in our reporting plugin. It just needs the block device names (sda/nvme0n1/pmem0/etc) and so this will very quickly enumerate that information. NOTE: The return of this method should match the keys retrieved when running `self.get_disks`. """ return get_disk_names() @private def get_disks(self, get_partitions=False, serial_only=False): ctx = pyudev.Context() disks = {} for dev in ctx.list_devices(subsystem='block', DEVTYPE='disk'): if dev.sys_name.startswith(DISKS_TO_IGNORE) or RE_NVME_PRIV.match(dev.sys_name): continue if is_iscsi_device(dev): continue try: if serial_only: disks[dev.sys_name] = self.get_disk_serial(dev) else: disks[dev.sys_name] = self.get_disk_details(ctx, dev, get_partitions) except Exception: self.logger.debug('Failed to retrieve disk details for %s', dev.sys_name, exc_info=True) return disks @private def get_disk_partitions(self, dev): parts = [] keys = tuple('ID_PART_ENTRY_' + i for i in ('TYPE', 'UUID', 'NUMBER', 'SIZE')) parent = dev.sys_name for i in filter(lambda x: all(x.get(k) for k in keys), dev.children): part_num = int(i['ID_PART_ENTRY_NUMBER']) part_name = self.middleware.call_sync('disk.get_partition_for_disk', parent, part_num) pinfo = get_partition_size_info(parent, int(i['ID_PART_ENTRY_OFFSET']), int(i['ID_PART_ENTRY_SIZE'])) part = { 'name': part_name, 'id': part_name, 'path': f'/dev/{parent}', 'disk': parent, 'fs_label': i.get('ID_FS_LABEL'), 'partition_type': i['ID_PART_ENTRY_TYPE'], 'partition_number': part_num, 'partition_uuid': i['ID_PART_ENTRY_UUID'], 'start_sector': pinfo.start_sector, 'end_sector': pinfo.end_sector, 'start': pinfo.start_byte, 'end': pinfo.end_byte, 'size': pinfo.total_bytes, 'encrypted_provider': None, } for attr in filter(lambda x: x.startswith('holders/md'), i.attributes.available_attributes): # looks like `holders/md123` part['encrypted_provider'] = f'/dev/{attr.split("/", 1)[1].strip()}' break parts.append(part) return parts @private def get_disk_details(self, ctx, dev, get_partitions=False): blocks = self.safe_retrieval(dev.attributes, 'size', None, asint=True) ident = serial = self.get_disk_serial(dev) model = descr = self.safe_retrieval(dev.properties, 'ID_MODEL', None) vendor = self.safe_retrieval(dev.properties, 'ID_VENDOR', None) is_nvme = dev.sys_name.startswith('nvme') or (vendor and vendor.lower().strip() == 'nvme') driver = self.safe_retrieval(dev.parent.properties, 'DRIVER', '') if not is_nvme else 'nvme' sectorsize = self.safe_retrieval(dev.attributes, 'queue/logical_block_size', None, asint=True) size = mediasize = None if blocks: size = mediasize = blocks * 512 disk = { 'name': dev.sys_name, 'sectorsize': sectorsize, 'number': dev.device_number, 'subsystem': self.safe_retrieval(dev.parent.properties, 'SUBSYSTEM', ''), 'driver': driver, 'hctl': self.safe_retrieval(dev.parent.properties, 'DEVPATH', '').split('/')[-1], 'size': size, 'mediasize': mediasize, 'vendor': vendor, 'ident': ident, 'serial': serial, 'model': model, 'descr': descr, 'lunid': self.safe_retrieval(dev.properties, 'ID_WWN', '').removeprefix('0x').removeprefix('eui.') or None, 'bus': self.safe_retrieval(dev.properties, 'ID_BUS', 'UNKNOWN').upper(), 'type': 'UNKNOWN', 'blocks': blocks, 'serial_lunid': None, 'rotationrate': None, 'stripesize': None, # remove this? (not used) 'parts': [], } if get_partitions: disk['parts'] = self.get_disk_partitions(dev) if self.safe_retrieval(dev.attributes, 'queue/rotational', None) == '1': disk['type'] = 'HDD' disk['rotationrate'] = self._get_rotation_rate(f'/dev/{dev.sys_name}') else: disk['type'] = 'SSD' disk['rotationrate'] = None if disk['serial'] and disk['lunid']: disk['serial_lunid'] = f'{disk["serial"]}_{disk["lunid"]}' disk['dif'] = self.is_dif_formatted(ctx, {'subsystem': disk['subsystem'], 'hctl': disk['hctl']}) return disk @private def is_dif_formatted(self, ctx, info): """ DIF is a feature added to the SCSI Standard. It adds 8 bytes to the end of each sector on disk. It increases the size of the commonly-used 512-byte disk block from 512 to 520 bytes. The extra bytes comprise the Data Integrity Field (DIF). The basic idea is that the HBA will calculate a checksum value for the data block on writes, and store it in the DIF. The storage device will confirm the checksum on receive, and store the data plus checksum. On a read, the checksum will be checked by the storage device and by the receiving HBA. The Data Integrity Extension (DIX) allows this check to move up the stack: the application calculates the checksum and passes it to the HBA, to be appended to the 512 byte data block. This provides a full end-to-end data integrity check. With support from the HBA, this means checksums will be computed/verified by the HBA for every block. This is redundant and a waste of bus bandwidth with ZFS. These disks should be reformatted to use a normal sector size without protection information before a pool can be created. """ dif = False if (info['subsystem'] != 'scsi') or (info['hctl'].count(':') != 3): # only check scsi devices return dif try: dev = pyudev.Devices.from_path(ctx, f'/sys/class/scsi_disk/{info["hctl"]}') except pyudev.DeviceNotFoundAtPathError: return dif except Exception: # logging this is painful because it'll spam so # ignore it for now... return dif else: # 0 == disabled, > 0 == enabled return bool(self.safe_retrieval(dev.attributes, 'protection_type', 0, asint=True)) @private def safe_retrieval(self, prop, key, default, asint=False): return safe_retrieval(prop, key, default, asint) @private def get_disk(self, name, get_partitions=False, serial_only=False): context = pyudev.Context() try: block_device = pyudev.Devices.from_name(context, 'block', name) if serial_only: return {'serial': self.get_disk_serial(block_device)} else: return self.get_disk_details(context, block_device, get_partitions) except pyudev.DeviceNotFoundByNameError: return except Exception: self.logger.debug('Failed to retrieve disk details for %s', name, exc_info=True) def _get_type_and_rotation_rate(self, disk_data, device_path): if disk_data['rota']: if self.HOST_TYPE == 'QEMU': # qemu/kvm guests do not support necessary ioctl for # retrieving rotational rate type_ = 'HDD' rotation_rate = None else: rotation_rate = self._get_rotation_rate(device_path) if rotation_rate: type_ = 'HDD' else: # Treat rotational devices without rotation rate as SSDs # (some USB bridges report SSDs as rotational devices, see # https://jira.ixsystems.com/browse/NAS-112230) type_ = 'SSD' else: type_ = 'SSD' rotation_rate = None return type_, rotation_rate def _get_rotation_rate(self, device_path): try: disk = libsgio.SCSIDevice(device_path) rotation_rate = disk.rotation_rate() except Exception: if device_path not in self.DISK_ROTATION_ERROR_LOG_CACHE: self.DISK_ROTATION_ERROR_LOG_CACHE.add(device_path) self.logger.error('Ioctl failed while retrieving rotational rate for disk %s', device_path) return else: self.DISK_ROTATION_ERROR_LOG_CACHE.discard(device_path) if rotation_rate in (0, 1): # 0 = not reported # 1 = SSD return return rotation_rate @private def get_gpus(self): gpus = get_gpus() to_isolate_gpus = self.middleware.call_sync('system.advanced.config')['isolated_gpu_pci_ids'] for gpu in gpus: gpu['available_to_host'] = gpu['addr']['pci_slot'] not in to_isolate_gpus return gpus
10,761
Python
.py
225
36.991111
119
0.598857
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,041
vrrp_events.py
truenas_middleware/src/middlewared/middlewared/plugins/device_/vrrp_events.py
from asyncio import sleep as asyncio_sleep from dataclasses import dataclass from collections import deque from logging import getLogger from math import floor from os import mkfifo from threading import Thread, Event from time import sleep, time from middlewared.service import Service from middlewared.utils.prctl import set_name LOGGER = getLogger('failover') # so logs show up in /var/log/failover.log @dataclass class VrrpObjs: fifo_thread = None event_thread = None event_queue = deque(maxlen=1) non_crit_ifaces = set() class VrrpThreadService(Service): class Config: cli_private = True private = True def pause_events(self): if VrrpObjs.fifo_thread is not None and VrrpObjs.fifo_thread.is_alive(): VrrpObjs.fifo_thread.pause() if VrrpObjs.event_thread is not None and VrrpObjs.event_thread.is_alive(): VrrpObjs.event_thread.pause() def unpause_events(self): if VrrpObjs.fifo_thread is not None and VrrpObjs.fifo_thread.is_alive(): VrrpObjs.fifo_thread.unpause() if VrrpObjs.event_thread is not None and VrrpObjs.event_thread.is_alive(): VrrpObjs.event_thread.unpause() def set_non_crit_ifaces(self): if VrrpObjs.fifo_thread is not None: VrrpObjs.fifo_thread.non_crit_ifaces = set( i['int_interface'] for i in self.middleware.call_sync( 'datastore.query', 'network.interfaces' ) if not i['int_critical'] ) class VrrpEventThread(Thread): def __init__(self, **kwargs): super(VrrpEventThread, self).__init__() self.middleware = kwargs.get('middleware') self.event_queue = VrrpObjs.event_queue self.shutdown_event = Event() self.pause_event = Event() self.grace_period = 0.5 self.user_provided_timeout = kwargs.get('timeout') or 2 self.max_wait = self.user_provided_timeout + self.grace_period self.settle_time = (self.max_wait / 2) + self.grace_period self.max_rapid_settle_time = 5 self.rapid_event_settle_time = min(2 * self.user_provided_timeout, self.max_rapid_settle_time) def shutdown(self): self.shutdown_event.set() def pause(self): self.pause_event.set() def unpause(self): self.pause_event.clear() @property def user_provided_timeout(self): return self.__upt @user_provided_timeout.setter def user_provided_timeout(self, value): self.__upt = value def run(self): set_name('vrrp_event_thread') LOGGER.info('vrrp event thread started') last_event, backoff = None, False while not self.shutdown_event.is_set(): if self.pause_event.is_set(): # A BACKUP event has to migrate all the VIPs # off of the controller and the only way to # (quickly) do that is to restart the vrrp service. # However, restarting the VRRP service triggers # more BACKUP events for the other interfaces # so we will pause this thread while we become # the backup controller and then unpause after last_event = None self.event_queue.clear() sleep(0.2) try: this_event = self.event_queue[-1] except IndexError: # loop is started but we've received no events sleep(0.2) continue if last_event is None: # first event (in the loop) so sleep `max_wait` # before we act upon it last_event = this_event sleep(self.max_wait) continue # These are the primary scenarios for which we need to handle # 1. receive 1 event within `max_wait` period # 2. receive 2 events with the most recent event being within # the timeframe of `max_wait` # 3. receive 2 events with the most recent event being greater # than the `max_wait` timeframe # 4. receive 2+ events with the most recent event being less # than the `max_wait` timeframe (i.e. rapid events) # The first 3 scenarios listed above are easy enough to handle # because we send those messages as-is to be processed. The # last scenario is the situation for which we need to try and # have a "settle" time. If we continue to receive a rapid # succesion of events, then we'll log a message and ignore the # event since it will wreak havoc on the HA system. time_diff_floor = floor((this_event['time'] - last_event['time'])) max_wait_floor = floor(self.max_wait) if last_event == this_event or time_diff_floor > max_wait_floor: # scenario #1 and scenario #3 listed above last_event = None backoff = False self.event_queue.pop() self.middleware.call_hook_sync('vrrp.fifo', data=this_event) elif time_diff_floor == max_wait_floor: # scenario #2 listed above # NOTE: # The events looke something like this: # RECEIVED: 'INSTANCE "eno1_v4" BACKUP 254\n' at time: 1701967219.244696 # RECEIVED: 'INSTANCE "eno1_v4" MASTER 254\n' at time: 1701967221.2902775 # In the messages above, the time difference is ~2seconds which is the default # timeout for not receiving a MASTER advertisement before VRRP takes over. So # we'll send this event down the pipe. last_event = None backoff = False self.event_queue.pop() self.middleware.call_hook_sync('vrrp.fifo', data=this_event) elif time_diff_floor < max_wait_floor: # scenario #4 listed above # NOTE: # The events could look like this: # RECEIVED: 'INSTANCE "eno1_v4" BACKUP 254\n' at time: 1701967219.244696 # RECEIVED: 'INSTANCE "eno1_v4" MASTER 254\n' at time: 1701967220.2902775 # This happens when both controllers of an HA system start near simultaneously # (i.e. power-outage event most often) OR it could be happening because of an # external networking problem. Either way, the VRRP service will send adverts # but the moment the MASTER controller is determined, it'll send that advert # and (while testing in-house), it is _always_ less than the default advert # timeout (max_wait). We obviously can't ignore that event because doing so # would prevent the HA system from coming up properly (no zpools, no fenced) if not backoff: backoff = True last_event = this_event sleep(self.rapid_event_settle_time) else: last_event = None backoff = False self.event_queue.pop() LOGGER.warning('Detected rapid succession of failover events: (%r)', this_event) else: LOGGER.warning('Unhandled failover event. last_event: %r, this_event: %r', last_event, this_event) last_event = None backoff = False self.event_queue.pop() class VrrpFifoThread(Thread): def __init__(self, *args, **kwargs): super(VrrpFifoThread, self).__init__() self._retry_timeout = 2 # timeout in seconds before retrying to connect to FIFO self._vrrp_file = '/var/run/vrrpd.fifo' self.pause_event = Event() self.middleware = kwargs.get('middleware') self.non_crit_ifaces = kwargs.get('non_crit_ifaces') or VrrpObjs.non_crit_ifaces self.event_queue = VrrpObjs.event_queue self.shutdown_line = '--SHUTDOWN--' def shutdown(self): with open(self._vrrp_file, 'w') as f: f.write(f'{self.shutdown_line}\n') def pause(self): self.pause_event.set() def unpause(self): self.pause_event.clear() def create_fifo(self): try: mkfifo(self._vrrp_file) except FileExistsError: pass except Exception: raise def format_fifo_msg(self, msg): if any(( not isinstance(msg, dict), not msg.get('event'), len(msg['event'].split()) != 4, not msg.get('time'), )): LOGGER.error('Ignoring unexpected VRRP event message: %r', msg) return try: info = msg['event'].split() ifname = info[1].split('_')[0].strip('"') # interface event = info[2] # the state that is being transititoned to except Exception: LOGGER.error('Failed parsing vrrp message', exc_info=True) return else: if event not in ('MASTER', 'BACKUP', 'FAULT'): return if event == 'FAULT': # a FAULT message is sent when iface goes down event = 'BACKUP' return {'ifname': ifname, 'event': event, 'time': msg['time']} def run(self): set_name('vrrp_fifo_thread') try: self.create_fifo() except Exception: LOGGER.error('FATAL: Unable to create VRRP fifo.', exc_info=True) return log_it = True while True: try: with open(self._vrrp_file) as f: LOGGER.info('vrrp fifo connection established') for line in f: if self.pause_event.is_set(): continue event = line.strip() if event == self.shutdown_line: return formatted = self.format_fifo_msg({'event': event, 'time': time()}) if not formatted: continue elif formatted['ifname'] in self.non_crit_ifaces: LOGGER.debug( 'Received an event (%r) for a non-critical interface, ignoring.', formatted ) continue self.event_queue.append(formatted) except Exception: if log_it: LOGGER.warning( 'vrrp fifo connection not established, retrying every %d seconds', self._retry_timeout, exc_info=True ) log_it = False sleep(self._retry_timeout) async def _start_stop_vrrp_threads(middleware): while not await middleware.call('system.ready'): await asyncio_sleep(0.2) licensed = await middleware.call('failover.licensed') if not licensed: # maybe the system is being downgraded to non-HA # (this is rare but still need to handle it) or # system is being restarted/shutdown etc if VrrpObjs.fifo_thread is not None and VrrpObjs.fifo_thread.is_alive(): await middleware.run_in_thread(VrrpObjs.fifo_thread.shutdown) VrrpObjs.fifo_thread = None if VrrpObjs.event_thread is not None and VrrpObjs.event_thread.is_alive(): await middleware.run_in_thread(VrrpObjs.event_thread.shutdown) VrrpObjs.event_thread = None VrrpObjs.event_queue.clear() else: # if this is a system that is being licensed for HA for the # first time (without being rebooted) then we need to make # sure we start these threads timeout = (await middleware.call('failover.config'))['timeout'] nci = set( i['int_interface'] for i in await middleware.call('datastore.query', 'network.interfaces') if not i['int_critical'] ) if VrrpObjs.fifo_thread is None or not VrrpObjs.fifo_thread.is_alive(): VrrpObjs.fifo_thread = VrrpFifoThread(middleware=middleware, non_crit_ifaces=nci) VrrpObjs.fifo_thread.start() if VrrpObjs.event_thread is None or not VrrpObjs.event_thread.is_alive(): VrrpObjs.event_thread = VrrpEventThread(middleware=middleware, timeout=timeout) VrrpObjs.event_thread.start() async def _post_license_update(middleware, *args, **kwargs): await _start_stop_vrrp_threads(middleware) async def setup(middleware): middleware.create_task(_start_stop_vrrp_threads(middleware)) middleware.register_hook('system.post_license_update', _post_license_update)
13,027
Python
.py
272
35.205882
114
0.582153
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,042
udev_events.py
truenas_middleware/src/middlewared/middlewared/plugins/device_/udev_events.py
import pyudev import subprocess import time from middlewared.service import private, Service from middlewared.utils import run from middlewared.utils.threading import start_daemon_thread class DeviceService(Service): @private async def settle_udev_events(self): cp = await run(['udevadm', 'settle'], stdout=subprocess.DEVNULL, check=False) if cp.returncode != 0: self.middleware.logger.error('Failed to settle udev events: %s', cp.stderr.decode()) @private async def trigger_udev_events(self, device): cp = await run(['udevadm', 'trigger', device], stdout=subprocess.DEVNULL, check=False) if cp.returncode != 0: self.middleware.logger.error('Failed to trigger udev events: %s', cp.stderr.decode()) def udev_events(middleware): _256MB = 268435456 # for large quantity disk systems (100's or more) while True: # We always want to keep polling udev, let's log what error we are # seeing and fix them as we come across them try: context = pyudev.Context() monitor = pyudev.Monitor.from_netlink(context) monitor.set_receive_buffer_size(_256MB) monitor.filter_by(subsystem='block') monitor.filter_by(subsystem='dlm') monitor.filter_by(subsystem='net') for device in iter(monitor.poll, None): middleware.call_hook_sync( f'udev.{device.subsystem}', data={**dict(device), 'SYS_NAME': device.sys_name} ) except Exception: middleware.logger.error('Polling udev failed', exc_info=True) time.sleep(10) def setup(middleware): start_daemon_thread(target=udev_events, args=(middleware,))
1,758
Python
.py
38
37.763158
98
0.660432
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,043
enums.py
truenas_middleware/src/middlewared/middlewared/plugins/jbof/enums.py
import enum class Transport(enum.Enum): NVME_ROCE = 'NVMe/ROCE' NVME_TCP = 'NVMe/TCP' def choices(): return [x.value for x in Transport] class AddressMechanism(enum.Enum): STATIC = 'static' STATIC_SET = 'static_set' DHCP = 'dhcp' def choices(): return [x.value for x in AddressMechanism] class ManagementProtocol(enum.Enum): REDFISH = 'redfish' def choices(): return [x.value for x in ManagementProtocol]
474
Python
.py
16
24.3125
52
0.670379
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,044
crud.py
truenas_middleware/src/middlewared/middlewared/plugins/jbof/crud.py
import asyncio import subprocess import time import middlewared.sqlalchemy as sa from middlewared.plugins.jbof.redfish import (InvalidCredentialsError, RedfishClient) from middlewared.schema import (Bool, Dict, Int, IPAddr, List, Password, Patch, Str, accepts, returns) from middlewared.service import (CallError, CRUDService, ValidationErrors, job, private) from middlewared.utils.license import LICENSE_ADDHW_MAPPING from middlewared.validators import Netmask, Range from .functions import (decode_static_ip, get_sys_class_nvme, initiator_ip_from_jbof_static_ip, initiator_static_ip, jbof_static_ip, jbof_static_ip_from_initiator_ip, static_ip_netmask_int, static_ip_netmask_str, static_mtu) class JBOFModel(sa.Model): __tablename__ = 'storage_jbof' id = sa.Column(sa.Integer(), primary_key=True) jbof_description = sa.Column(sa.String(120), nullable=True) # When performing static (code-based) assignment of data-plane IPs, we # want each JBOD to have a deterministic unique index that counts up from # zero (without any gaps, which rules out using the id). This will not be # part of the public API. jbof_index = sa.Column(sa.Integer(), unique=True) jbof_uuid = sa.Column(sa.Text(), nullable=False, unique=True) jbof_mgmt_ip1 = sa.Column(sa.String(45), nullable=False) jbof_mgmt_ip2 = sa.Column(sa.String(45)) jbof_mgmt_username = sa.Column(sa.String(120)) jbof_mgmt_password = sa.Column(sa.EncryptedText()) class JBOFService(CRUDService): class Config: service = 'jbof' datastore = 'storage.jbof' datastore_prefix = "jbof_" cli_private = True role_prefix = 'JBOF' ENTRY = Dict( 'jbof_entry', Int('id', required=True), Str('description'), # Redfish IPAddr('mgmt_ip1', required=True), IPAddr('mgmt_ip2', required=False), Str('mgmt_username', required=True), Password('mgmt_password', required=True), ) # Number of seconds we need to wait for a ES24N to start responding on # a newly configured IP JBOF_CONFIG_DELAY_SECS = 20 @private async def add_index(self, data): """Add a private unique index (0-255) to the entry if not already present.""" if 'index' not in data: index = await self.middleware.call('jbof.next_index') if index is not None: data['index'] = index else: raise CallError('Could not generate an index.') return data @private async def validate(self, data, schema_name, old=None): verrors = ValidationErrors() # Check license license_count = await self.middleware.call("jbof.licensed") if license_count == 0: verrors.add(f"{schema_name}.mgmt_ip1", "This feature is not licensed") else: if old is None: # We're adding a new JBOF - have we exceeded the license? count = await self.middleware.call('jbof.query', [], {'count': True}) if count >= license_count: verrors.add(f"{schema_name}.mgmt_ip1", f"Already configured the number of licensed emclosures: {license_count}") # Ensure redfish connects to mgmt1 (incl login) mgmt_ip1 = data.get('mgmt_ip1') if not RedfishClient.is_redfish(mgmt_ip1): verrors.add(f"{schema_name}.mgmt_ip1", "Not a redfish management interface") else: redfish1 = RedfishClient(f'https://{mgmt_ip1}') try: redfish1.login(data['mgmt_username'], data['mgmt_password']) RedfishClient.cache_set(mgmt_ip1, redfish1) except InvalidCredentialsError: verrors.add(f"{schema_name}.mgmt_username", "Invalid username or password") # If mgmt_ip2 was supplied, ensure it matches to the same system as mgmt_ip1 mgmt_ip2 = data.get('mgmt_ip2') if mgmt_ip2: if not RedfishClient.is_redfish(mgmt_ip2): verrors.add(f"{schema_name}.mgmt_ip2", "Not a redfish management interface") else: redfish2 = RedfishClient(f'https://{mgmt_ip2}') if redfish1.product != redfish2.product: verrors.add(f"{schema_name}.mgmt_ip2", "Product does not match other IP address.") if redfish1.uuid != redfish2.uuid: verrors.add(f"{schema_name}.mgmt_ip2", "UUID does not match other IP address.") # When adding a new JBOF - do we have a UUID clash? if old is None: existing_uuids = [d['uuid'] for d in (await self.middleware.call('jbof.query', [], {'select': ['uuid']}))] if redfish1.uuid in existing_uuids: verrors.add(f"{schema_name}.mgmt_ip1", "Supplied JBOF already in database (UUID)") else: # Inject UUID data['uuid'] = redfish1.uuid await self.add_index(data) return verrors, data @accepts( Patch( 'jbof_entry', 'jbof_create', ('rm', {'name': 'id'}), register=True ) ) async def do_create(self, data): """ Create a new JBOF. This will use the supplied Redfish credentials to configure the data plane on the expansion shelf for direct connection to ROCE capable network cards on the TrueNAS head unit. `description` Optional description of the JBOF. `mgmt_ip1` IP of 1st Redfish management interface. `mgmt_ip2` Optional IP of 2nd Redfish management interface. `mgmt_username` Redfish administrative username. `mgmt_password` Redfish administrative password. """ verrors, data = await self.validate(data, 'jbof_create') verrors.check() mgmt_ip = data['mgmt_ip1'] shelf_index = data['index'] # Everything looks good so far. Attempt to hardwire the dataplane. try: await self.middleware.call('jbof.hardwire_dataplane', mgmt_ip, shelf_index, 'jbof_create.mgmt_ip1', verrors) if verrors: await self.middleware.call('jbof.unwire_dataplane', mgmt_ip, shelf_index) except Exception as e: self.logger.error('Failed to add JBOF', exc_info=True) # Try a cleanup try: await self.middleware.call('jbof.unwire_dataplane', mgmt_ip, shelf_index) except Exception: pass verrors.add('jbof_create.mgmt_ip1', f'Failed to add JBOF: {e}') verrors.check() # If the caller just supplied mgmt_ip1, let's fetch mgmt_ip2 to store in the DB if data.get('mgmt_ip2') in ['', None]: try: if ip := await self.middleware.call('jbof.alt_mgmt_ip', mgmt_ip): data['mgmt_ip2'] = ip self.logger.info('Detected additional JBOF mgmt IP %r', ip) else: self.logger.warning('Unable to determine additional JBOF mgmt IP') except Exception: self.logger.warning('Unable to detect additional JBOF mgmt IP', exc_info=True) data['id'] = await self.middleware.call( 'datastore.insert', self._config.datastore, data, {'prefix': self._config.datastore_prefix}) return await self.get_instance(data['id']) @accepts( Int('id', required=True), Patch( 'jbof_create', 'jbof_update', ('attr', {'update': True}) ) ) async def do_update(self, id_, data): """ Update JBOF of `id` """ old = await self.get_instance(id_) new = old.copy() new.update(data) verrors, data = await self.validate(new, 'jbof_update', old) verrors.check() if old['uuid'] != new['uuid']: self.logger.debug('Changed UUID of JBOF from %s to %s', old['uuid'], new['uuid']) await self.middleware.call('jbof.unwire_dataplane', old['mgmt_ip1'], old['index']) await self.middleware.call('jbof.hardwire_dataplane', new['mgmt_ip1'], new['index'], 'jbof_update.mgmt_ip1', verrors) await self.middleware.call( 'datastore.update', self._config.datastore, id_, new, {'prefix': self._config.datastore_prefix}) return await self.get_instance(id_) @accepts(Int('id'), Bool('force', default=False)) async def do_delete(self, id_, force): """ Delete a JBOF by ID. """ # Will make a best-effort un tear down existing connections / wiring # To do that we first need to fetch the config. data = await self.get_instance(id_) try: await self.middleware.run_in_thread(self.ensure_redfish_client_cached, data) except Exception as e: if force: # If we have lost communication with the redfish interface for any reason # we might still want to proceed with removing the JBOF, even without tearing # down the shelf configuration. However, we wil still want to undo the # host configuration. self.logger.debug('Unable to ensure redfish client for JBOF %r. Forcing.', data['id']) else: raise e try: await self.middleware.call('jbof.unwire_dataplane', data['mgmt_ip1'], data['index']) except Exception: self.logger.debug('Unable to unwire JBOF @%r', data['mgmt_ip1']) await self.middleware.call('alert.oneshot_create', 'JBOFTearDownFailure', None) # Now delete the entry response = await self.middleware.call('datastore.delete', self._config.datastore, id_) return response @accepts() async def reapply_config(self): """ Reapply the JBOF configuration to attached JBOFs. If an IOM is replaced in a JBOF, then it is expected to be configured to have the same redfish IP, user & password as was previously the case. This API can then be called to configure each JBOF with the expected data-plane IP configuration, and then attach NVMe drives. """ verrors = ValidationErrors() await self.middleware.call('jbof.hardwire_shelves') await self.middleware.call('jbof.attach_drives', 'jbof.reapply_config', verrors) verrors.check() @private def get_mgmt_ips(self, mgmt_ip): redfish = RedfishClient.cache_get(mgmt_ip) return redfish.mgmt_ips() @private def alt_mgmt_ip(self, mgmt_ip): other_mgmt_ips = filter(lambda x: x != mgmt_ip, self.get_mgmt_ips(mgmt_ip)) for ip in other_mgmt_ips: if RedfishClient.is_redfish(ip): return ip # If unable to talk to one, pick the first one. Maybe connectivity will be restored later. if len(other_mgmt_ips): self.logger.info('Unable to validate connectivity to alternate JBOF mgmt IP %r', other_mgmt_ips[0]) @private def ensure_redfish_client_cached(self, data): """Synchronous function to ensure we have a redfish client in cache.""" mgmt_ip = data['mgmt_ip1'] username = data.get('mgmt_username') password = data.get('mgmt_password') try: return RedfishClient.cache_get(mgmt_ip) except KeyError: # This could take a while to login, etc ... hence synchronous wrapper. redfish = RedfishClient(f'https://{mgmt_ip}', username, password) RedfishClient.cache_set(mgmt_ip, redfish) return redfish @accepts(roles=['JBOF_READ']) @returns(Int()) async def licensed(self): """Return a count of the number of JBOF units licensed.""" result = 0 # Do we have a license at all? license_ = await self.middleware.call('system.license') if not license_: return result # check if this node's system serial matches the serial in the license local_serial = (await self.middleware.call('system.dmidecode_info'))['system-serial-number'] if local_serial not in (license_['system_serial'], license_['system_serial_ha']): return result # Check to see if we're licensed to attach a JBOF if license_['addhw']: for quantity, code in license_['addhw']: if code not in LICENSE_ADDHW_MAPPING: self.logger.warning('Unknown additional hardware code %d', code) continue name = LICENSE_ADDHW_MAPPING[code] if name == 'ES24N': result += quantity return result @private @accepts( Int('id', required=True), Str('iom', enum=['IOM1', 'IOM2'], required=True), Dict( 'iom_network', Bool('dhcp'), Str('fqdn'), Str('hostname'), List('ipv4_static_addresses', items=[Dict( 'ipv4_static_address', IPAddr('address', v6=False), Str('netmask', validators=[Netmask(ipv6=False, prefix_length=False)]), IPAddr('gateway', v6=False))], default=None), List('ipv6_static_addresses', items=[Dict( 'ipv6_static_address', IPAddr('address', v4=False), Int('prefixlen', validators=[Range(min_=1, max_=64)]))], default=None), List('nameservers', items=[IPAddr('nameserver')], default=None), ), Int('ethindex', default=1), Bool('force', default=False), Bool('check', default=True), ) def set_mgmt_ip(self, id_, iom, data, ethindex, force, check): """Change the mamagement IP for a particular IOM""" # Fetch the existing JBOF config config = self.get_instance__sync(id_) config_mgmt_ips = set([config['mgmt_ip1'], config['mgmt_ip2']]) redfish = self.ensure_redfish_client_cached(config) old_iom_mgmt_ips = set(redfish.iom_mgmt_ips(iom)) if not check: if data.get('dhcp'): raise CallError('Can not bypass check when setting DHCP') try: new_static_ip = data.get('ipv4_static_addresses', [])[0]['address'] except Exception: raise CallError('Can not determine new static IP') if config['mgmt_ip1'] in old_iom_mgmt_ips: ip_to_update = 'mgmt_ip1' elif config['mgmt_ip2'] in old_iom_mgmt_ips: ip_to_update = 'mgmt_ip2' else: raise CallError('Can not determine whether updating mgmt_ip1 or mgmt_ip2') if not force: # Do we need to switch redfish to the other IOM if redfish.mgmt_ip() in old_iom_mgmt_ips: other_iom = 'IOM2' if iom == 'IOM1' else 'IOM1' for mgmt_ip in redfish.iom_mgmt_ips(other_iom): if mgmt_ip in config_mgmt_ips: redfish = self.ensure_redfish_client_cached({'mgmt_ip1': mgmt_ip, 'mgmt_username': config['mgmt_username'], 'mgmt_password': config['mgmt_password']}) break if redfish.mgmt_ip() in redfish.iom_mgmt_ips(iom): raise CallError('Can not modify IOM network config thru same IOM') # Read the existing config via redfish uri = f'/redfish/v1/Managers/{iom}/EthernetInterfaces/{ethindex}' r = redfish.get(uri) if not r.ok: raise CallError('Unable to read existing network configuration of {iom}/{ethindex}') orig_net_config = r.json() newdata = {} olddata = {} if (dhcp := data.get('dhcp')) is not None: newdata.update({'DHCPv4': {'DHCPEnabled': dhcp}}) olddata.update({'DHCPv4': orig_net_config['DHCPv4']}) if (fqdn := data.get('fqdn')) is not None: newdata.update({'FQDN': fqdn}) olddata.update({'FQDN': orig_net_config['FQDN']}) if (hostname := data.get('hostname')) is not None: newdata.update({'HostName': hostname}) olddata.update({'HostName': orig_net_config['HostName']}) if (ipv4_static_addresses := data.get('ipv4_static_addresses')) is not None: newitems = [] for item in ipv4_static_addresses: newitems.append({'Address': item['address'], 'Gateway': item['gateway'], 'SubnetMask': item['netmask']}) newdata.update({'IPv4StaticAddresses': newitems}) olddata.update({'IPv4StaticAddresses': orig_net_config['IPv4StaticAddresses']}) if (ipv6_static_addresses := data.get('ipv6_static_addresses')) is not None: newitems = [] for item in ipv6_static_addresses: newitems.append({'Address': item['address'], 'PrefixLength': item['prefixlen']}) newdata.update({'IPv6StaticAddresses': newitems}) olddata.update({'IPv6StaticAddresses': orig_net_config['IPv6StaticAddresses']}) if (nameservers := data.get('nameservers')) is not None: newdata.update({'NameServers': nameservers}) olddata.update({'NameServers': orig_net_config['NameServers']}) try: removed_active = False added_active = False redfish.post(uri, data=newdata) # Give a few seconds for the changes to take effect time.sleep(10) if check: new_iom_mgmt_ips = set(redfish.iom_mgmt_ips(iom)) if old_iom_mgmt_ips != new_iom_mgmt_ips: # IPs have changed. # 1. Was the IP that changed on one of the stored mgmt_ips for removed_ip in old_iom_mgmt_ips - new_iom_mgmt_ips: if removed_ip in config_mgmt_ips: removed_active = True break if removed_active: for added_ip in new_iom_mgmt_ips - old_iom_mgmt_ips: if RedfishClient.is_redfish(added_ip): added_active = True break if not added_active: raise CallError(f'Unable to access redfish IP on {iom}') # Update the config to reflect the new IP if removed_ip == config['mgmt_ip1']: self.middleware.call_sync( 'jbof.update', config['id'], {'mgmt_ip1': added_ip} ) else: self.middleware.call_sync( 'jbof.update', config['id'], {'mgmt_ip2': added_ip} ) else: # IPs did not change, still want to test connectivity for ip in config_mgmt_ips: if ip in old_iom_mgmt_ips: if not RedfishClient.is_redfish(ip): raise CallError(f'Unable to access redfish IP {ip}') else: # check is False ... don't attempt to communicate with the new IP # just update the database. new = config.copy() new.update({ip_to_update: new_static_ip}) self.middleware.call_sync( 'datastore.update', self._config.datastore, config['id'], new, {'prefix': self._config.datastore_prefix} ) except Exception as e: self.logger.error(f'Unable to modify mgmt ip for {iom}/{ethindex}', exc_info=True) try: redfish.post(uri, data=olddata) except Exception: self.logger.error(f'Unable to restore original mgmt ip for {iom}/{ethindex}', exc_info=True) raise e @private async def next_index(self): existing_indices = [d['index'] for d in (await self.middleware.call('jbof.query', [], {'select': ['index']}))] for index in range(256): if index not in existing_indices: return index @private async def hardwire_dataplane(self, mgmt_ip, shelf_index, schema, verrors): """Hardware the dataplane interfaces of the specified JBOF. Configure the data plane network interfaces on the JBOF to previously determined subnets. Then attempt to connect using all the available RDMA capable interfaces. """ await self.middleware.call('jbof.hardwire_shelf', mgmt_ip, shelf_index) await self.middleware.call('jbof.hardwire_host', mgmt_ip, shelf_index, schema, verrors) if not verrors: await self.middleware.call('jbof.attach_drives', schema, verrors) @private def fabric_interface_choices(self, mgmt_ip): redfish = RedfishClient.cache_get(mgmt_ip) return redfish.fabric_ethernet_interfaces() @private def fabric_interface_macs(self, mgmt_ip): """Return a dict keyed by IP address where the value is the corresponding MAC address.""" redfish = RedfishClient.cache_get(mgmt_ip) macs = {} for uri in self.fabric_interface_choices(mgmt_ip): netdata = redfish.get_uri(uri) for address in netdata['IPv4Addresses']: macs[address['Address']] = netdata['MACAddress'] return macs @private def hardwire_shelf(self, mgmt_ip, shelf_index): redfish = RedfishClient.cache_get(mgmt_ip) shelf_interfaces = redfish.fabric_ethernet_interfaces() # Let's record the link status for each interface up_before = set() for uri in shelf_interfaces: status = redfish.link_status(uri) if status == 'LinkUp': up_before.add(uri) # Modify all the interfaces for (eth_index, uri) in enumerate(shelf_interfaces): address = jbof_static_ip(shelf_index, eth_index) redfish.configure_fabric_interface(uri, address, static_ip_netmask_str(address), mtusize=static_mtu()) # Wait for all previously up interfaces to come up again up_after = set() retries = 0 while retries < JBOFService.JBOF_CONFIG_DELAY_SECS and up_before - up_after: for uri in up_before: if uri not in up_after: status = redfish.link_status(uri) if status == 'LinkUp': up_after.add(uri) time.sleep(1) retries += 1 if up_before - up_after: self.logger.debug('Timed-out waiting for interfaces to come up') # Allow this to continue as we still might manage to ping it. else: self.logger.debug('Configured JBOF #%r', shelf_index) @private async def hardwire_shelves(self): """Apply the expected datapath IPs to all configured shelves.""" jbofs = await self.middleware.call('jbof.query') if jbofs: exceptions = await asyncio.gather( *[self.middleware.call('jbof.hardwire_shelf', jbof['mgmt_ip1'], jbof['index']) for jbof in jbofs], return_exceptions=True ) failures = [] for jbof, exc in zip(jbofs, exceptions): if isinstance(exc, Exception): failures.append(str(exc)) else: self.logger.info('Successfully hardwired JBOF %r (index %r)', jbof['description'], jbof['index']) if failures: self.logger.error(f'Failure hardwiring JBOFs: {", ".join(failures)}') @private def unwire_shelf(self, mgmt_ip): redfish = RedfishClient.cache_get(mgmt_ip) for uri in redfish.fabric_ethernet_interfaces(): redfish.configure_fabric_interface(uri, '0.0.0.0', '255.255.255.0', True, mtusize=1500) @private async def hardwire_host(self, mgmt_ip, shelf_index, schema, verrors): """Discover which direct links exist to the specified expansion shelf.""" # See how many interfaces are available on the expansion shelf shelf_ip_to_mac = await self.middleware.call('jbof.fabric_interface_macs', mgmt_ip) # Setup a dict with the expected IP pairs shelf_ip_to_host_ip = {} for idx, _ in enumerate((await self.middleware.call('jbof.fabric_interface_choices', mgmt_ip))): shelf_ip_to_host_ip[jbof_static_ip(shelf_index, idx)] = initiator_static_ip(shelf_index, idx) # Let's check that we have the expected hardwired IPs on the shelf if set(shelf_ip_to_mac) != set(shelf_ip_to_host_ip): # This should not happen verrors.add(schema, 'JBOF does not have expected IPs.' f'Expected: {shelf_ip_to_host_ip}, has: {shelf_ip_to_mac}') return if await self.middleware.call('failover.licensed'): # HA system if not await self.middleware.call('failover.remote_connected'): verrors.add(schema, 'Unable to contact remote controller') return this_node = await self.middleware.call('failover.node') if this_node == 'MANUAL': verrors.add(schema, 'Unable to determine this controllers position in chassis') return connected_shelf_ips = [] results = await asyncio.gather( *[self.hardwire_node(node, shelf_index, shelf_ip_to_mac) for node in ('A', 'B')] ) for (node, connected_shelf_ips) in zip(('A', 'B'), results): if not connected_shelf_ips: # Failed to connect any IPs => error verrors.add(schema, f'Unable to communicate with the expansion shelf (node {node})') return elif len(connected_shelf_ips) > 1: # Too many connections exist (currently do not support multipath) verrors.add(schema, f'Too many connections wired to the expansion shelf (node {node})') return self.logger.debug('Configured node %r: %r', node, connected_shelf_ips) else: connected_shelf_ips = await self.hardwire_node('', shelf_index, shelf_ip_to_mac) if not connected_shelf_ips: # Failed to connect any IPs => error verrors.add(schema, 'Unable to communicate with the expansion shelf') return elif len(connected_shelf_ips) > 1: # Too many connections exist (currently do not support multipath) verrors.add(schema, 'Too many connections wired to the expansion shelf') return self.logger.debug('Configured node: %r', connected_shelf_ips) @private async def hardwire_node(self, node, shelf_index, shelf_ip_to_mac, skip_ips=[]): localnode = not node or node == await self.middleware.call('failover.node') # Next see what RDMA-capable links are available on the host # Also setup a map for frequent use below if localnode: links = await self.middleware.call('rdma.get_link_choices') else: try: links = await self.middleware.call('failover.call_remote', 'rdma.get_link_choices') except CallError as e: if e.errno != CallError.ENOMETHOD: raise self.logger.warning('Cannot hardwire remote node') return [] # First check to see if any interfaces that were previously configured # for this shelf are no longer applicable (they might have been moved to # a different port on the JBOF). connected_shelf_ips = set() dirty = False configured_interfaces = await self.middleware.call('rdma.interface.query') configured_interface_names = [interface['ifname'] for interface in configured_interfaces if interface['node'] == node] for interface in configured_interfaces: if node and node != interface['node']: continue host_ip = interface['address'] shelf_ip = jbof_static_ip_from_initiator_ip(host_ip) value = decode_static_ip(host_ip) if value and value[0] == shelf_index: # This is supposed to be connected to our shelf. Check connectivity. if await self.middleware.call('rdma.interface.ping', node, interface['ifname'], shelf_ip, shelf_ip_to_mac[shelf_ip]): # This config looks good, keep it. connected_shelf_ips.add(shelf_ip) if node: self.logger.info(f'Validated existing link on node {node}: {host_ip} -> {shelf_ip}') else: self.logger.info(f'Validated existing link: {host_ip} -> {shelf_ip}') else: self.logger.info('Removing RDMA interface that cannot connect to JBOF') await self.middleware.call('rdma.interface.delete', interface['id']) dirty = True for shelf_ip in shelf_ip_to_mac: if shelf_ip in connected_shelf_ips or shelf_ip in skip_ips: continue # Try each remaining interface if dirty: configured_interfaces = await self.middleware.call('rdma.interface.query') configured_interface_names = [interface['ifname'] for interface in configured_interfaces if interface['node'] == node] dirty = False for link in links: ifname = link['rdma'] if ifname not in configured_interface_names: host_ip = initiator_ip_from_jbof_static_ip(shelf_ip) payload = { 'ifname': ifname, 'address': host_ip, 'prefixlen': static_ip_netmask_int(), 'mtu': static_mtu(), 'check': {'ping_ip': shelf_ip, 'ping_mac': shelf_ip_to_mac[shelf_ip]} } if node: payload['node'] = node if await self.middleware.call('rdma.interface.create', payload): dirty = True connected_shelf_ips.add(shelf_ip) # break out of the ifname loop if node: self.logger.info(f'Created link on node {node}: {host_ip} -> {shelf_ip}') else: self.logger.info(f'Created link: {host_ip} -> {shelf_ip}') break return list(connected_shelf_ips) @private async def attach_drives(self, schema, verrors): """Attach drives from all configured JBOF expansion shelves.""" if await self.middleware.call('failover.licensed'): # HA system if not await self.middleware.call('failover.remote_connected'): verrors.add(schema, 'Unable to contact remote controller') return this_node = await self.middleware.call('failover.node') if this_node == 'MANUAL': verrors.add(schema, 'Unable to determine this controllers position in chassis') return await asyncio.gather(*[self.attach_drives_to_node(node) for node in ('A', 'B')]) else: await self.attach_drives_to_node('') @private async def attach_drives_to_node(self, node): localnode = not node or node == await self.middleware.call('failover.node') configured_interfaces = await self.middleware.call('rdma.interface.query') if localnode: for interface in configured_interfaces: if interface['node'] != node: continue jbof_ip = jbof_static_ip_from_initiator_ip(interface['address']) await self.middleware.call('jbof.nvme_connect', jbof_ip) else: for interface in configured_interfaces: if interface['node'] != node: continue jbof_ip = jbof_static_ip_from_initiator_ip(interface['address']) try: await self.middleware.call('failover.call_remote', 'jbof.nvme_connect', [jbof_ip]) except CallError as e: if e.errno != CallError.ENOMETHOD: raise @private def nvme_connect(self, ip, nr_io_queues=16): command = ['nvme', 'connect-all', '-t', 'rdma', '-a', ip, '--persistent', '-i', f'{nr_io_queues}'] ret = subprocess.run(command, capture_output=True) if ret.returncode: error = ret.stderr.decode() if ret.stderr else ret.stdout.decode() if not error: error = 'No error message reported' self.logger.debug('Failed to execute command: %r with error: %r', " ".join(command), error) raise CallError(f'Failed connect NVMe disks: {error}') return True @private def nvme_disconnect(self, ips): """Iterate through all nvme devices that have a transport protocol of RDMA and disconnect from this host""" nqns = [] for nvme, info in get_sys_class_nvme().items(): if info['transport_protocol'] == 'rdma' and any(ip == info['transport_address'] for ip in ips): nqns.append(info['subsysnqn']) len_nqns = len(nqns) if len_nqns > 0: self.logger.debug('Disconnecting %r NQNs', len_nqns) command = ['nvme', 'disconnect', '-n', ','.join(nqns)] ret = subprocess.run(command, capture_output=True) if ret.returncode: error = ret.stderr.decode() if ret.stderr else ret.stdout.decode() if not error: error = 'No error message reported' raise CallError(f'Failed disconnect NVMe disks: {error}') @private async def shelf_interface_count(self, mgmt_ip): try: return len(await self.middleware.call('jbof.fabric_interface_choices', mgmt_ip)) except Exception: # Really only expect 4, but we'll over-estimate for now, as we check them anyway return 6 @private async def unwire_host(self, mgmt_ip, shelf_index): """Unware the dataplane interfaces of the specified JBOF.""" possible_host_ips = [] possible_shelf_ips = [] shelf_interface_count = await self.shelf_interface_count(mgmt_ip) # If shelf_interface_count is e.g. 4 then we want to iterate over [0,1,2,3] for eth_index in range(shelf_interface_count): possible_host_ips.append(initiator_static_ip(shelf_index, eth_index)) possible_shelf_ips.append(jbof_static_ip(shelf_index, eth_index)) # Disconnect NVMe disks if await self.middleware.call('failover.licensed'): # HA system try: await asyncio.gather( self.middleware.call('jbof.nvme_disconnect', possible_shelf_ips), self.middleware.call('failover.call_remote', 'jbof.nvme_disconnect', [possible_shelf_ips]) ) except CallError as e: if e.errno != CallError.ENOMETHOD: raise # If other controller is not updated to include this method then nothing to tear down else: await self.middleware.call('jbof.nvme_disconnect', possible_shelf_ips) # Disconnect interfaces for interface in await self.middleware.call('rdma.interface.query', [['address', 'in', possible_host_ips]]): await self.middleware.call('rdma.interface.delete', interface['id']) @private async def unwire_dataplane(self, mgmt_ip, shelf_index): """Unware the dataplane interfaces of the specified JBOF.""" await self.middleware.call('jbof.unwire_host', mgmt_ip, shelf_index) await self.middleware.call('jbof.unwire_shelf', mgmt_ip) @private async def configure(self): interfaces = await self.middleware.call('rdma.interface.configure') for interface in interfaces: jbof_ip = jbof_static_ip_from_initiator_ip(interface['address']) await self.middleware.call('jbof.nvme_connect', jbof_ip) @private async def configure_jbof(self, node, shelf_index): """Bring up a particular previously-configured JBOF on this node.""" possible_host_ips = [] jbof_ips = [] # Seeing as we're just going to be using possible_host_ips to filter a DB query, # don't bother checking with the JBOF for its shelf_interface_count ... just # assume a ridiculously high number (12) instead, so we interate over [0,1,11] shelf_interface_count = 12 for eth_index in range(shelf_interface_count): possible_host_ips.append(initiator_static_ip(shelf_index, eth_index)) # First bring up the interfaces on this host interfaces = await self.middleware.call('rdma.interface.query', [['address', 'in', possible_host_ips], ['node', '=', node]]) for interface in interfaces: await self.middleware.call('rdma.interface.local_configure_interface', interface['ifname'], interface['address'], interface['prefixlen'], interface['mtu']) jbof_ips.append(jbof_static_ip_from_initiator_ip(interface['address'])) # Next do the NVMe connect # Include some retry code, but expect it won't get used. retries = 5 while retries: retries -= 1 # Note that we iterate over a COPY of jbof_ips so that we can remove items for jbof_ip in jbof_ips[:]: try: await self.middleware.call('jbof.nvme_connect', jbof_ip) jbof_ips.remove(jbof_ip) self.logger.debug(f'Connected NVMe/RoCE: {jbof_ip}') except CallError: if retries: self.logger.info(f'Failed to connect to {jbof_ip}, will retry') await asyncio.sleep(1) else: raise if not jbof_ips: return @private @job(lock='configure_job') async def configure_job(self, job, reload_fenced=False): """Bring up any previously configured JBOF NVMe/RoCE configuration. Each JBOF will be brought up in parallel. Result will be a dict with keys 'failed' (boolean) and 'message' (str). """ job.set_progress(0, 'Configure RDMA interfaces') failed = False if await self.middleware.call('failover.licensed'): node = await self.middleware.call('failover.node') else: node = '' jbofs = await self.middleware.call('jbof.query') if not jbofs: err = 'No JBOFs need to be configured' job.set_progress(100, err) return {'failed': failed, 'message': err} # Bring up the JBOFs in parallel. exceptions = await asyncio.gather( *[self.configure_jbof(node, jbof['index']) for jbof in jbofs], return_exceptions=True ) failures = [] for exc in exceptions: if isinstance(exc, Exception): failures.append(str(exc)) # Report progress so far. if reload_fenced: percent_available = 90 else: percent_available = 100 if failures: # We know all_count is > 0 because of the return above. all_count = len(jbofs) fail_count = len(failures) percent = (percent_available * (all_count - fail_count)) // all_count err = f'Failure connecting {fail_count} JBOFs: {", ".join(failures)}' self.logger.error(err) job.set_progress(percent, err) failed = True else: percent = percent_available err = 'Completed boot-time bring up of NVMe/RoCE' job.set_progress(percent, err) # Reload fenced if requested if reload_fenced and (await self.middleware.call('failover.fenced.run_info'))['running']: try: await self.middleware.call('failover.fenced.signal', {'reload': True}) self.logger.debug('Reloaded fenced') job.set_progress(percent + 10, err + ', reloaded fenced') except Exception: self.logger.error('Unhandled exception reloading fenced', exc_info=True) job.set_progress(percent, err + ', failed to reload fenced') failed = True else: job.set_progress(percent + 10, err) # This gets returned as the job.result return {'failed': failed, 'message': err} async def _clear_reboot_alerts(middleware, event_type, args): await middleware.call('alert.oneshot_delete', 'JBOFTearDownFailure', None) async def setup(middleware): RedfishClient.setup() # Deliberately do NOT handle the case where the system is already # ready, as we only want the following to occur after a boot, not # on a middlwared restart. middleware.event_subscribe("system.ready", _clear_reboot_alerts)
42,614
Python
.py
838
37.597852
120
0.577597
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,045
functions.py
truenas_middleware/src/middlewared/middlewared/plugins/jbof/functions.py
import ipaddress from pathlib import Path from middlewared.utils.disks import RE_IS_PART def jbof_static_ip(shelf_index, eth_index): return f'169.254.{20 + shelf_index}.{(eth_index <<2)+1}' def initiator_static_ip(shelf_index, eth_index): return f'169.254.{20 + shelf_index}.{(eth_index <<2)+2}' def static_ip_netmask_int(): return 30 def static_ip_netmask_str(ip='169.254.20.0'): return str(ipaddress.IPv4Network(f'{ip}/{static_ip_netmask_int()}', strict=False).netmask) def static_mtu(): return 5000 def decode_static_ip(ip): """Decode a static IP. Returns a tuple of (shelf_index, eth_index) or None """ try: ipaddress.ip_address(ip) if ip.startswith('169.254.'): vals = [int(x) for x in ip.split('.')] if vals[2] < 20: return shelf_index = vals[2] - 20 eth_index = vals[3] >> 2 return (shelf_index, eth_index) except ValueError: pass def jbof_static_ip_from_initiator_ip(ip): result = decode_static_ip(ip) if result: return jbof_static_ip(*result) def initiator_ip_from_jbof_static_ip(ip): result = decode_static_ip(ip) if result: return initiator_static_ip(*result) def get_sys_class_nvme(): data = dict() for i in filter(lambda x: x.is_dir(), Path('/sys/class/nvme').iterdir()): data[i.name] = { 'model': (i / 'model').read_text().strip(), 'serial': (i / 'serial').read_text().strip(), 'subsysnqn': (i / 'subsysnqn').read_text().strip(), 'transport_address': (i / 'address').read_text().strip(), 'transport_protocol': (i / 'transport').read_text().strip(), 'state': (i / 'state').read_text().strip(), } if data[i.name]['transport_protocol'] == 'rdma': data[i.name]['hostnqn'] = (i / 'hostnqn').read_text().strip() data[i.name]['transport_address'] = data[i.name]['transport_address'].split('=')[1].split(',')[0].strip() namespaces, partitions = list(), list() for j in filter(lambda x: x.is_dir() and x.name.startswith(f'{i.name}n'), i.iterdir()): # nvme1n1/n2/n3 etc namespaces.append(j.name) for k in filter(lambda x: RE_IS_PART.search(x.name), j.iterdir()): # nvme1n1p1/p2/p3 etc partitions.append(k.name) data[i.name]['namespaces'] = namespaces data[i.name]['partitions'] = partitions return data
2,534
Python
.py
60
34.116667
117
0.589555
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,046
client.py
truenas_middleware/src/middlewared/middlewared/plugins/jbof/redfish/client.py
import asyncio import enum import errno import json import logging import socket from urllib.parse import urlencode import aiohttp import requests from middlewared.service import CallError from middlewared.utils import MIDDLEWARE_RUN_DIR, filter_list from truenas_api_client import Client from urllib3.exceptions import InsecureRequestWarning DEFAULT_REDFISH_TIMEOUT_SECS = 10 HEADER = {'Content-Type': 'application/json', 'Vary': 'accept'} REDFISH_ROOT_PATH = '/redfish/v1' ODATA_ID = '@odata.id' LOGGER = logging.getLogger(__name__) REDFISH_SESSIONS = '/redfish/v1/SessionService/Sessions' class InvalidCredentialsError(Exception): pass class AuthMethod(enum.Enum): BASIC = 'basic' SESSION = 'session' def choices(): return [x.value for x in AuthMethod] def authtype_to_enum(authtype): if authtype in (AuthMethod.BASIC, AuthMethod.BASIC.value): return AuthMethod.BASIC elif authtype in (AuthMethod.SESSION, AuthMethod.SESSION.value): return AuthMethod.SESSION raise ValueError('Invalid auth method', authtype) class AbstractRedfishClient: @property def uuid(self): return self.root['UUID'] @property def product(self): return self.root['Product'] def _members(self, data): result = {} for manager in data['Members']: uri = manager[ODATA_ID] result[uri.split('/')[-1]] = uri return result class RedfishClient(AbstractRedfishClient): client_cache = {} def __init__( self, base_url, username=None, password=None, authtype=AuthMethod.BASIC, default_prefix=REDFISH_ROOT_PATH, verify=False, timeout=DEFAULT_REDFISH_TIMEOUT_SECS ): self.log_requests = False self.base_url = base_url.rstrip('/') self.username = username self.password = password self.authtype = AuthMethod.authtype_to_enum(authtype) self.prefix = default_prefix self.verify = verify self.auth = None self.auth_token = None self.session_key = None self.authorization_key = None self.session_location = None self.timeout = timeout self.cache = {} self.root = self.get_root_object() try: self.login_url = self.root['Links']['Sessions']['@odata.id'] except KeyError: self.login_url = REDFISH_SESSIONS if username and password: self.login() @classmethod def setup(cls): # Silence InsecureRequestWarning: Unverified HTTPS request is being made to host requests.packages.urllib3.disable_warnings(InsecureRequestWarning) @classmethod def ping(cls, mgmt_ip, timeout=DEFAULT_REDFISH_TIMEOUT_SECS): r = requests.get(f'https://{mgmt_ip.rstrip("/")}{REDFISH_ROOT_PATH}', verify=False, timeout=timeout) if r.ok: try: return r.json() except requests.exceptions.JSONDecodeError: pass @classmethod def is_redfish(cls, mgmt_ip, timeout=DEFAULT_REDFISH_TIMEOUT_SECS): try: data = cls.ping(mgmt_ip, timeout) return data and 'RedfishVersion' in data except requests.exceptions.Timeout: LOGGER.debug('Timed out querying redfish host %r', mgmt_ip) return False @classmethod def cache_set(cls, key, value): cls.client_cache[key] = value @classmethod def cache_get(cls, mgmt_ip, jbof_query=None): try: return cls.client_cache[mgmt_ip] except KeyError: redfish, jbofs = None, list() filters, options = [['OR', [['mgmt_ip1', '=', mgmt_ip], ['mgmt_ip2', '=', mgmt_ip]]]], dict() if jbof_query is not None: jbofs = jbof_query else: with Client(f'ws+unix://{MIDDLEWARE_RUN_DIR}/middlewared-internal.sock', py_exceptions=True) as c: jbofs = c.call('jbof.query') for jbof in filter_list(jbofs, filters, options): redfish = RedfishClient(f'https://{mgmt_ip}', jbof['mgmt_username'], jbof['mgmt_password']) RedfishClient.cache_set(mgmt_ip, redfish) return redfish def _cached_fetch(self, cache_key, uri, use_cached=True): if use_cached and cache_key in self.cache: return self.cache[cache_key] r = self.get(uri) if r.ok: self.cache[cache_key] = self._members(r.json()) return self.cache[cache_key] def chassis(self, use_cached=True): return self._cached_fetch('chassis', '/Chassis', use_cached) def managers(self, use_cached=True): return self._cached_fetch('managers', '/Managers', use_cached) def mgmt_ethernet_interfaces(self, iom, use_cached=True): uri = f'{self.managers()[iom]}/EthernetInterfaces' return self._cached_fetch(f'{iom}/mgmt_ethernet_interfaces', uri, use_cached) def mgmt_ip(self): return socket.gethostbyname(self.base_url.split('/')[-1]) def iom_eth_mgmt_ips(self, eth_uri): result = [] # Do not want any cached value. IPs can change. data = self.get_uri(eth_uri, False) for ipv4_address in data.get('IPv4Addresses', []): addr = ipv4_address.get('Address') if addr: result.append(addr) return result def iom_mgmt_ips(self, iom): result = [] for eth_uri in self.mgmt_ethernet_interfaces(iom).values(): # Do not want any cached value. IPs can change. result.extend(self.iom_eth_mgmt_ips(eth_uri)) return result def mgmt_ips(self): result = [] for iom in self.managers(): result.extend(self.iom_mgmt_ips(iom)) return result def network_device_functions(self, iom, use_cached=True): return self._cached_fetch( f'{iom}/network_device_functions', f'/Chassis/{iom}/NetworkAdapters/1/NetworkDeviceFunctions', use_cached ) def fabric_ethernet_interfaces(self, use_cached=True): result = [] for iom in self.managers(use_cached): for ndfuri in self.network_device_functions(iom, use_cached).values(): result.append(f'{ndfuri}/EthernetInterfaces/1') result.sort() return result def get_uri(self, uri, use_cached=True): if use_cached and uri in self.cache: return self.cache[uri] r = self.get(uri) if r.ok: self.cache[uri] = r.json() return self.cache[uri] def get_root_object(self): return self.get(self.prefix).json() def login(self, username=None, password=None, authtype=None): self.username = username if username else self.username self.password = password if password else self.password if authtype: self.authtype = AuthMethod.authtype_to_enum(authtype) if self.authtype == AuthMethod.BASIC: self.get(self.login_url, auth=(self.username, self.password)) # No exception thrown ... self.auth = (self.username, self.password) elif self.authtype == AuthMethod.SESSION: data = {'UserName': self.username, 'Password': self.password} resp = self.post(self.login_url, data=data) self.resp = resp if not resp.ok: raise InvalidCredentialsError('Could not authenticate credentials supplied') self.auth_token = resp.headers.get('X-Auth-Token') if self.auth_token: self.session_id = resp.json()['Id'] self.session_location = resp.headers.get('Location') else: raise ValueError('Invalid auth supplied:', authtype) def logout(self): if self.authtype == AuthMethod.BASIC: self.auth = None elif self.authtype == AuthMethod.SESSION: self.delete(self.session_location) self.auth_token = self.session_id = self.session_location = None self.username = None self.password = None def get(self, url, **kwargs): return self._make_request('get', url, **kwargs) def post(self, url, **kwargs): return self._make_request('post', url, **kwargs) def put(self, url, **kwargs): return self._make_request('put', url, **kwargs) def delete(self, url, **kwargs): return self._make_request('delete', url, **kwargs) def _make_request(self, method, url, **kwargs): """ Function is responsible for sending the API request. `method`: String representing what type of http request to make (i.e. get, put, post, delete) `url`: String representing the api endpoint to send the https request. Can provide the endpoint by itself (i.e. /Chassis/IOM1/NetworkAdapters) and the correct prefix will be added or you can provide the full url to the endpoint (i.e. http://ip-here/redfish/v1/endpoint-here) `kwargs['data']`: Dict representing the "payload" to send along with the http request. """ if method == 'get': req = requests.get elif method == 'post': req = requests.post elif method == 'put': req = requests.put elif method == 'delete': req = requests.delete else: raise ValueError(f'Invalid request type: {method}') if not url.startswith('https://'): if url.startswith(self.prefix): url = f'{self.base_url}{url}' else: url = f'{self.base_url}{self.prefix}{url}' if 'auth' in kwargs: auth = kwargs['auth'] else: auth = self.auth timeout = kwargs.get('timeout', self.timeout) payload = kwargs.get('data', {}) headers = kwargs.get('headers', {}) if self.log_requests: LOGGER.debug('%r %r %r', method.upper(), url, payload) if payload: if isinstance(payload, dict) or isinstance(payload, list): if headers.get('Content-Type', None) == 'multipart/form-data': # See python-redfish-library on how to handle if ever necessary raise ValueError('Currently do not support this content-type') else: headers['Content-Type'] = 'application/json' payload = json.dumps(payload) elif isinstance(payload, bytes): headers['Content-Type'] = 'application/octet-stream' payload = payload else: headers['Content-Type'] = 'application/x-www-form-urlencoded' payload = urlencode(payload) if self.authtype == AuthMethod.BASIC: r = req(url, auth=auth, verify=self.verify, headers=headers, data=payload, timeout=timeout) else: if self.auth_token: headers.update({'X-Auth-Token': self.auth_token}) r = req(url, verify=self.verify, headers=headers, data=payload, timeout=timeout) if r.status_code == 401: raise InvalidCredentialsError('HTTP 401 Unauthorized returned: Invalid credentials supplied') return r def configure_fabric_interface( self, uri, address, subnet_mask, dhcp_enabled=False, gateway='0.0.0.0', mtusize=5000, enabled=True ): return self.post(uri, data={ 'DHCPv4': {'DHCPEnabled': dhcp_enabled}, 'IPv4StaticAddresses': [{'Address': address, 'Gateway': gateway, 'SubnetMask': subnet_mask}], 'MTUSize': mtusize, 'InterfaceEnabled': enabled, }) def link_status(self, uri): r = self.get(uri) if r.ok: return r.json().get('LinkStatus') class AsyncRedfishClient(AbstractRedfishClient): """ Asynchronous Redfish client which supports multipath. Various instantiation mechanisms are available, including `cache_get` where objects will be cached for re-use. """ # Cache where objects will be stored, keyed by JBOF UUID. client_cache = {} def __init__( self, base_urls, username=None, password=None, authtype=AuthMethod.BASIC, default_prefix=REDFISH_ROOT_PATH, timeout=DEFAULT_REDFISH_TIMEOUT_SECS ): self.log_requests = False self.base_urls = [base_url.rstrip('/') for base_url in base_urls] self.username = username self.password = password self.authtype = AuthMethod.authtype_to_enum(authtype) self.prefix = default_prefix self.auth = None self.auth_token = None self.session_key = None self.authorization_key = None self.session_location = {} self.timeout = aiohttp.ClientTimeout(total=timeout) self.cache = {} self.root = None self._sessions = {} # Implement a little value cache self._attributes = {} self.logged = set() def _add_session(self, base_url, session): """Add a session corresponding to the base_url""" self._sessions[base_url] = session async def _del_session(self, base_url, session): """This is called when a session is no longer responding""" if session: # Don't attempt to logout. The session is kaput await session.close() if base_url in self._sessions: del self._sessions[base_url] def sessions(self): """Generator to yield the (base_url, session), good ones first""" good = set(self._sessions.keys()) bad = set(self.base_urls) - good for base_url in good: yield base_url, self._sessions[base_url] for base_url in bad: yield base_url, None def get_attribute(self, name, defval=None): """Retrieve some data associated with the client.""" return self._attributes.get(name, defval) def set_attribute(self, name, value): """Associate some data with the client.""" self._attributes[name] = value @classmethod async def create(cls, base_urls, username=None, password=None, authtype=AuthMethod.BASIC, default_prefix=REDFISH_ROOT_PATH, timeout=DEFAULT_REDFISH_TIMEOUT_SECS): """Async factory to create an AsyncRedfishClient object.""" self = cls(base_urls, username, password, authtype, default_prefix, timeout) self.root = await self.get_root_object() try: self.login_url = self.root['Links']['Sessions']['@odata.id'] except KeyError: self.login_url = REDFISH_SESSIONS return self @classmethod def cache_set(cls, key, value): cls.client_cache[key] = value @classmethod def cache_unset(cls, key): del cls.client_cache[key] @classmethod async def cache_get(cls, uuid, jbof_query=None): """Fetch AsyncRedfishClient object from cache, creating if necessary.""" try: return cls.client_cache[uuid] except KeyError: redfish, jbofs = None, list() filters, options = [['uuid', '=', uuid]], dict() if jbof_query is not None: jbofs = jbof_query else: with Client(f'ws+unix://{MIDDLEWARE_RUN_DIR}/middlewared-internal.sock', py_exceptions=True) as c: jbofs = c.call('jbof.query', filters) for jbof in filter_list(jbofs, filters, options): base_urls = [] for key in ['mgmt_ip1', 'mgmt_ip2']: if mgmt_ip := jbof.get(key): base_urls.append(f'https://{mgmt_ip}') redfish = await cls.create(base_urls, jbof['mgmt_username'], jbof['mgmt_password']) cls.cache_set(uuid, redfish) return redfish async def _login(self, base_url, username=None, password=None, authtype=None): """Login using the specified path and credentials.""" self.username = username if username else self.username self.password = password if password else self.password if authtype: self.authtype = AuthMethod.authtype_to_enum(authtype) async with aiohttp.ClientSession(base_url, timeout=self.timeout) as session: if self.authtype == AuthMethod.BASIC: auth = aiohttp.BasicAuth(self.username, self.password) async with session.get(self.login_url, ssl=False, auth=auth) as response: if not response.ok: raise InvalidCredentialsError('Could not authenticate credentials supplied') newsession = aiohttp.ClientSession(base_url, timeout=self.timeout, raise_for_status=True, auth=auth) elif self.authtype == AuthMethod.SESSION: data = {'UserName': self.username, 'Password': self.password} async with session.post(self.login_url, ssl=False, json=data) as response: if not response.ok: raise InvalidCredentialsError('Could not authenticate credentials supplied') auth_token = response.headers.get('X-Auth-Token') if auth_token: # Save the Location for logout purposes self.session_location[base_url] = response.headers.get('Location') newsession = aiohttp.ClientSession(base_url, timeout=self.timeout, raise_for_status=True, headers={'X-Auth-Token': auth_token}) else: raise ValueError('Invalid auth supplied:', authtype) # Save the session for reuse self._add_session(base_url, newsession) # Clear any silenced exception logging try: self.logged.remove(base_url) except KeyError: pass return newsession async def _logout(self, base_url, session): if self.authtype == AuthMethod.BASIC: self.auth = None elif self.authtype == AuthMethod.SESSION: if location := self.session_location.get(base_url): async with session.delete(location, ssl=False) as response: if response.ok: try: del self.session_location[base_url] except KeyError: # Should not occur, but protect in case parallel calls pass async def get_root_object(self): """Fetch the root object.""" # We're not going to try to be clever about which base_urls are used for this. base_url_count = len(self.base_urls) for index, base_url in enumerate(self.base_urls, 1): try: async with aiohttp.ClientSession(base_url, timeout=self.timeout, raise_for_status=True) as session: async with session.get(self.prefix, ssl=False) as response: return await response.json() except asyncio.TimeoutError: if index == base_url_count: raise CallError('Connection timed out', errno.ETIMEDOUT) else: continue except Exception: continue raise CallError('Failed to obtain root object', errno.EBADMSG) async def get(self, uri): if not uri.startswith(self.prefix): uri = f'{self.prefix}{uri}' # Iterate over the available paths for base_url, session in self.sessions(): try: if not session: session = await self._login(base_url) async with session.get(uri, ssl=False) as response: if response.ok: return await response.json() except asyncio.TimeoutError: LOGGER.debug('Timed out GET %r: %r', base_url, uri) await self._del_session(base_url, session) continue except Exception: if base_url not in self.logged: LOGGER.debug('Failed GET %r: %r', base_url, uri, exc_info=True) self.logged.add(base_url) await self._del_session(base_url, session) continue raise CallError(f'Failed to GET {uri}:', errno.EBADMSG) async def post(self, uri, **kwargs): if not uri.startswith(self.prefix): uri = f'{self.prefix}{uri}' payload = kwargs.get('data', {}) # Iterate over the available paths for base_url, session in self.sessions(): try: if not session: session = await self._login(base_url) async with session.post(uri, ssl=False, json=payload) as response: if response.ok: return await response.json() except asyncio.TimeoutError: LOGGER.debug('Timed out POST %r: %r', base_url, uri) await self._del_session(base_url, session) continue except Exception: if base_url not in self.logged: LOGGER.debug('Failed POST %r: %r', base_url, uri, exc_info=True) self.logged.add(base_url) await self._del_session(base_url, session) continue raise CallError(f'Failed to POST {uri}:', errno.EBADMSG) async def close(self): for base_url, session in self._sessions.items(): await self._logout(base_url, session) await session.close() self._sessions = {} async def _cached_fetch(self, cache_key, uri, use_cached=True): if use_cached and cache_key in self.cache: return self.cache[cache_key] r = await self.get(uri) if r: self.cache[cache_key] = self._members(r) return self.cache[cache_key] async def chassis(self, use_cached=True): return await self._cached_fetch('chassis', '/Chassis', use_cached) async def managers(self, use_cached=True): return await self._cached_fetch('managers', '/Managers', use_cached)
22,674
Python
.py
516
32.813953
143
0.593747
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,047
target.py
truenas_middleware/src/middlewared/middlewared/plugins/jbof/redfish/target.py
import requests import urllib3 from middlewared.service import Service DEFAULT_REDFISH_TIMEOUT_SECS = 10 HEADER = {'Content-Type': 'application/json', 'Vary': 'accept'} REDFISH_ROOT_PATH = '/redfish/v1' class JBOFRedfishService(Service): class Config: namespace = 'jbof.redfish' private = True def ping(self, mgmt_ip, timeout=DEFAULT_REDFISH_TIMEOUT_SECS): base_url = f'https://{mgmt_ip}/{REDFISH_ROOT_PATH}' r = requests.get(base_url, verify=False, timeout=timeout) try: return r.json() except requests.exceptions.JSONDecodeError: return None def is_redfish(self, mgmt_ip, timeout=DEFAULT_REDFISH_TIMEOUT_SECS): try: data = self.ping(mgmt_ip, timeout) if data: return "RedfishVersion" in data except requests.exceptions.Timeout: self.logger.debug('Failed to query redfish host %s', mgmt_ip) return False def _members(self, mgmt_ip, mgmt_username, mgmt_password, path, timeout=DEFAULT_REDFISH_TIMEOUT_SECS): """Return a list containing the names of the IO Managers.""" result = [] r = self.make_request(mgmt_ip, 'get', path, username=mgmt_username, password=mgmt_password, timeout=timeout) for member in r.get('Members', []): if member['@odata.id'].startswith(path): result.append(member['@odata.id'].split('/')[-1]) return result def managers(self, mgmt_ip, mgmt_username, mgmt_password, timeout=DEFAULT_REDFISH_TIMEOUT_SECS): """Return a list containing the names of the IO Managers.""" return self._members(mgmt_ip, mgmt_username, mgmt_password, f'{REDFISH_ROOT_PATH}/Managers', timeout) def ethernet_interfaces(self, mgmt_ip, mgmt_username, mgmt_password, manager, timeout=DEFAULT_REDFISH_TIMEOUT_SECS): """Return a list containing the names of the EthernetInterfaces in an IO Managers.""" return self._members(mgmt_ip, mgmt_username, mgmt_password, f'{REDFISH_ROOT_PATH}/Managers/{manager}/EthernetInterfaces', timeout) def ethernet_interface(self, mgmt_ip, mgmt_username, mgmt_password, manager, interface, timeout=DEFAULT_REDFISH_TIMEOUT_SECS): """Return a list containing the names of the EthernetInterfaces in an IO Managers.""" return self.make_request(mgmt_ip, 'get', f'{REDFISH_ROOT_PATH}/Managers/{manager}/EthernetInterfaces/{interface}', username=mgmt_username, password=mgmt_password, timeout=timeout) def make_request(self, mgmt_ip, _type, url, **kwargs): """ Function is responsible for sending the API request. `_type`: String representing what type of http request to make (i.e. get, put, post, delete) `url`: String representing the api endpoint to send the https request. Can provide the endpoint by itself (i.e. /Chassis/IOM1/NetworkAdapters) and the correct prefix will be added or you can provide the full url to the endpoint (i.e. http://ip-here/redfish/v1/endpoint-here) `kwargs['data']`: Dict representing the "payload" to send along with the http request. """ if _type == 'get': req = requests.get elif _type == 'post': req = requests.post elif _type == 'put': req = requests.put elif _type == 'delete': req = requests.delete else: raise ValueError(f'Invalid request type: {_type}') if not url.startswith('https://'): if url.startswith(REDFISH_ROOT_PATH): url = f'https://{mgmt_ip}{url}' else: url = f'https://{mgmt_ip}{REDFISH_ROOT_PATH}{url}' auth = kwargs.get('auth', None) if not auth: auth = (kwargs.get('username', 'Admin'), kwargs.get('password', '')) return req(url, auth=auth, verify=False, data=kwargs.get('data', {})).json() async def setup(middleware): # Silence InsecureRequestWarning: Unverified HTTPS request is being made to host requests.packages.urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
4,471
Python
.py
84
41.392857
120
0.61232
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,048
utils_snmp_user.py
truenas_middleware/src/middlewared/middlewared/plugins/snmp_/utils_snmp_user.py
import os import subprocess from cryptography.fernet import Fernet from logging import getLogger from middlewared.service import CallError from middlewared.utils.crypto import generate_string LOGGER = getLogger(__name__) class SNMPSystem(): # This is an snmpd auto-generated file. We use it to create the SNMPv3 users. PRIV_CONF = '/var/lib/snmp/snmpd.conf' # SNMP System User authentication PRIV_KEY = None # SNMP System User info SYSTEM_USER = { 'name': 'snmpSystemUser', 'auth_type': 'SHA', 'key': None, 'size': 0 } def _get_authuser_secret(): """ Get the auth user saved secret. Internal helper function for use by this module. Return decoded string. """ secret = "" if not SNMPSystem.SYSTEM_USER['key']: # No system user key registered LOGGER.debug("No system user key is registered") return secret if SNMPSystem.PRIV_KEY: secret = Fernet(SNMPSystem.SYSTEM_USER['key']).decrypt(SNMPSystem.PRIV_KEY).decode() return secret def _set_authuser_secret(secret): """ Save the auth user secret. Internal helper function for use by this module. INPUT: ascii string (not encoded) """ SNMPSystem.PRIV_KEY = Fernet(SNMPSystem.SYSTEM_USER['key']).encrypt(secret.encode()) # noqa: (F841, assigned but not used) return def _add_system_user(): """ Add the v3 system user. For internal use by this module. NOTES: SNMP must be stopped before calling. The private config file is assumed to be in a regenerated state with no v3 users """ SNMPSystem.SYSTEM_USER['key'] = Fernet.generate_key() auth_pwd = generate_string(32) priv_config = { 'v3_username': SNMPSystem.SYSTEM_USER['name'], 'v3_authtype': SNMPSystem.SYSTEM_USER['auth_type'], 'v3_password': f"{auth_pwd}" } add_snmp_user(priv_config) _set_authuser_secret(auth_pwd) def add_snmp_user(snmp): """ Build the createUser message and add it to the private config file. NOTE: The SNMP daemon should be stopped before calling this routine and the new user will be available after starting SNMP. """ # The private config file must exist, i.e. SNMP must have been started at least once if not os.path.exists(SNMPSystem.PRIV_CONF): return # BuilSNMPSystem. 'createUser' message create_v3_user = f"createUser {snmp['v3_username']} " user_pwd = snmp['v3_password'] create_v3_user += f'{snmp["v3_authtype"]} "{user_pwd}" ' if snmp.get('v3_privproto'): user_phrase = snmp['v3_privpassphrase'] create_v3_user += f'{snmp["v3_privproto"]} "{user_phrase}" ' create_v3_user += '\n' # Example: createUser newPrivUser MD5 "abcd1234" DES "abcd1234" with open(SNMPSystem.PRIV_CONF, 'a') as f: f.write(create_v3_user) def delete_snmp_user(user): """ Delete the SNMPv3 user RETURN: stdout message NOTE: SNMP must be running for this call to succeed """ if pwd := _get_authuser_secret(): # snmpusm -v3 -l authPriv -u JoeUser -a MD5 -A "abcd1234" -x AES -X "A pass phrase" localhost delete JoeUser cmd = [ 'snmpusm', '-v3', '-u', f'{SNMPSystem.SYSTEM_USER["name"]}', '-l', 'authNoPriv', '-a', f'{SNMPSystem.SYSTEM_USER["auth_type"]}', '-A', f'{pwd}', 'localhost', 'delete', user ] # This call will timeout if SNMP is not running subprocess.run(cmd, capture_output=True) else: raise CallError def get_users_cmd(): cmd = [] if pwd := _get_authuser_secret(): # snmpwalk -v3 -u ixAuthUser -l authNoPriv -a MD5 -A "abcd1234" localhost iso.3.6.1.6.3.15.1.2.2.1.3 cmd = ['snmpwalk', '-v3', '-u', f'{SNMPSystem.SYSTEM_USER["name"]}', '-l', 'authNoPriv', '-a', f'{SNMPSystem.SYSTEM_USER["auth_type"]}', '-A', f'{pwd}', 'localhost', 'iso.3.6.1.6.3.15.1.2.2.1.3'] else: LOGGER.debug("Unable to get authuser secret") return cmd
4,059
Python
.py
101
34.009901
127
0.64664
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,049
m_series_bios.py
truenas_middleware/src/middlewared/middlewared/plugins/hardware/m_series_bios.py
import datetime from middlewared.service import Service from middlewared.utils.functools_ import cache class MseriesBiosService(Service): class Config: private = True namespace = 'mseries.bios' @cache def is_old_version(self): chassis = self.middleware.call_sync("truenas.get_chassis_hardware") bios_dates = { "TRUENAS-M40": datetime.date(2020, 2, 20), "TRUENAS-M50": datetime.date(2020, 12, 3), "TRUENAS-M60": datetime.date(2020, 12, 3), } min_bios_date = next((v for k, v in bios_dates.items() if chassis.startswith(k)), None) if min_bios_date and (bios := self.middleware.call_sync("system.dmidecode_info")["bios-release-date"]): return bios < min_bios_date
782
Python
.py
18
35.722222
111
0.649539
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,050
mem_info.py
truenas_middleware/src/middlewared/middlewared/plugins/hardware/mem_info.py
from pathlib import Path from middlewared.service import Service from middlewared.schema import accepts, returns, Dict class HardwareMemoryService(Service): class Config: namespace = 'hardware.memory' cli_namespace = 'system.hardware.memory' @accepts() @returns(Dict('mem_ctrl', additional_attrs=True)) def error_info(self): results = {} mc_path = Path('/sys/devices/system/edac/mc') if not mc_path.exists(): return results dimm_or_rank = 'dimm' mc_idx = 0 for mc in filter(lambda x: x.is_dir() and x.name.startswith('mc'), mc_path.iterdir()): mc_info = {mc.name: {}} if mc_idx == 0 and not (mc / f'{dimm_or_rank}{mc_idx}').exists(): # AMD systems use "rank" as top-level dir while Intel uses dimm dimm_or_rank = 'rank' # top-level memory controller information for key, _file in ( ('corrected_errors', 'ce_count'), ('uncorrected_errors', 'ue_count'), ('corrected_errors_with_no_dimm_info', 'ce_noinfo_count'), ('uncorrected_errors_with_no_dimm_info', 'ue_noinfo_count'), ): try: value = int((mc / _file).read_text().strip()) except (FileNotFoundError, ValueError): value = None mc_info[mc.name].update({key: value}) # specific dimm module memory information for dimm in filter(lambda x: x.is_dir() and x.name.startswith(dimm_or_rank), mc.iterdir()): # looks like /sys/devices/edac/mc0/dimm(or rank){0/1/2} mc_info[mc.name][dimm.name] = {} for key, _file in ( ('corrected_errors', 'dimm_ce_count'), ('uncorrected_errors', 'dimm_ue_count'), ): try: value = int((dimm / _file).read_text().strip()) except (FileNotFoundError, ValueError): value = None mc_info[mc.name][dimm.name].update({key: value}) mc_idx += 1 results.update(mc_info) return results
2,265
Python
.py
49
32.653061
103
0.529946
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,051
m_series_nvdimm.py
truenas_middleware/src/middlewared/middlewared/plugins/hardware/m_series_nvdimm.py
import glob import re import subprocess from middlewared.service import Service class MseriesNvdimmService(Service): class Config: private = True namespace = 'mseries.nvdimm' def run_ixnvdimm(self, nvmem_dev): out = subprocess.run( ["ixnvdimm", nvmem_dev], stdout=subprocess.PIPE, stderr=subprocess.STDOUT, encoding="utf-8", errors="ignore", ).stdout specrev = subprocess.run( ['ixnvdimm', '-r', nvmem_dev, 'SPECREV'], stdout=subprocess.PIPE, stderr=subprocess.STDOUT, encoding="utf-8", errors="ignore", ).stdout return out, specrev def get_running_firmware_vers_and_detect_old_bios(self, output): result = {'running_firmware': None, 'old_bios': True} if m := re.search(r"selected: [0-9]+ running: ([0-9]+)", output): running_slot = int(m.group(1)) if m := re.search(rf"slot{running_slot}: ([0-9])([0-9])", output): result['running_firmware'] = f"{m.group(1)}.{m.group(2)}" result['old_bios'] = False return result def get_module_health(self, output): if (m := re.search(r"Module Health:[^\n]+", output)): return m.group().split("Module Health: ")[-1].strip() def vendor_info(self, output): mapping = { '0x2c80_0x4e32_0x31_0x3480_0x4131_0x01': { 'vendor': '0x2c80', 'device': '0x4e32', 'rev_id': '0x31', 'subvendor': '0x3480', 'subdevice': '0x4131', 'subrev_id': '0x01', 'part_num': '18ASF2G72PF12G6V21AB', 'size': '16GB', 'clock_speed': '2666MHz', 'qualified_firmware': ['2.6'], 'recommended_firmware': '2.6', }, '0x2c80_0x4e36_0x31_0x3480_0x4231_0x02': { 'vendor': '0x2c80', 'device': '0x4e36', 'rev_id': '0x31', 'subvendor': '0x3480', 'subdevice': '0x4231', 'subrev_id': '0x02', 'part_num': '18ASF2G72PF12G9WP1AB', 'size': '16GB', 'clock_speed': '2933MHz', 'qualified_firmware': ['2.2'], 'recommended_firmware': '2.2', }, '0x2c80_0x4e33_0x31_0x3480_0x4231_0x01': { 'vendor': '0x2c80', 'device': '0x4e33', 'rev_id': '0x31', 'subvendor': '0x3480', 'subdevice': '0x4231', 'subrev_id': '0x01', 'part_num': '36ASS4G72PF12G9PR1AB', 'size': '32GB', 'clock_speed': '2933MHz', 'qualified_firmware': ['2.4'], 'recommended_firmware': '2.4', }, '0xce01_0x4e38_0x33_0xc180_0x4331_0x01': { 'vendor': '0xce01', 'device': '0x4e38', 'rev_id': '0x33', 'subvendor': '0xc180', 'subdevice': '0x4331', 'subrev_id': '0x01', 'part_num': 'AGIGA8811-016ACA', 'size': '16GB', 'clock_speed': '2933MHz', 'qualified_firmware': ['0.8'], 'recommended_firmware': '0.8', }, '0xce01_0x4e42_0x31_0xc180_0x4331_0x01': { 'vendor': '0xce01', 'device': '0x4e42', 'rev_id': '0x31', 'subvendor': '0xc180', 'subdevice': '0x4331', 'subrev_id': '0x01', 'part_num': 'AGIGA8811-016BCA', 'size': '16GB', 'clock_speed': '2933MHz', 'qualified_firmware': ['3.0'], 'recommended_firmware': '3.0', }, '0xce01_0x4e39_0x34_0xc180_0x4331_0x01': { 'vendor': '0xce01', 'device': '0x4e39', 'rev_id': '0x34', 'subvendor': '0xc180', 'subdevice': '0x4331', 'subrev_id': '0x01', 'part_num': 'AGIGA8811-032ACA', 'size': '32GB', 'clock_speed': '2933MHz', 'qualified_firmware': ['0.8'], 'recommended_firmware': '0.8', }, 'unknown': { 'vendor': None, 'device': None, 'rev_id': None, 'subvendor': None, 'subdevice': None, 'subrev_id': None, 'part_num': None, 'size': None, 'clock_speed': None, 'qualified_firmware': [], 'recommended_firmware': None, } } result = mapping['unknown'] vend_key = subvend_key = None if (match := re.search(r'vendor: (?P<v>\w+) device: (?P<d>\w+) revision: (?P<r>\w+)', output)): vend_key = '_'.join([f'0x{v}' for v in match.groupdict().values()]) if (match := re.search(r'subvendor: (?P<v>\w+) subdevice: (?P<d>\w+) subrevision: (?P<r>\w+)', output)): subvend_key = '_'.join([f'0x{v}' for v in match.groupdict().values()]) if all((vend_key, subvend_key)): result = mapping.get(f'{vend_key}_{subvend_key}', mapping['unknown']) return result def health_info(self, output): result = { 'critical_health_info': {}, 'nvm_health_info': {}, 'nvm_error_threshold_status': {}, 'nvm_warning_threshold_status': {}, 'nvm_lifetime': None, 'nvm_temperature': None, 'es_lifetime': None, 'es_temperature': None, } if m := re.search(r'Critical Health Info: (.*)', output): bit, vals = m.group(1).split(' ', 1) result['critical_health_info'][bit] = [i for i in vals.lstrip('<').rstrip('>').split(',') if i] if m := re.search(r'Module Health: (.*)', output): bit, vals = m.group(1).split(' ', 1) result['nvm_health_info'][bit] = [i for i in vals.lstrip('<').rstrip('>').split(',') if i] if m := re.search(r'Error Threshold Status: (.*)', output): bit, vals = m.group(1).split(' ', 1) result['nvm_error_threshold_status'][bit] = [i for i in vals.lstrip('<').rstrip('>').split(',') if i] if m := re.search(r'Warning Threshold Status: (.*)', output): bit, vals = m.group(1).split(' ', 1) result['nvm_warning_threshold_status'][bit] = [i for i in vals.lstrip('<').rstrip('>').split(',') if i] if m := re.search(r'NVM Lifetime: (.*)', output): result['nvm_lifetime'] = m.group(1).split(' ', 1)[0] if m := re.search(r'Module Current Temperature: (.*)', output): result['nvm_temperature'] = m.group(1).split(' ', 1)[0] if m := re.search(r'ES Lifetime Percentage: (.*)', output): result['es_lifetime'] = m.group(1).split(' ', 1)[0] if m := re.search(r'ES Current Temperature: (.*)', output): result['es_temperature'] = m.group(1).split(' ', 1)[0] return result def state_flags(self, nmem): try: with open(f'/sys/bus/nd/devices/{nmem.removeprefix("/dev/")}/nfit/flags') as f: state_flags = f.read().strip().split() except Exception: state_flags = [] return state_flags def info(self): results = [] sys = ("TRUENAS-M40", "TRUENAS-M50", "TRUENAS-M60") if not self.middleware.call_sync("truenas.get_chassis_hardware").startswith(sys): return results try: for nmem in glob.glob("/dev/nmem*"): output, specrev = self.run_ixnvdimm(nmem) info = { 'index': int(nmem[len('/dev/nmem')]), 'dev': nmem.removeprefix('/dev/'), 'dev_path': nmem, 'specrev': int(specrev.strip()), 'state_flags': self.state_flags(nmem), } info.update(self.health_info(output)) info.update(self.vendor_info(output)) info.update(self.get_running_firmware_vers_and_detect_old_bios(output)) results.append(info) except Exception: self.logger.error("Unhandled exception obtaining nvdimm info", exc_info=True) else: return results
8,094
Python
.py
165
36.121212
115
0.50822
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,052
cpu.py
truenas_middleware/src/middlewared/middlewared/plugins/hardware/cpu.py
from pathlib import Path from middlewared.service import Service from middlewared.service_exception import ValidationError from middlewared.schema import accepts, returns, Dict, Str from middlewared.utils.functools_ import cache class HardwareCpuService(Service): class Config: namespace = 'hardware.cpu' cli_namespace = 'system.hardware.cpu' @accepts() @returns(Dict('governor', additional_attrs=True)) @cache def available_governors(self): """Return available cpu governors""" try: with open('/sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors') as f: return {i: i for i in f.read().split()} except FileNotFoundError: # doesn't support changing governor return dict() @accepts() @returns(Str('governor')) def current_governor(self): """Returns currently set cpu governor""" try: with open('/sys/devices/system/cpu/cpu0/cpufreq/scaling_governor') as f: return f.read().strip() except FileNotFoundError: # doesn't support changing governor return @accepts(Str('governor', required=True)) @returns() def set_governor(self, governor): """Set the cpu governor to `governor` on all cpus""" curr_gov = self.current_governor() if curr_gov is None: raise ValidationError('hardware.cpu.governor', 'Changing cpu governor is not supported') elif curr_gov == governor: # current governor is already set to what is being requested return elif governor not in self.available_governors(): raise ValidationError('hardware.cpu.governor', f'{governor} is not available') for i in Path('/sys/devices/system/cpu').iterdir(): if i.is_dir() and i.name.startswith('cpu'): cpug = (i / 'cpufreq/scaling_governor') if cpug.exists(): cpug.write_text(governor)
2,028
Python
.py
47
34.06383
100
0.641156
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,053
rdma.py
truenas_middleware/src/middlewared/middlewared/plugins/rdma/rdma.py
import json import subprocess from pathlib import Path from .constants import RDMAprotocols from middlewared.api import api_method from middlewared.api.current import ( RdmaLinkConfigArgs, RdmaLinkConfigResult, RdmaCardConfigArgs, RdmaCardConfigResult, RdmaCapableServicesArgs, RdmaCapableServicesResult ) from middlewared.service import Service, private from middlewared.service_exception import CallError from middlewared.utils.functools_ import cache from middlewared.plugins.rdma.interface import RDMAInterfaceService # noqa (just import to start the service) PRODUCT_NAME_PREFIX = 'Product Name: ' SERIAL_NUMBER_PREFIX = '[SN] Serial number: ' PART_NUMBER_PREFIX = '[PN] Part number: ' class RDMAService(Service): class Config: private = True @private def get_pci_vpd(self, pci_addr): lspci_cmd = ['lspci', '-vv', '-s', pci_addr] ret = subprocess.run(lspci_cmd, capture_output=True) if ret.returncode: error = ret.stderr.decode() if ret.stderr else ret.stdout.decode() if not error: error = 'No error message reported' self.logger.debug('Failed to execute command: %r with error: %r', " ".join(lspci_cmd), error) raise CallError(f'Failed to determine serial number/product: {error}') result = {} for line in ret.stdout.decode().split('\n'): sline = line.strip() if sline.startswith(PRODUCT_NAME_PREFIX): result['product'] = sline[len(PRODUCT_NAME_PREFIX):] elif sline.startswith(SERIAL_NUMBER_PREFIX): result['serial'] = sline[len(SERIAL_NUMBER_PREFIX):] elif sline.startswith(PART_NUMBER_PREFIX): result['part'] = sline[len(PART_NUMBER_PREFIX):] return result @api_method(RdmaLinkConfigArgs, RdmaLinkConfigResult, private=True) async def get_link_choices(self, all): """Return a list containing dictionaries with keys 'rdma' and 'netdev'. Unless all is set to True, configured interfaces will be excluded.""" all_links = await self.middleware.call('rdma._get_link_choices') if all: return all_links existing = await self.middleware.call('interface.get_configured_interfaces') return list(filter(lambda x: x['netdev'] not in existing, all_links)) @private @cache def _get_link_choices(self): """Return a list containing dictionaries with keys 'rdma' and 'netdev'. Since these are just the hardware present in the system, we cache the result.""" self.logger.info('Fetching RDMA link netdev choices') link_cmd = ['rdma', '-j', 'link'] ret = subprocess.run(link_cmd, capture_output=True) if ret.returncode: error = ret.stderr.decode() if ret.stderr else ret.stdout.decode() if not error: error = 'No error message reported' self.logger.debug('Failed to execute command: %r with error: %r', " ".join(link_cmd), error) raise CallError(f'Failed to determine RDMA links: {error}') result = [] for link in json.loads(ret.stdout.decode()): if 'netdev' in link: result.append({'rdma': link['ifname'], 'netdev': link['netdev']}) return result @api_method(RdmaCardConfigArgs, RdmaCardConfigResult) @cache def get_card_choices(self): """Return a list containing details about each RDMA card. Dual cards will contain two RDMA links.""" self.logger.info('Fetching RDMA card choices') links = self.middleware.call_sync('rdma.get_link_choices') grouper = {} for link in links: rdma = link["rdma"] p = Path(f'/sys/class/infiniband/{rdma}') if not p.is_symlink(): # Should never happen self.logger.debug(f'Not a symlink: {p}') continue pci_addr = p.readlink().parent.parent.name if ':' not in pci_addr: # Should never happen self.logger.debug(f'{rdma} symlink {p} does not yield a PCI address: {pci_addr}') continue vpd = self.middleware.call_sync('rdma.get_pci_vpd', pci_addr) serial = vpd.get('serial') if not serial: # Should never happen self.logger.debug(f'Could not find serial number for {rdma} / {pci_addr}') continue part_number = vpd.get('part', '') # We'll use part_number:serial as the key, just in case we had different # device types with the same serial number (unlikely) key = f'{part_number}:{serial}' if key not in grouper: grouper[key] = {'serial': serial, 'product': vpd.get('product', ''), 'part_number': part_number, 'links': [link]} else: grouper[key]['links'].append(link) # Now that we have finished processing, generate a name that can be used # to store in the database. We will concatenate the rdma names in each # card. for k, v in grouper.items(): names = [link['rdma'] for link in v['links']] v['name'] = ':'.join(sorted(names)) return list(grouper.values()) @api_method(RdmaCapableServicesArgs, RdmaCapableServicesResult) async def capable_services(self): result = [] is_ent = await self.middleware.call('system.is_enterprise') if is_ent and 'MINI' not in await self.middleware.call('truenas.get_chassis_hardware'): if await self.middleware.call('rdma.get_link_choices', True): result.append(RDMAprotocols.NFS.value) return result
5,884
Python
.py
120
38.6
110
0.618824
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,054
constants.py
truenas_middleware/src/middlewared/middlewared/plugins/rdma/constants.py
import enum class RDMAprotocols(enum.Enum): NFS = 'NFS' ISER = 'iSER' def values(): return [a.value for a in RDMAprotocols]
147
Python
.py
6
19.666667
47
0.652174
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,055
crud.py
truenas_middleware/src/middlewared/middlewared/plugins/rdma/interface/crud.py
import middlewared.sqlalchemy as sa from middlewared.schema import (Dict, Int, IPAddr, Patch, Str, accepts) from middlewared.service import CallError, CRUDService from pyroute2 import NDB from pyroute2.ndb.transaction import CheckProcessException class ConnectionChecker: def __init__(self, middleware, ifname, ip, mac=None): self.middleware = middleware self.ifname = ifname self.ip = ip self.mac = mac def commit(self): if not self.middleware.call_sync('rdma.interface.local_ping', self.ifname, self.ip, self.mac): raise CheckProcessException('CheckProcess failed') class RDMAInterfaceModel(sa.Model): __tablename__ = 'rdma_interface' __table_args__ = (sa.UniqueConstraint('rdmaif_node', 'rdmaif_ifname'),) id = sa.Column(sa.Integer(), primary_key=True, autoincrement=True) rdmaif_node = sa.Column(sa.String(120), nullable=False) rdmaif_ifname = sa.Column(sa.String(120), nullable=False) rdmaif_address = sa.Column(sa.String(45), nullable=False) rdmaif_prefixlen = sa.Column(sa.Integer(), nullable=False) rdmaif_mtu = sa.Column(sa.Integer(), nullable=False) class RDMAInterfaceService(CRUDService): class Config: namespace = 'rdma.interface' private = True cli_private = True datastore = 'rdma.interface' datastore_prefix = "rdmaif_" ENTRY = Dict( 'rdma_interface_entry', Str('id', required=True), Str('node', default=''), Str('ifname', required=True), IPAddr('address', required=True), Int('prefixlen', required=True), Int('mtu', default=5000), ) async def compress(self, data): if 'check' in data: del data['check'] @accepts( Patch( 'rdma_interface_entry', 'rdma_interface_create', ('rm', {'name': 'id'}), ('add', Dict('check', Str('ping_ip'), Str('ping_mac'))), ) ) async def do_create(self, data): result = await self.middleware.call('rdma.interface.configure_interface', data['node'], data['ifname'], data['address'], data['prefixlen'], data['mtu'], data.get('check')) if result: await self.compress(data) data['id'] = await self.middleware.call( 'datastore.insert', self._config.datastore, data, {'prefix': self._config.datastore_prefix}) return await self.get_instance(data['id']) else: return None @accepts( Int('id', required=True), Patch( 'rdma_interface_entry', 'rdma_interface_update', ('add', Dict('check', Str('ping_ip'), Str('ping_mac'))), ('attr', {'update': True}) ) ) async def do_update(self, id_, data): """ Update RDMA interface of `id` """ old = await self.get_instance(id_) new = old.copy() new.update(data) if old['node'] != new['node'] or old['ifname'] != new['ifname']: await self.middleware.call('rdma.interface.configure_interface', data['node'], data['ifname'], None) result = await self.middleware.call('rdma.interface.configure_interface', new['node'], new['ifname'], new['address'], new['prefixlen'], new['mtu'], new.get('check')) if result: await self.compress(new) await self.middleware.call( 'datastore.update', self._config.datastore, id_, new, {'prefix': self._config.datastore_prefix}) return await self.get_instance(id_) else: raise CallError("Failed to update active RDMA interface configuration") @accepts(Int('id')) async def do_delete(self, id_): """ Delete a RDMA interface by ID. """ data = await self.get_instance(id_) # Attempt to remove the live configuration result = await self.middleware.call('rdma.interface.configure_interface', data['node'], data['ifname'], None) if not result: self.logger.warn("Failed to delete active RDMA interface configuration") # Now delete the entry return await self.middleware.call('datastore.delete', self._config.datastore, id_) async def configure_interface(self, node, ifname, address, prefixlen=None, mtu=None, check=None): if not node or node == await self.middleware.call('failover.node'): # Local return await self.middleware.call('rdma.interface.local_configure_interface', ifname, address, prefixlen, mtu, check) else: # Remote try: return await self.middleware.call('failover.call_remote', 'rdma.interface.local_configure_interface', [ifname, address, prefixlen, mtu, check]) except CallError as e: if e.errno != CallError.ENOMETHOD: raise self.logger.warning('Failed to configure remote RDMA interface') return False def local_configure_interface(self, ifname, address, prefixlen=None, mtu=None, check=None): """Configure the interface.""" netdev = self.middleware.call_sync('rdma.interface.ifname_to_netdev', ifname) if not netdev: self.logger.error('Could not find netdev associated with %s', ifname) return False try: with NDB(log='off') as ndb: with ndb.interfaces[netdev] as dev: with ndb.begin() as ctx: # First we will check to see if any change is required dirty = True if address: for addr in dev.ipaddr: if address == addr['address'] and prefixlen == addr['prefixlen']: dirty = False if mtu and not dirty: dirty = mtu != dev['mtu'] # Now reconfigure if necessary if dirty: for addr in dev.ipaddr: ctx.push(dev.del_ip(address=addr['address'], prefixlen=addr['prefixlen'], family=addr['family'])) if mtu: dev['mtu'] = mtu if address: ctx.push(dev.add_ip(address=address, prefixlen=prefixlen).set('state', 'up')) else: # not dirty if dev['state'] != 'up': ctx.push(dev.set('state', 'up')) if check: ctx.push(ConnectionChecker(self.middleware, ifname, check['ping_ip'], check.get('ping_mac'))) if dirty and not address: with NDB(log='off') as ndb: with ndb.interfaces[netdev] as dev: dev.set('state', 'down') if check: self.logger.info(f'Validated communication of {netdev} with IP {check["ping_ip"]}') return True except CheckProcessException: self.logger.info(f'Failed to validate communication of {netdev} with IP {check["ping_ip"]}') except Exception: self.logger.error('Failed to configure RDMA interface', exc_info=True) return False async def ping(self, node, ifname, ip, mac=None): if not node or node == await self.middleware.call('failover.node'): # Local result = await self.middleware.call('rdma.interface.local_ping', ifname, ip, mac) else: # Remote try: result = await self.middleware.call('failover.call_remote', 'rdma.interface.local_ping', [ifname, ip, mac]) except CallError as e: if e.errno != CallError.ENOMETHOD: raise self.logger.warning('Failed to ping from remote RDMA interface') return False return result async def ifname_to_netdev(self, ifname): links = await self.middleware.call('rdma.get_link_choices') for link in links: if link['rdma'] == ifname: return link['netdev'] async def local_ping(self, ifname, ip, mac=None): netdev = await self.middleware.call('rdma.interface.ifname_to_netdev', ifname) if not await self.middleware.call('core.ping_remote', {'hostname': ip, 'timeout': 1, 'count': 4, 'interface': netdev, 'interval': '0.1'}): # Common case no logging needed return False if mac: macs = await self.middleware.call('core.arp', {'ip': ip, 'interface': netdev}) if ip not in macs: # If we can ping it, we should be able to get the MAC address self.middleware.logger.debug('Could not obtain arp info for IP: %s', ip) return False if macs[ip] != mac: # MAC address mismatch self.middleware.logger.debug('MAC address mismatch for IP %s', ip) return False return True async def internal_interfaces(self, all=False): # We must fetch all link choices. If we did not there would # be a circular call chain between interface and rdma links = await self.middleware.call('rdma._get_link_choices') ifname_to_netdev = {} for link in links: ifname_to_netdev[link['rdma']] = link['netdev'] if all: # Treat all RDMA interfaces as internal return list(ifname_to_netdev.values()) else: # Otherwise we only treat used RDMA interfaces as internal result = set() interfaces = await self.query() for interface in interfaces: result.add(ifname_to_netdev[interface['ifname']]) return list(result) async def configure(self): if await self.middleware.call('failover.licensed'): node = await self.middleware.call('failover.node') else: node = '' interfaces = await self.middleware.call('rdma.interface.query', [['node', '=', node]]) for interface in interfaces: await self.middleware.call('rdma.interface.local_configure_interface', interface['ifname'], interface['address'], interface['prefixlen'], interface['mtu']) return interfaces
11,653
Python
.py
240
32.55
117
0.521132
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,056
config.py
truenas_middleware/src/middlewared/middlewared/plugins/system_advanced/config.py
import os import re import tempfile import textwrap import warnings from copy import deepcopy import middlewared.sqlalchemy as sa from middlewared.schema import accepts, Bool, Dict, Int, List, Patch, Password, returns, Str from middlewared.service import ConfigService, private, no_auth_required, ValidationErrors from middlewared.validators import Range from middlewared.utils import run class SystemAdvancedModel(sa.Model): __tablename__ = 'system_advanced' id = sa.Column(sa.Integer(), primary_key=True) adv_consolemenu = sa.Column(sa.Boolean(), default=False) adv_serialconsole = sa.Column(sa.Boolean(), default=False) adv_serialport = sa.Column(sa.String(120), default='ttyS0') adv_serialspeed = sa.Column(sa.String(120), default='9600') adv_powerdaemon = sa.Column(sa.Boolean(), default=False) adv_overprovision = sa.Column(sa.Integer(), nullable=True, default=None) adv_traceback = sa.Column(sa.Boolean(), default=True) adv_advancedmode = sa.Column(sa.Boolean(), default=False) adv_autotune = sa.Column(sa.Boolean(), default=False) adv_debugkernel = sa.Column(sa.Boolean(), default=False) adv_uploadcrash = sa.Column(sa.Boolean(), default=True) adv_anonstats = sa.Column(sa.Boolean(), default=True) adv_anonstats_token = sa.Column(sa.Text()) adv_motd = sa.Column(sa.Text(), default='Welcome') adv_login_banner = sa.Column(sa.Text(), default='') adv_boot_scrub = sa.Column(sa.Integer(), default=7) adv_fqdn_syslog = sa.Column(sa.Boolean(), default=False) adv_sed_user = sa.Column(sa.String(120), default='user') adv_sed_passwd = sa.Column(sa.EncryptedText(), default='') adv_sysloglevel = sa.Column(sa.String(120), default='f_info') adv_syslogserver = sa.Column(sa.String(120), default='') adv_syslog_transport = sa.Column(sa.String(12), default='UDP') adv_syslog_tls_certificate_id = sa.Column(sa.ForeignKey('system_certificate.id'), index=True, nullable=True) adv_syslog_tls_certificate_authority_id = sa.Column( sa.ForeignKey('system_certificateauthority.id'), index=True, nullable=True ) adv_syslog_audit = sa.Column(sa.Boolean(), default=False) adv_kmip_uid = sa.Column(sa.String(255), nullable=True, default=None) adv_kdump_enabled = sa.Column(sa.Boolean(), default=False) adv_isolated_gpu_pci_ids = sa.Column(sa.JSON(), default=[]) adv_kernel_extra_options = sa.Column(sa.Text(), default='', nullable=False) class SystemAdvancedService(ConfigService): class Config: datastore = 'system.advanced' datastore_prefix = 'adv_' datastore_extend = 'system.advanced.system_advanced_extend' namespace = 'system.advanced' cli_namespace = 'system.advanced' role_prefix = 'SYSTEM_ADVANCED' ENTRY = Dict( 'system_advanced_entry', Bool('advancedmode', required=True), Bool('autotune', required=True), Bool('kdump_enabled', required=True), Int('boot_scrub', validators=[Range(min_=1)], required=True), Bool('consolemenu', required=True), Bool('consolemsg', required=True), Bool('debugkernel', required=True), Bool('fqdn_syslog', required=True), Str('motd', required=True), Str('login_banner', required=True, max_length=4096), Bool('powerdaemon', required=True), Bool('serialconsole', required=True), Str('serialport', required=True), Str('anonstats_token', required=True), Str('serialspeed', enum=['9600', '19200', '38400', '57600', '115200'], required=True), Int('overprovision', validators=[Range(min_=0)], null=True, required=True), Bool('traceback', required=True), Bool('uploadcrash', required=True), Bool('anonstats', required=True), Str('sed_user', enum=['USER', 'MASTER'], required=True), Str('sysloglevel', enum=[ 'F_EMERG', 'F_ALERT', 'F_CRIT', 'F_ERR', 'F_WARNING', 'F_NOTICE', 'F_INFO', 'F_DEBUG', ], required=True), Str('syslogserver'), Str('syslog_transport', enum=['UDP', 'TCP', 'TLS'], required=True), Int('syslog_tls_certificate', null=True, required=True), Int('syslog_tls_certificate_authority', null=True, required=True), Bool('syslog_audit'), List('isolated_gpu_pci_ids', items=[Str('pci_id')], required=True), Str('kernel_extra_options', required=True), Int('id', required=True), ) @private async def system_advanced_extend(self, data): data['consolemsg'] = (await self.middleware.call('system.general.config'))['ui_consolemsg'] if data.get('sed_user'): data['sed_user'] = data.get('sed_user').upper() for k in filter(lambda k: data[k], ['syslog_tls_certificate_authority', 'syslog_tls_certificate']): data[k] = data[k]['id'] data.pop('sed_passwd') data.pop('kmip_uid') return data async def __validate_fields(self, schema, data): verrors = ValidationErrors() serial_choice = data.get('serialport') if data.get('serialconsole'): if not serial_choice: verrors.add( f'{schema}.serialport', 'Please specify a serial port when serial console option is checked' ) elif serial_choice not in await self.middleware.call('system.advanced.serial_port_choices'): verrors.add( f'{schema}.serialport', 'Serial port specified has not been identified by the system' ) ups_port = (await self.middleware.call('ups.config'))['port'] if not verrors and os.path.join('/dev', serial_choice or '') == ups_port: verrors.add( f'{schema}.serialport', 'Serial port must be different then the port specified for UPS Service' ) syslog_server = data.get('syslogserver') if syslog_server: match = re.match(r"^\[?[\w\.\-\:\%]+\]?(\:\d+)?$", syslog_server) if not match: verrors.add( f'{schema}.syslogserver', 'Invalid syslog server format' ) elif ']:' in syslog_server or (':' in syslog_server and not ']' in syslog_server): port = int(syslog_server.split(':')[-1]) if port < 0 or port > 65535: verrors.add( f'{schema}.syslogserver', 'Port must be in the range of 0 to 65535.' ) if data['syslog_transport'] == 'TLS': if not data['syslog_tls_certificate_authority']: verrors.add( f'{schema}.syslog_tls_certificate_authority', 'This is required when using TLS as syslog transport' ) ca_cert = await self.middleware.call( 'certificateauthority.query', [['id', '=', data['syslog_tls_certificate_authority']]] ) if not ca_cert: verrors.add(f'{schema}.syslog_tls_certificate_authority', 'Unable to locate specified CA') elif ca_cert[0]['revoked']: verrors.add(f'{schema}.syslog_tls_certificate_authority', 'Specified CA has been revoked') if data['syslog_tls_certificate']: verrors.extend(await self.middleware.call( 'certificate.cert_services_validation', data['syslog_tls_certificate'], f'{schema}.syslog_tls_certificate', False )) elif data['syslog_tls_certificate_authority'] or data['syslog_tls_certificate']: data['syslog_tls_certificate_authority'] = data['syslog_tls_certificate'] = None for invalid_char in ('\n', '"'): if invalid_char in data['kernel_extra_options']: verrors.add('kernel_extra_options', f'{invalid_char!r} is an invalid character and not allowed') with tempfile.NamedTemporaryFile(mode="w+", encoding="utf-8") as f: f.write(textwrap.dedent(f"""\ menuentry 'TrueNAS' {{ linux /boot/vmlinuz {data['kernel_extra_options']} }} """)) f.flush() result = await run(["grub-script-check", f.name], check=False) if result.returncode != 0: verrors.add('kernel_extra_options', 'Invalid syntax') invalid_param = 'systemd.unified_cgroup_hierarchy' if invalid_param in data['kernel_extra_options']: # TODO: we don't normalize values being passed into us which # allows a comical amount of potential foot-shooting verrors.add('kernel_extra_options', f'Modifying {invalid_param!r} is not allowed') return verrors, data @accepts( Patch( 'system_advanced_entry', 'system_advanced_update', ('rm', {'name': 'id'}), ('rm', {'name': 'anonstats_token'}), ('rm', {'name': 'isolated_gpu_pci_ids'}), ('add', Password('sed_passwd')), ('attr', {'update': True}), ), audit='System advanced update' ) async def do_update(self, data): """ Update System Advanced Service Configuration. `consolemenu` should be disabled if the menu at console is not desired. It will default to standard login in the console if disabled. `autotune` when enabled executes autotune script which attempts to optimize the system based on the installed hardware. When `syslogserver` is defined, logs of `sysloglevel` or above are sent. If syslog_audit is also set then the remote syslog server will also receive audit messages. `consolemsg` is a deprecated attribute and will be removed in further releases. Please, use `consolemsg` attribute in the `system.general` plugin. """ consolemsg = None if 'consolemsg' in data: consolemsg = data.pop('consolemsg') warnings.warn("`consolemsg` has been deprecated and moved to `system.general`", DeprecationWarning) config_data = await self.config() config_data['sed_passwd'] = await self.sed_global_password() config_data.pop('consolemsg') original_data = deepcopy(config_data) config_data.update(data) verrors, config_data = await self.__validate_fields('advanced_settings_update', config_data) verrors.check() if config_data != original_data: if original_data.get('sed_user'): original_data['sed_user'] = original_data['sed_user'].lower() if config_data.get('sed_user'): config_data['sed_user'] = config_data['sed_user'].lower() if not config_data['sed_passwd'] and config_data['sed_passwd'] != original_data['sed_passwd']: # We want to make sure kmip uid is None in this case adv_config = await self.middleware.call('datastore.config', self._config.datastore) self.middleware.create_task( self.middleware.call('kmip.reset_sed_global_password', adv_config['adv_kmip_uid']) ) config_data['kmip_uid'] = None await self.middleware.call( 'datastore.update', self._config.datastore, config_data['id'], config_data, {'prefix': self._config.datastore_prefix} ) if original_data['boot_scrub'] != config_data['boot_scrub']: await self.middleware.call('service.restart', 'cron') generate_grub = original_data['kernel_extra_options'] != config_data['kernel_extra_options'] if original_data['motd'] != config_data['motd']: await self.middleware.call('etc.generate', 'motd') if original_data['login_banner'] != config_data['login_banner']: await self.middleware.call('service.reload', 'ssh') if original_data['powerdaemon'] != config_data['powerdaemon']: await self.middleware.call('service.restart', 'powerd') if original_data['fqdn_syslog'] != config_data['fqdn_syslog']: await self.middleware.call('service.restart', 'syslogd') if ( original_data['sysloglevel'].lower() != config_data['sysloglevel'].lower() or original_data['syslogserver'] != config_data['syslogserver'] or original_data['syslog_transport'] != config_data['syslog_transport'] or original_data['syslog_tls_certificate'] != config_data['syslog_tls_certificate'] or original_data['syslog_audit'] != config_data['syslog_audit'] or original_data['syslog_tls_certificate_authority'] != config_data['syslog_tls_certificate_authority'] ): await self.middleware.call('service.restart', 'syslogd') if config_data['sed_passwd'] and original_data['sed_passwd'] != config_data['sed_passwd']: await self.middleware.call('kmip.sync_sed_keys') if config_data['kdump_enabled'] != original_data['kdump_enabled']: # kdump changes require a reboot to take effect. So just generating the kdump config # should be enough await self.middleware.call('etc.generate', 'kdump') generate_grub = True if original_data['debugkernel'] != config_data['debugkernel']: generate_grub = True await self.middleware.call('system.advanced.configure_tty', original_data, config_data, generate_grub) if consolemsg is not None: await self.middleware.call('system.general.update', {'ui_consolemsg': consolemsg}) return await self.config() @accepts(roles=['SYSTEM_ADVANCED_READ']) @returns(Bool('sed_global_password_is_set')) async def sed_global_password_is_set(self): """Returns a boolean identifying whether or not a global SED password has been set""" return bool(await self.sed_global_password()) @accepts(roles=['SYSTEM_ADVANCED_READ']) @returns(Password('sed_global_password')) async def sed_global_password(self): """Returns configured global SED password in clear-text if one is configured, otherwise an empty string""" passwd = (await self.middleware.call( 'datastore.config', 'system.advanced', {'prefix': self._config.datastore_prefix} ))['sed_passwd'] return passwd if passwd else await self.middleware.call('kmip.sed_global_password') @no_auth_required @accepts() @returns(Str()) def login_banner(self): """Returns user set login banner""" return self.middleware.call_sync('datastore.config', 'system.advanced')['adv_login_banner']
15,025
Python
.py
277
43.256318
119
0.616206
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,057
serial.py
truenas_middleware/src/middlewared/middlewared/plugins/system_advanced/serial.py
from middlewared.schema import accepts, Dict, returns from middlewared.service import private, Service from middlewared.utils import run class SystemAdvancedService(Service): class Config: namespace = 'system.advanced' cli_namespace = 'system.advanced' @accepts() @returns(Dict('serial_port_choices', additional_attrs=True)) async def serial_port_choices(self): """ Get available choices for `serialport`. """ ports = {e['name']: e['name'] for e in await self.middleware.call('device.get_info', {'type': 'SERIAL'})} if not ports or (await self.middleware.call('system.advanced.config'))['serialport'] == 'ttyS0': # We should always add ttyS0 if ports is false or current value is the default one in db # i.e ttyS0 ports['ttyS0'] = 'ttyS0' return ports @private async def configure_tty(self, old, new, generate_grub=False): restart_ttys = any(old[k] != new[k] for k in ('serialconsole', 'serialspeed', 'serialport')) if old['serialconsole'] != new['serialconsole']: if old['serialport'] == new['serialport']: action = 'enable' if new['serialconsole'] else 'disable' cp = await run( ['systemctl', action, f'serial-getty@{old["serialport"]}.service'], check=False ) if cp.returncode: self.logger.error('Failed to %r serialconsole: %r', action, cp.stderr.decode()) if old['serialport'] != new['serialport']: for command in [ ['systemctl', 'disable', f'serial-getty@{old["serialport"]}.service'], ['systemctl', 'stop', f'serial-getty@{old["serialport"]}.service'], ] + ( [['systemctl', 'enable', f'serial-getty@{new["serialport"]}.service']] if new['serialconsole'] else [] ): cp = await run(command, check=False) if cp.returncode: self.logger.error( 'Failed to %r %r serialport service: %r', command[1], command[2], cp.stderr.decode() ) if restart_ttys or old['consolemenu'] != new['consolemenu']: serial_action = 'restart' if new['serialconsole'] else 'stop' cp = await run(['systemctl', serial_action, f'serial-getty@{new["serialport"]}.service'], check=False) if cp.returncode: self.middleware.logger.error( 'Failed to %r %r serial port: %r', serial_action, new['serialport'], cp.stderr.decode() ) if old['consolemenu'] != new['consolemenu']: cp = await run(['systemctl', 'restart', 'getty@tty1.service'], check=False) if cp.returncode: self.middleware.logger.error('Failed to restart tty service: %r', cp.stderr.decode()) if generate_grub or restart_ttys: await self.middleware.call('etc.generate', 'grub')
3,026
Python
.py
55
42.763636
118
0.582095
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,058
gpu.py
truenas_middleware/src/middlewared/middlewared/plugins/system_advanced/gpu.py
from middlewared.schema import accepts, Dict, List, returns, Str from middlewared.service import private, Service, ValidationErrors from middlewared.utils.gpu import get_gpus class SystemAdvancedService(Service): class Config: namespace = 'system.advanced' cli_namespace = 'system.advanced' @accepts(roles=['SYSTEM_ADVANCED_READ']) @returns(Dict(additional_attrs=True)) def get_gpu_pci_choices(self): """ This endpoint gives all the gpu pci ids/slots that can be isolated. """ configured_value = self.middleware.call_sync('system.advanced.config')['isolated_gpu_pci_ids'] gpus = { f'{gpu["description"]} [{gpu["addr"]["pci_slot"]}]': gpu['addr']['pci_slot'] for gpu in get_gpus() if not gpu['uses_system_critical_devices'] } for slot in filter(lambda gpu: gpu not in gpus.values(), configured_value): gpus[f'Unknown {slot!r} slot'] = slot return gpus @accepts(List('isolated_gpu_pci_ids', items=[Str('pci_id')], required=True), roles=['SYSTEM_ADVANCED_WRITE']) @returns() async def update_gpu_pci_ids(self, isolated_gpu_pci_ids): """ `isolated_gpu_pci_ids` is a list of PCI ids which are isolated from host system. """ verrors = ValidationErrors() if isolated_gpu_pci_ids: verrors = await self.validate_gpu_pci_ids(isolated_gpu_pci_ids, verrors, 'gpu_settings') if await self.middleware.call('system.is_ha_capable') and isolated_gpu_pci_ids: verrors.add( 'gpu_settings.isolated_gpu_pci_ids', 'HA capable systems do not support PCI passthrough' ) verrors.check() await self.middleware.call( 'datastore.update', 'system.advanced', (await self.middleware.call('system.advanced.config'))['id'], {'isolated_gpu_pci_ids': isolated_gpu_pci_ids}, {'prefix': 'adv_'} ) await self.middleware.call('boot.update_initramfs') @private async def validate_gpu_pci_ids(self, isolated_gpu_pci_ids, verrors, schema): available = set() critical_gpus = set() for gpu in await self.middleware.call('device.get_gpus'): available.add(gpu['addr']['pci_slot']) if gpu['uses_system_critical_devices']: critical_gpus.add(gpu['addr']['pci_slot']) provided = set(isolated_gpu_pci_ids) not_available = provided - available cannot_isolate = provided & critical_gpus if not_available: verrors.add( f'{schema}.isolated_gpu_pci_ids', f'{", ".join(not_available)} GPU pci slot(s) are not available or a GPU is not configured.' ) if cannot_isolate: verrors.add( f'{schema}.isolated_gpu_pci_ids', f'{", ".join(cannot_isolate)} GPU pci slot(s) consists of devices ' 'which cannot be isolated from host.' ) if len(available - provided) < 1: verrors.add( f'{schema}.isolated_gpu_pci_ids', 'A minimum of 1 GPU is required for the host to ensure it functions as desired.' ) return verrors
3,324
Python
.py
72
35.75
113
0.60315
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,059
syslog.py
truenas_middleware/src/middlewared/middlewared/plugins/system_advanced/syslog.py
from middlewared.schema import accepts, Dict, returns from middlewared.service import Service class SystemAdvancedService(Service): class Config: namespace = 'system.advanced' cli_namespace = 'system.advanced' @accepts() @returns(Dict( additional_attrs=True, title='Syslog Certificate Choices', )) async def syslog_certificate_choices(self): """ Return choices of certificates which can be used for `syslog_tls_certificate`. """ return { i['id']: i['name'] for i in await self.middleware.call('certificate.query', [('cert_type_CSR', '=', False)]) } @accepts() @returns(Dict( additional_attrs=True, title='Syslog Certificate Authority Choices', )) async def syslog_certificate_authority_choices(self): """ Return choices of certificate authorities which can be used for `syslog_tls_certificate_authority`. """ return { i['id']: i['name'] for i in await self.middleware.call('certificateauthority.query', [['revoked', '=', False]]) }
1,149
Python
.py
32
28.125
107
0.626799
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,060
device.py
truenas_middleware/src/middlewared/middlewared/plugins/virt/device.py
import usb.core from middlewared.service import CallError, Service from middlewared.api import api_method from middlewared.api.current import ( VirtDeviceUSBChoicesArgs, VirtDeviceUSBChoicesResult, VirtDeviceGPUChoicesArgs, VirtDeviceGPUChoicesResult, VirtDeviceDiskChoicesArgs, VirtDeviceDiskChoicesResult, ) class VirtDeviceService(Service): class Config: namespace = 'virt.device' cli_namespace = 'virt.device' @api_method(VirtDeviceUSBChoicesArgs, VirtDeviceUSBChoicesResult, roles=['VIRT_INSTANCE_READ']) def usb_choices(self): """ Provide choices for USB devices. """ choices = {} for i in usb.core.find(find_all=True): name = f'usb_{i.bus}_{i.address}' choices[name] = { 'vendor_id': format(i.idVendor, '04x'), 'product_id': format(i.idProduct, '04x'), 'bus': i.bus, 'dev': i.address, 'product': i.product, 'manufacturer': i.manufacturer, } return choices @api_method(VirtDeviceGPUChoicesArgs, VirtDeviceGPUChoicesResult, roles=['VIRT_INSTANCE_READ']) async def gpu_choices(self, instance_type, gpu_type): """ Provide choices for GPU devices. """ choices = {} if gpu_type != 'PHYSICAL': raise CallError('Only PHYSICAL type is supported for now.') if instance_type != 'CONTAINER': raise CallError('Only CONTAINER supported for now.') for i in await self.middleware.call('device.get_gpus'): if not i['available_to_host'] or i['uses_system_critical_devices']: continue choices[i['addr']['pci_slot']] = { 'bus': i['addr']['bus'], 'slot': i['addr']['slot'], 'description': i['description'], 'vendor': i['vendor'], } return choices @api_method(VirtDeviceDiskChoicesArgs, VirtDeviceDiskChoicesResult, roles=['VIRT_INSTANCE_READ']) async def disk_choices(self): """ Returns disk (zvol) choices for device type "DISK". """ out = {} zvols = await self.middleware.call( 'zfs.dataset.unlocked_zvols_fast', [ ['OR', [['attachment', '=', None], ['attachment.method', '=', 'virt.instance.query']]], ['ro', '=', False], ], {}, ['ATTACHMENT', 'RO'] ) for zvol in zvols: out[zvol['path']] = zvol['name'] return out
2,601
Python
.py
65
29.830769
103
0.577091
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,061
global.py
truenas_middleware/src/middlewared/middlewared/plugins/virt/global.py
from typing import TYPE_CHECKING import subprocess import middlewared.sqlalchemy as sa from middlewared.api import api_method from middlewared.api.current import ( VirtGlobalEntry, VirtGlobalUpdateArgs, VirtGlobalUpdateResult, VirtGlobalBridgeChoicesArgs, VirtGlobalBridgeChoicesResult, VirtGlobalPoolChoicesArgs, VirtGlobalPoolChoicesResult, VirtGlobalGetNetworkArgs, VirtGlobalGetNetworkResult, ) from middlewared.service import job, private from middlewared.service import ConfigService, ValidationErrors from middlewared.service_exception import CallError from middlewared.utils import run from middlewared.plugins.boot import BOOT_POOL_NAME_VALID from .utils import Status, incus_call if TYPE_CHECKING: from middlewared.main import Middleware INCUS_PATH = '/var/lib/incus' INCUS_BRIDGE = 'incusbr0' BRIDGE_AUTO = '[AUTO]' POOL_DISABLED = '[DISABLED]' class NoPoolConfigured(Exception): pass class LockedDataset(Exception): pass class VirtGlobalModel(sa.Model): __tablename__ = 'virt_global' id = sa.Column(sa.Integer(), primary_key=True) pool = sa.Column(sa.String(120), nullable=True) bridge = sa.Column(sa.String(120), nullable=True) v4_network = sa.Column(sa.String(120), nullable=True) v6_network = sa.Column(sa.String(120), nullable=True) class VirtGlobalService(ConfigService): class Config: datastore = 'virt_global' datastore_extend = 'virt.global.extend' namespace = 'virt.global' cli_namespace = 'virt.global' role_prefix = 'VIRT_GLOBAL' entry = VirtGlobalEntry @private async def extend(self, data): if data['pool']: data['dataset'] = f'{data["pool"]}/.ix-virt' else: data['dataset'] = None try: data['state'] = await self.middleware.call('cache.get', 'VIRT_STATE') except KeyError: data['state'] = Status.INITIALIZING.value return data @private async def validate(self, new: dict, schema_name: str, verrors: ValidationErrors): bridge = new['bridge'] if not bridge: bridge = BRIDGE_AUTO if bridge not in await self.bridge_choices(): verrors.add(f'{schema_name}.bridge', 'Invalid bridge') if bridge == BRIDGE_AUTO: new['bridge'] = None pool = new['pool'] if not pool: pool = POOL_DISABLED if pool not in await self.pool_choices(): verrors.add(f'{schema_name}.pool', 'Invalid pool') if pool == POOL_DISABLED: new['pool'] = None @api_method(VirtGlobalUpdateArgs, VirtGlobalUpdateResult) @job() async def do_update(self, job, data): """ Update global virtualization settings. `pool` which pool to use to store instances. None will disable the service. `bridge` which bridge interface to use by default. None means it will automatically create one. """ old = await self.config() new = old.copy() new.update(data) verrors = ValidationErrors() await self.validate(new, 'virt_global_update', verrors) verrors.check() # Not part of the database new.pop('state') new.pop('dataset') await self.middleware.call( 'datastore.update', self._config.datastore, new['id'], new, ) job = await self.middleware.call('virt.global.setup') await job.wait(raise_error=True) return await self.config() @api_method(VirtGlobalBridgeChoicesArgs, VirtGlobalBridgeChoicesResult, roles=['VIRT_GLOBAL_READ']) async def bridge_choices(self): """ Bridge choices for virtualization purposes. Empty means it will be managed/created automatically. """ choices = {BRIDGE_AUTO: 'Automatic'} # We do not allow custom bridge on HA because it might have bridge STP issues # causing failover problems. if not await self.middleware.call('failover.licensed'): choices.update({ i['name']: i['name'] for i in await self.middleware.call('interface.query', [['type', '=', 'BRIDGE']]) }) return choices @api_method(VirtGlobalPoolChoicesArgs, VirtGlobalPoolChoicesResult, roles=['VIRT_GLOBAL_READ']) async def pool_choices(self): """ Pool choices for virtualization purposes. """ pools = {POOL_DISABLED: '[Disabled]'} for p in (await self.middleware.call('zfs.pool.query_imported_fast')).values(): if p['name'] in BOOT_POOL_NAME_VALID: continue ds = await self.middleware.call( 'pool.dataset.get_instance_quick', p['name'], {'encryption': True}, ) if not ds['locked']: pools[p['name']] = p['name'] return pools @private async def internal_interfaces(self): return [INCUS_BRIDGE] @private async def check_initialized(self, config=None): if config is None: config = await self.config() if config['state'] != Status.INITIALIZED.value: raise CallError('Virtualization not initialized.') @private async def get_default_profile(self): result = await incus_call('1.0/profiles/default', 'get') if result.get('status_code') != 200: raise CallError(result.get('error')) return result['metadata'] @api_method(VirtGlobalGetNetworkArgs, VirtGlobalGetNetworkResult, roles=['VIRT_GLOBAL_READ']) async def get_network(self, name): """ Details for the given network. """ await self.check_initialized() result = await incus_call(f'1.0/networks/{name}', 'get') if result.get('status_code') != 200: raise CallError(result.get('error')) data = result['metadata'] return { 'type': data['type'].upper(), 'managed': data['managed'], 'ipv4_address': data['config']['ipv4.address'], 'ipv4_nat': data['config']['ipv4.nat'], 'ipv6_address': data['config']['ipv6.address'], 'ipv6_nat': data['config']['ipv6.nat'], } @private @job() async def setup(self, job): """ Sets up incus through their API. Will create necessary storage datasets if required. """ try: await self.middleware.call('cache.put', 'VIRT_STATE', Status.INITIALIZING.value) await self._setup_impl(job) except NoPoolConfigured: await self.middleware.call('cache.put', 'VIRT_STATE', Status.NO_POOL.value) except LockedDataset: await self.middleware.call('cache.put', 'VIRT_STATE', Status.LOCKED.value) except Exception: await self.middleware.call('cache.put', 'VIRT_STATE', Status.ERROR.value) raise else: await self.middleware.call('cache.put', 'VIRT_STATE', Status.INITIALIZED.value) async def _setup_impl(self, job): config = await self.config() if not config['pool']: if await self.middleware.call('service.started', 'incus'): job = await self.middleware.call('virt.global.reset') await job.wait(raise_error=True) self.logger.debug('No pool set for virtualization, skipping.') raise NoPoolConfigured() else: await self.middleware.call('service.start', 'incus') try: ds = await self.middleware.call( 'zfs.dataset.get_instance', config['dataset'], { 'extra': { 'retrieve_children': False, 'user_properties': False, 'properties': ['encryption', 'keystatus'], } }, ) except Exception: ds = None if not ds: await self.middleware.call('zfs.dataset.create', { 'name': config['dataset'], 'properties': { 'aclmode': 'discard', 'acltype': 'posix', 'exec': 'on', 'casesensitivity': 'sensitive', 'atime': 'off', }, }) else: if ds['encrypted'] and not ds['key_loaded']: self.logger.info('Dataset %r not unlocked, skipping virt setup.', ds['name']) raise LockedDataset() import_storage = True storage = await incus_call('1.0/storage-pools/default', 'get') if storage['type'] != 'error': if storage['metadata']['config']['source'] == config['dataset']: self.logger.debug('Storage pool for virt already configured.') import_storage = False else: job = await self.middleware.call('virt.global.reset', True) await job.wait(raise_error=True) # If no bridge interface has been set, use incus managed if not config['bridge']: result = await incus_call(f'1.0/networks/{INCUS_BRIDGE}', 'get') # Create INCUS_BRIDGE if it doesn't exist if result.get('status') != 'Success': # Reuse v4/v6 network from database if there is one result = await incus_call('1.0/networks', 'post', {'json': { 'config': { 'ipv4.address': config['v4_network'] or 'auto', 'ipv4.nat': 'true', 'ipv6.address': config['v6_network'] or 'auto', 'ipv6.nat': 'true', }, 'description': '', 'name': INCUS_BRIDGE, 'type': 'bridge', }}) if result.get('status_code') != 200: raise CallError(result.get('error')) result = await incus_call(f'1.0/networks/{INCUS_BRIDGE}', 'get') if result.get('status_code') != 200: raise CallError(result.get('error')) # Update automatically selected networks into our database # so it can persist upgrades. await self.middleware.call('datastore.update', 'virt_global', config['id'], { 'v4_network': result['metadata']['config']['ipv4.address'], 'v6_network': result['metadata']['config']['ipv6.address'], }) nic = { 'name': 'eth0', 'network': INCUS_BRIDGE, 'type': 'nic', } else: nic = { 'name': 'eth0', 'type': 'nic', 'nictype': 'bridged', 'parent': config['bridge'], } if import_storage: payload = { 'pools': [{ 'config': {'source': config['dataset']}, 'description': '', 'name': 'default', 'driver': 'zfs', }], } result = await incus_call('internal/recover/validate', 'post', {'json': payload}) if result.get('status') == 'Success': if result['metadata']['DependencyErrors']: raise CallError('Missing depedencies: ' + ', '.join(result['metadata']['DependencyErrors'])) result = await incus_call('internal/recover/import', 'post', {'json': payload}) if result.get('status') != 'Success': raise CallError(result.get('error')) else: raise CallError('Invalid storage') result = await incus_call('1.0/profiles/default', 'put', {'json': { 'config': {}, 'description': 'Default TrueNAS profile', 'devices': { 'root': { 'path': '/', 'pool': 'default', 'type': 'disk', }, 'eth0': nic, }, }}) if result.get('status') != 'Success': raise CallError(result.get('error')) # If storage was imported we need to restart incus service so instances # with autostart can be started if import_storage: await self.middleware.call('service.restart', 'incus') @private @job() async def reset(self, job, start: bool = False): config = await self.config() if await self.middleware.call('service.started', 'incus'): # Stop running instances params = [ [i['id'], {'force': True, 'timeout': 10}] for i in await self.middleware.call( 'virt.instance.query', [('status', '=', 'RUNNING')] ) ] job = await self.middleware.call('core.bulk', 'virt.instance.stop', params, 'Stopping instances') await job.wait() if await self.middleware.call('virt.instance.query', [('status', '=', 'RUNNING')]): raise CallError('Failed to stop instances') await self.middleware.call('service.stop', 'incus') if await self.middleware.call('service.started', 'incus'): raise CallError('Failed to stop virtualization service') if not config['bridge']: # Make sure we delete in case it exists try: await run(['ip', 'link', 'show', INCUS_BRIDGE], check=True) except subprocess.CalledProcessError: pass else: await run(['ip', 'link', 'delete', INCUS_BRIDGE], check=True) # Have incus start fresh # Use subprocess because shutil.rmtree will traverse filesystems # and we do have instances datasets that might be mounted beneath await run(f'rm -rf --one-file-system {INCUS_PATH}/*', shell=True, check=True) if start and not await self.middleware.call('service.start', 'incus'): raise CallError('Failed to start virtualization service') async def _event_system_ready(middleware: 'Middleware', event_type, args): if not await middleware.call('failover.licensed'): middleware.create_task(middleware.call('virt.global.setup')) async def setup(middleware: 'Middleware'): middleware.event_subscribe('system.ready', _event_system_ready) # Should only happen if middlewared crashes or during development failover_licensed = await middleware.call('failover.licensed') ready = await middleware.call('system.ready') if ready and not failover_licensed: await middleware.call('virt.global.setup')
14,865
Python
.py
342
32.040936
112
0.570945
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,062
utils.py
truenas_middleware/src/middlewared/middlewared/plugins/virt/utils.py
import asyncio import aiohttp import enum from collections.abc import Callable from .websocket import IncusWS from middlewared.service import CallError SOCKET = '/var/lib/incus/unix.socket' HTTP_URI = 'http://unix.socket' class Status(enum.Enum): INITIALIZING = 'INITIALIZING' INITIALIZED = 'INITIALIZED' NO_POOL = 'NO_POOL' LOCKED = 'LOCKED' ERROR = 'ERROR' async def incus_call(path: str, method: str, request_kwargs: dict = None): async with aiohttp.UnixConnector(path=SOCKET) as conn: async with aiohttp.ClientSession(connector=conn) as session: methodobj = getattr(session, method) r = await methodobj(f'{HTTP_URI}/{path}', **(request_kwargs or {})) return await r.json() async def incus_call_and_wait( path: str, method: str, request_kwargs: dict = None, running_cb: Callable[[dict], None] = None, timeout: int = 300, ): result = await incus_call(path, method, request_kwargs) if result.get('type') == 'error': raise CallError(result['error']) async def callback(data): if data['metadata']['status'] == 'Failure': return ('ERROR', data['metadata']['err']) if data['metadata']['status'] == 'Success': return ('SUCCESS', data['metadata']['metadata']) if data['metadata']['status'] == 'Running': if running_cb: await running_cb(data) return ('RUNNING', None) task = asyncio.ensure_future(IncusWS().wait(result['metadata']['id'], callback)) try: await asyncio.wait_for(task, timeout) except asyncio.TimeoutError: raise CallError('Timed out') return task.result()
1,693
Python
.py
42
33.880952
84
0.650397
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,063
instance_device.py
truenas_middleware/src/middlewared/middlewared/plugins/virt/instance_device.py
import errno from typing import Any from middlewared.service import ( CallError, Service, ValidationErrors, private ) from middlewared.api import api_method from middlewared.api.current import ( VirtInstanceDeviceListArgs, VirtInstanceDeviceListResult, VirtInstanceDeviceAddArgs, VirtInstanceDeviceAddResult, VirtInstanceDeviceDeleteArgs, VirtInstanceDeviceDeleteResult, ) from .utils import incus_call_and_wait class VirtInstanceDeviceService(Service): class Config: namespace = 'virt.instance' cli_namespace = 'virt.instance' @api_method(VirtInstanceDeviceListArgs, VirtInstanceDeviceListResult, roles=['VIRT_INSTANCE_READ']) async def device_list(self, id): """ List all devices associated to an instance. """ instance = await self.middleware.call('virt.instance.get_instance', id, {'extra': {'raw': True}}) # Grab devices from default profile (e.g. nic and disk) profile = await self.middleware.call('virt.global.get_default_profile') # Flag devices from the profile as readonly, cannot be modified only overridden raw_devices = profile['devices'] for v in raw_devices.values(): v['readonly'] = True raw_devices.update(instance['raw']['devices']) devices = [] for k, v in raw_devices.items(): if (device := await self.incus_to_device(k, v)) is not None: devices.append(device) return devices @private async def incus_to_device(self, name: str, incus: dict[str, Any]): device = {'name': name, 'readonly': incus.get('readonly') or False} def unsupported(): self.logger.trace('Proxy device not supported by API, skipping.') return None match incus['type']: case 'disk': device['dev_type'] = 'DISK' device['source'] = incus.get('source') device['destination'] = incus.get('path') case 'nic': device['dev_type'] = 'NIC' device['network'] = incus.get('network') case 'proxy': device['dev_type'] = 'PROXY' # For now follow docker lead for simplification # only allowing to bind on host (host -> container) if incus.get('bind') == 'instance': return unsupported() proto, addr, ports = incus['listen'].split(':') if proto == 'unix' or '-' in ports or ',' in ports: return unsupported() device['source_proto'] = proto.upper() device['source_port'] = int(ports) proto, addr, ports = incus['connect'].split(':') if proto == 'unix' or '-' in ports or ',' in ports: return unsupported() device['dest_proto'] = proto.upper() device['dest_port'] = int(ports) case 'tpm': device['dev_type'] = 'TPM' device['path'] = incus.get('path') device['pathrm'] = incus.get('pathrm') case 'usb': device['dev_type'] = 'USB' if incus.get('busnum') is not None: device['bus'] = int(incus['busnum']) if incus.get('devnum') is not None: device['dev'] = int(incus['devnum']) if incus.get('productid') is not None: device['product_id'] = incus['productid'] if incus.get('vendorid') is not None: device['vendir_id'] = incus['vendorid'] case 'gpu': device['dev_type'] = 'USB' device['id'] = incus['id'] device['gpu_type'] = incus['gputype'].upper() match incus['gputype']: case 'physical': pass case 'mdev': pass case 'mig': device['mig_uuid'] = incus['mig.uuid'] case 'sriov': pass case _: return unsupported() return device @private async def device_to_incus(self, instance_type: str, device: dict[str, Any]) -> dict[str, Any]: new = {} match device['dev_type']: case 'DISK': new['type'] = 'disk' source = device.get('source') or '' if not source.startswith(('/dev/zvol/', '/mnt/')): raise CallError('Only pool paths are allowed.') new['source'] = device['source'] if source.startswith('/mnt/'): if source.startswith('/mnt/.ix-apps'): raise CallError('Invalid source') if not device.get('destination'): raise CallError('Destination is required for filesystem paths.') if instance_type == 'VM': raise CallError('Destination is not valid for VM') new['path'] = device['destination'] case 'NIC': new['type'] = 'nic' new['network'] = device['network'] case 'PROXY': new['type'] = 'proxy' # For now follow docker lead for simplification # only allowing to bind on host (host -> container) new['bind'] = 'host' new['listen'] = f'{device["source_proto"].lower()}:0.0.0.0:{device["source_port"]}' new['connect'] = f'{device["dest_proto"].lower()}:0.0.0.0:{device["dest_port"]}' case 'USB': new['type'] = 'usb' if device.get('bus') is not None: new['busnum'] = str(device['bus']) if device.get('dev') is not None: new['devnum'] = str(device['dev']) if device.get('product_id') is not None: new['productid'] = device['product_id'] if device.get('vendor_id') is not None: new['vendorid'] = device['vendor_id'] case 'TPM': new['type'] = 'tpm' if device.get('path'): if instance_type == 'VM': raise CallError('Path is not valid for VM') new['path'] = device['path'] elif instance_type == 'CONTAINER': new['path'] = '/dev/tpm0' if device.get('pathrm'): if instance_type == 'VM': raise CallError('Pathrm is not valid for VM') new['pathrm'] = device['pathrm'] elif instance_type == 'CONTAINER': new['pathrm'] = '/dev/tpmrm0' case 'GPU': new['type'] = 'gpu' # new['id'] = device['id'] match device['gpu_type']: case 'PHYSICAL': new['gputype'] = 'physical' new['pci'] = device['pci'] case 'MDEV': new['gputype'] = 'mdev' case 'MIG': new['gputype'] = 'mig' if not device.get('mig_uuid'): raise CallError('UUID is required for MIG') new['mig.uuid'] = device['mig_uuid'] case 'SRIOV': new['gputype'] = 'sriov' case _: raise Exception('Invalid device type') return new async def __generate_device_name(self, device_names: list[str], device_type: str) -> str: name = device_type.lower() i = 0 while True: new_name = f'{name}{i}' if new_name not in device_names: name = new_name break i += 1 return name async def __validate_device(self, device, schema, verrors: ValidationErrors): match device['dev_type']: case 'PROXY': verror = await self.middleware.call('port.validate_port', schema, device['source_port']) verrors.extend(verror) case 'DISK': if device['source'] and device['source'].startswith('/dev/zvol/'): if device['source'] not in await self.middleware.call('virt.device.disk_choices'): verrors.add(schema, 'Invalid ZVOL choice.') @api_method(VirtInstanceDeviceAddArgs, VirtInstanceDeviceAddResult, roles=['VIRT_INSTANCE_WRITE']) async def device_add(self, id, device): """ Add a device to an instance. """ instance = await self.middleware.call('virt.instance.get_instance', id, {'extra': {'raw': True}}) data = instance['raw'] if device['name'] is None: device['name'] = await self.__generate_device_name(data['devices'].keys(), device['dev_type']) verrors = ValidationErrors() await self.__validate_device(device, 'virt_device_add', verrors) verrors.check() data['devices'][device['name']] = await self.device_to_incus(instance['type'], device) await incus_call_and_wait(f'1.0/instances/{id}', 'put', {'json': data}) return True @api_method(VirtInstanceDeviceDeleteArgs, VirtInstanceDeviceDeleteResult, roles=['VIRT_INSTANCE_DELETE']) async def device_delete(self, id, device): """ Delete a device from an instance. """ instance = await self.middleware.call('virt.instance.get_instance', id, {'extra': {'raw': True}}) data = instance['raw'] if device not in data['devices']: raise CallError('Device not found.', errno.ENOENT) data['devices'].pop(device) await incus_call_and_wait(f'1.0/instances/{id}', 'put', {'json': data}) return True
10,039
Python
.py
211
33.099526
109
0.515306
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,064
websocket.py
truenas_middleware/src/middlewared/middlewared/plugins/virt/websocket.py
import asyncio from collections.abc import Callable from collections import defaultdict from typing import TYPE_CHECKING import aiohttp import logging from middlewared.service import CallError if TYPE_CHECKING: from middlewared.main import Middleware logger = logging.getLogger(__name__) SOCKET = '/var/lib/incus/unix.socket' class Singleton(type): instance = None def __call__(cls, *args, **kwargs): if cls.instance is None: cls.instance = super(Singleton, cls).__call__(*args, **kwargs) return cls.instance class IncusWS(object, metaclass=Singleton): def __init__(self, middleware): self.middleware = middleware self._incoming = defaultdict(list) self._waiters = defaultdict(list) self._task = None async def run(self): while True: try: await self._run_impl() except aiohttp.client_exceptions.UnixClientConnectorError as e: logger.warning('Failed to connect to incus socket: %r', e) except Exception: logger.warning('Incus websocket failure', exc_info=True) await asyncio.sleep(1) async def _run_impl(self): async with aiohttp.UnixConnector(path=SOCKET) as conn: async with aiohttp.ClientSession(connector=conn) as session: async with session.ws_connect('ws://unix.socket/1.0/events') as ws: async for msg in ws: if msg.type != aiohttp.WSMsgType.TEXT: continue data = msg.json() match data['type']: case 'operation': if 'metadata' in data and 'id' in data['metadata']: self._incoming[data['metadata']['id']].append(data) for i in self._waiters[data['metadata']['id']]: i.set() case 'logging': if data['metadata']['message'] == 'Instance agent started': self.middleware.send_event( 'virt.instance.agent_running', 'CHANGED', id=data['metadata']['context']['instance'], ) async def wait(self, id: str, callback: Callable[[str], None]): event = asyncio.Event() self._waiters[id].append(event) try: while True: if not self._incoming[id]: await event.wait() event.clear() for i in list(self._incoming[id]): self._incoming[id].remove(i) if (result := await callback(i)) is None: continue status, data = result match status: case 'SUCCESS': return data case 'ERROR': raise CallError(data) case 'RUNNING': pass case _: raise CallError(f'Unknown status: {status}') finally: self._waiters[id].remove(event) async def start(self): if not self._task: self._task = asyncio.ensure_future(self.run()) async def stop(self): if self._task: self._task.cancel() self._task = None async def __event_system_shutdown(middleware, event_type, args): await IncusWS().stop() async def setup(middleware: 'Middleware'): middleware.event_register( 'virt.instance.agent_running', 'Agent is running on guest.', roles=['VIRT_INSTANCE_READ'], ) IncusWS(middleware) middleware.event_subscribe('system.shutdown', __event_system_shutdown)
4,023
Python
.py
92
28.391304
98
0.515865
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,065
instance.py
truenas_middleware/src/middlewared/middlewared/plugins/virt/instance.py
import aiohttp from middlewared.service import ( CRUDService, ValidationErrors, filterable, job, private ) from middlewared.utils import filter_list from middlewared.api import api_method from middlewared.api.current import ( VirtInstanceEntry, VirtInstanceCreateArgs, VirtInstanceCreateResult, VirtInstanceUpdateArgs, VirtInstanceUpdateResult, VirtInstanceDeleteArgs, VirtInstanceDeleteResult, VirtInstanceStartArgs, VirtInstanceStartResult, VirtInstanceStopArgs, VirtInstanceStopResult, VirtInstanceRestartArgs, VirtInstanceRestartResult, VirtInstanceImageChoicesArgs, VirtInstanceImageChoicesResult, ) from .utils import incus_call, incus_call_and_wait LC_IMAGES_SERVER = 'https://images.linuxcontainers.org' LC_IMAGES_JSON = f'{LC_IMAGES_SERVER}/streams/v1/images.json' class VirtInstanceService(CRUDService): class Config: namespace = 'virt.instance' cli_namespace = 'virt.instance' entry = VirtInstanceEntry role_prefix = 'VIRT_INSTANCE' @filterable async def query(self, filters, options): """ Query all instances with `query-filters` and `query-options`. """ config = await self.middleware.call('virt.global.config') if config['state'] != 'INITIALIZED': return [] results = (await incus_call('1.0/instances?filter=&recursion=2', 'get'))['metadata'] entries = [] for i in results: # If entry has no config or state its probably in an unknown state, skip it if not i.get('config') or not i.get('state'): continue entry = { 'id': i['name'], 'name': i['name'], 'type': 'CONTAINER' if i['type'] == 'container' else 'VM', 'status': i['state']['status'].upper(), 'cpu': i['config'].get('limits.cpu'), 'autostart': i['config'].get('boot.autostart') or False, 'environment': {}, 'aliases': [], } if options.get('extra').get('raw'): entry['raw'] = i if memory := i['config'].get('limits.memory'): # Handle all units? e.g. changes done through CLI if memory.endswith('MiB'): memory = int(memory[:-3]) * 1024 * 1024 else: memory = None entry['memory'] = memory for k, v in i['config'].items(): if not k.startswith('environment.'): continue entry['environment'][k[12:]] = v entries.append(entry) for v in (i['state']['network'] or {}).values(): for address in v['addresses']: if address['scope'] != 'global': continue entry['aliases'].append({ 'type': address['family'].upper(), 'address': address['address'], 'netmask': int(address['netmask']), }) return filter_list(entries, filters, options) @private async def validate(self, new, schema_name, verrors, old=None): if not old and await self.query([('name', '=', new['name'])]): verrors.add(f'{schema_name}.name', f'Name {new["name"]!r} already exists') # Do not validate image_choices because its an expansive operation, just fail on creation def __data_to_config(self, data): config = {} if data.get('environment'): for k, v in data['environment'].items(): config[f'environment.{k}'] = v if data.get('cpu'): config['limits.cpu'] = data['cpu'] if data.get('memory'): config['limits.memory'] = str(int(data['memory'] / 1024 / 1024)) + 'MiB' if data.get('autostart') is not None: config['boot.autostart'] = str(data['autostart']).lower() return config @api_method(VirtInstanceImageChoicesArgs, VirtInstanceImageChoicesResult, roles=['VIRT_INSTANCE_READ']) async def image_choices(self, data): """ Provice choices for instance image from a remote repository. """ choices = {} if data['remote'] == 'LINUX_CONTAINERS': url = LC_IMAGES_JSON async with aiohttp.ClientSession() as session: async with session.get(url) as resp: data = await resp.json() for v in data['products'].values(): alias = v['aliases'].split(',', 1)[0] choices[alias] = { 'label': f'{v["os"]} {v["release"]} ({v["arch"]}, {v["variant"]})', 'os': v['os'], 'release': v['release'], 'arch': v['arch'], 'variant': v['variant'], } return choices @api_method(VirtInstanceCreateArgs, VirtInstanceCreateResult) @job() async def do_create(self, job, data): """ Create a new virtualizated instance. """ await self.middleware.call('virt.global.check_initialized') verrors = ValidationErrors() await self.validate(data, 'virt_instance_create', verrors) devices = {} for i in (data['devices'] or []): await self.middleware.call('virt.instance.validate', i, 'virt_instance_create', verrors) devices[i['name']] = await self.middleware.call('virt.instance.device_to_incus', data['instance_type'], i) verrors.check() async def running_cb(data): if 'metadata' in data['metadata'] and (metadata := data['metadata']['metadata']): if 'download_progress' in metadata: job.set_progress(None, metadata['download_progress']) if 'create_instance_from_image_unpack_progress' in metadata: job.set_progress(None, metadata['create_instance_from_image_unpack_progress']) if data['remote'] == 'LINUX_CONTAINERS': url = LC_IMAGES_SERVER source = { 'type': 'image', } result = await incus_call(f'1.0/images/{data["image"]}', 'get') if result['status_code'] == 200: source['fingerprint'] = result['metadata']['fingerprint'] else: source.update({ 'server': url, 'protocol': 'simplestreams', 'mode': 'pull', 'alias': data['image'], }) await incus_call_and_wait('1.0/instances', 'post', {'json': { 'name': data['name'], 'ephemeral': False, 'config': self.__data_to_config(data), 'devices': devices, 'source': source, 'type': 'container' if data['instance_type'] == 'CONTAINER' else 'virtual-machine', 'start': True, }}, running_cb) return await self.middleware.call('virt.instance.get_instance', data['name']) @api_method(VirtInstanceUpdateArgs, VirtInstanceUpdateResult) @job() async def do_update(self, job, id, data): """ Update instance. """ await self.middleware.call('virt.global.check_initialized') instance = await self.middleware.call('virt.instance.get_instance', id, {'extra': {'raw': True}}) verrors = ValidationErrors() await self.validate(data, 'virt_instance_create', verrors, old=instance) verrors.check() instance['raw']['config'].update(self.__data_to_config(data)) await incus_call_and_wait(f'1.0/instances/{id}', 'put', {'json': instance['raw']}) return await self.middleware.call('virt.instance.get_instance', id) @api_method(VirtInstanceDeleteArgs, VirtInstanceDeleteResult) @job() async def do_delete(self, job, id): """ Delete an instance. """ await self.middleware.call('virt.global.check_initialized') instance = await self.middleware.call('virt.instance.get_instance', id) if instance['status'] == 'RUNNING': await incus_call_and_wait(f'1.0/instances/{id}/state', 'put', {'json': { 'action': 'stop', 'timeout': -1, 'force': True, }}) await incus_call_and_wait(f'1.0/instances/{id}', 'delete') return True @api_method(VirtInstanceStartArgs, VirtInstanceStartResult, roles=['VIRT_INSTANCE_WRITE']) @job() async def start(self, job, id): """ Start an instance. """ await self.middleware.call('virt.global.check_initialized') await incus_call_and_wait(f'1.0/instances/{id}/state', 'put', {'json': { 'action': 'start', }}) return True @api_method(VirtInstanceStopArgs, VirtInstanceStopResult, roles=['VIRT_INSTANCE_WRITE']) @job() async def stop(self, job, id, data): """ Stop an instance. Timeout is how long it should wait for the instance to shutdown cleanly. """ await self.middleware.call('virt.global.check_initialized') await incus_call_and_wait(f'1.0/instances/{id}/state', 'put', {'json': { 'action': 'stop', 'timeout': data['timeout'], 'force': data['force'], }}) return True @api_method(VirtInstanceRestartArgs, VirtInstanceRestartResult, roles=['VIRT_INSTANCE_WRITE']) @job() async def restart(self, job, id, data): """ Restart an instance. Timeout is how long it should wait for the instance to shutdown cleanly. """ await self.middleware.call('virt.global.check_initialized') instance = await self.middleware.call('virt.instance.get_instance', id) if instance['state'] == 'RUNNING': await incus_call_and_wait(f'1.0/instances/{id}/state', 'put', {'json': { 'action': 'stop', 'timeout': data['timeout'], 'force': data['force'], }}) await incus_call_and_wait(f'1.0/instances/{id}/state', 'put', {'json': { 'action': 'start', }}) return True
10,298
Python
.py
230
33.556522
118
0.566853
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,066
stats.py
truenas_middleware/src/middlewared/middlewared/plugins/virt/stats.py
import time from middlewared.event import EventSource from middlewared.schema import Dict, Str from middlewared.plugins.reporting.realtime_reporting.cgroup import get_cgroup_stats class VirtInstacesMetricsEventSource(EventSource): ACCEPTS = Dict( Str('id'), ) def run_sync(self): instance_id = self.arg['id'] while not self._cancel_sync.is_set(): netdata_metrics = None # TODO: code duplication with reporting.realtime # this gathers the most recent metric recorded via netdata (for all charts) retries = 2 while retries > 0: try: netdata_metrics = self.middleware.call_sync('netdata.get_all_metrics') except Exception: retries -= 1 if retries <= 0: raise time.sleep(0.5) else: break data = {} if not bool(netdata_metrics): data['error'] = True data['errname'] = 'FAILED_TO_CONNECT' else: data.update(get_cgroup_stats(netdata_metrics, [instance_id])[instance_id]) data['error'] = False data['errname'] = None self.send_event('ADDED', fields=data) time.sleep(2) async def setup(middleware): middleware.register_event_source( 'virt.instance.metrics', VirtInstacesMetricsEventSource, roles=['VIRT_INSTANCE_READ'] )
1,550
Python
.py
39
27.512821
93
0.567802
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,067
attachments.py
truenas_middleware/src/middlewared/middlewared/plugins/virt/attachments.py
from typing import TYPE_CHECKING from middlewared.common.attachment import FSAttachmentDelegate from middlewared.common.ports import PortDelegate if TYPE_CHECKING: from middlewared.main import Middleware class VirtFSAttachmentDelegate(FSAttachmentDelegate): name = 'virt' title = 'Virtualization' async def query(self, path, enabled, options=None): config = await self.middleware.call('virt.global.config') instances = [] for i in await self.middleware.call('virt.instance.query'): append = False if path != f'/mnt/{config["pool"]}': for device in await self.middleware.call('virt.instance.device_list', i['id']): if device['dev_type'] != 'DISK': continue if device['source'] is None: continue if await self.middleware.call('filesystem.is_child', device['source'], path): append = True break else: append = True if append: instances.append({ 'id': i['id'], 'name': i['name'], }) return instances async def delete(self, attachments): if attachments: job = await self.middleware.call('virt.global.update', {'pool': ''}) await job.wait(raise_error=True) async def toggle(self, attachments, enabled): for attachment in attachments: action = 'start' if enabled else 'stop' try: job = await self.middleware.call(f'virt.instance.{action}', attachment['id']) await job.wait(raise_error=True) except Exception as e: self.middleware.logger.warning('Unable to %s %r: %s', action, attachment['id'], e) async def stop(self, attachments): await self.toggle(attachments, False) async def start(self, attachments): await self.toggle(attachments, True) class VirtPortDelegate(PortDelegate): name = 'virt devices' namespace = 'virt.device' title = 'Virtualization Device' async def get_ports(self): ports = [] for instance in await self.middleware.call('virt.instance.query'): instance_ports = [] for device in await self.middleware.call('virt.instance.device_list', instance['id']): if device['dev_type'] != 'PROXY': continue instance_ports.append(('0.0.0.0', device['source_port'])) instance_ports.append(('::', device['source_port'])) if instance_ports: ports.append({ 'description': f'{instance["id"]!r} instance', 'ports': instance_ports, }) return ports async def setup(middleware: 'Middleware'): middleware.create_task( middleware.call( 'pool.dataset.register_attachment_delegate', VirtFSAttachmentDelegate(middleware), ) ) await middleware.call('port.register_attachment_delegate', VirtPortDelegate(middleware))
3,198
Python
.py
73
31.90411
98
0.581537
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,068
client.py
truenas_middleware/src/middlewared/middlewared/plugins/apps_images/client.py
import aiohttp import aiohttp.client_exceptions import async_timeout import asyncio import urllib.parse from middlewared.service import CallError from .utils import ( DEFAULT_DOCKER_REGISTRY, DOCKER_AUTH_SERVICE, DOCKER_AUTH_HEADER, DOCKER_AUTH_URL, DOCKER_CONTENT_DIGEST_HEADER, DOCKER_MANIFEST_LIST_SCHEMA_V2, DOCKER_MANIFEST_SCHEMA_V1, DOCKER_MANIFEST_SCHEMA_V2, DOCKER_RATELIMIT_URL, parse_auth_header, parse_digest_from_schema, ) class ContainerRegistryClientMixin: @staticmethod async def _api_call(url, options=None, headers=None, mode='get'): options = options or {} timeout = options.get('timeout', 15) assert mode in ('get', 'head') response = {'error': None, 'response': {}, 'response_obj': None} try: async with async_timeout.timeout(timeout): async with aiohttp.ClientSession( raise_for_status=True, trust_env=True, ) as session: req = await getattr(session, mode)(url, headers=headers) except asyncio.TimeoutError: response['error'] = f'Unable to connect with {url} in {timeout} seconds.' except aiohttp.ClientResponseError as e: response['error'] = str(e) else: response['response_obj'] = req if req.status != 200: response['error'] = f'Received response code {req.status}' + ( f' ({req.content})' if req.content else '' ) else: try: response['response'] = await req.json() except aiohttp.client_exceptions.ContentTypeError as e: # quay.io registry returns malformed content type header which aiohttp fails to parse # even though the content returned by registry is valid json response['error'] = f'Unable to parse response: {e}' except asyncio.TimeoutError: response['error'] = 'Timed out waiting for a response' return response async def _get_token(self, scope, auth_url=DOCKER_AUTH_URL, service=DOCKER_AUTH_SERVICE): query_params = urllib.parse.urlencode({ 'service': service, 'scope': scope, }) response = await self._api_call(f'{auth_url}?{query_params}') if response['error']: raise CallError(f'Unable to retrieve token for {scope!r}: {response["error"]}') return response['response']['token'] async def _get_manifest_response(self, registry, image, tag, headers, mode, raise_error): manifest_url = f'https://{registry}/v2/{image}/manifests/{tag}' # 1) try getting manifest response = await self._api_call(manifest_url, headers=headers, mode=mode) if (error := response['error']) and isinstance(error, aiohttp.ClientResponseError): if error.status == 401: # 2) try to get token from manifest api call's response headers auth_data = parse_auth_header(error.headers[DOCKER_AUTH_HEADER]) headers['Authorization'] = f'Bearer {await self._get_token(**auth_data)}' # 3) Redo the manifest call with updated token response = await self._api_call(manifest_url, headers=headers, mode=mode) if raise_error and response['error']: raise CallError( f'Unable to retrieve latest image digest for registry={registry} ' f'image={image} tag={tag}: {response["error"]}' ) return response async def get_manifest_call_headers(self, registry, image, headers): if registry == DEFAULT_DOCKER_REGISTRY: headers['Authorization'] = f'Bearer {await self._get_token(scope=f"repository:{image}:pull")}' return headers async def _get_repo_digest(self, registry, image, tag): response = await self._get_manifest_response( registry, image, tag, await self.get_manifest_call_headers(registry, image, { 'Accept': (f'{DOCKER_MANIFEST_SCHEMA_V2}, ' f'{DOCKER_MANIFEST_LIST_SCHEMA_V2}, ' f'{DOCKER_MANIFEST_SCHEMA_V1}') }), 'get', True ) digests = parse_digest_from_schema(response) digests.append(response['response_obj'].headers.get(DOCKER_CONTENT_DIGEST_HEADER)) return digests async def get_docker_hub_rate_limit_preview(self): return await self._api_call( url=DOCKER_RATELIMIT_URL, headers={'Authorization': f'Bearer {await self._get_token(scope="repository:ratelimitpreview/test:pull")}'}, mode='head' )
4,747
Python
.py
91
40.67033
120
0.61512
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,069
update_alerts.py
truenas_middleware/src/middlewared/middlewared/plugins/apps_images/update_alerts.py
import contextlib import logging from collections import defaultdict from middlewared.service import CallError, Service from .client import ContainerRegistryClientMixin from .utils import normalize_reference logger = logging.getLogger('docker_image') class ContainerImagesService(Service, ContainerRegistryClientMixin): class Config: namespace = 'app.image.op' private = True IMAGE_CACHE = defaultdict(lambda: False) async def get_update_cache(self, normalized=False): return { normalize_reference(i)['complete_tag']: v for i, v in self.IMAGE_CACHE.items() } if normalized else self.IMAGE_CACHE def normalize_reference(self, reference: str) -> dict: return normalize_reference(reference=reference) async def check_update(self): images = await self.middleware.call('app.image.query') for image in images: for tag in image['repo_tags']: try: await self.check_update_for_image(tag, image) except CallError as e: logger.error(str(e)) async def retrieve_digest(self, reference: str): repo_digests = [] parsed_reference = self.normalize_reference(reference=reference) with contextlib.suppress(CallError): repo_digests = await self._get_repo_digest( parsed_reference['registry'], parsed_reference['image'], parsed_reference['tag'], ) return repo_digests async def check_update_for_image(self, tag, image_details): if not image_details['dangling']: parsed_reference = self.normalize_reference(tag) self.IMAGE_CACHE[tag] = await self.compare_id_digests( image_details, parsed_reference['registry'], parsed_reference['image'], parsed_reference['tag'] ) async def clear_update_flag_for_tag(self, tag): self.IMAGE_CACHE[tag] = False async def compare_id_digests(self, image_details, registry, image_str, tag_str): """ Returns whether an update is available for an image. """ digest = await self._get_repo_digest(registry, image_str, tag_str) return not any( digest.split('@', 1)[-1] == upstream_digest for upstream_digest in digest for digest in image_details['repo_digests'] ) if image_details['repo_digests'] else False async def remove_from_cache(self, image): for tag in image['repo_tags']: self.IMAGE_CACHE.pop(tag, None)
2,648
Python
.py
60
34.25
90
0.635939
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,070
images.py
truenas_middleware/src/middlewared/middlewared/plugins/apps_images/images.py
from middlewared.plugins.apps.ix_apps.docker.images import delete_image, list_images, pull_image from middlewared.schema import accepts, Bool, Dict, Int, List, returns, Str from middlewared.service import CRUDService, filterable, job from middlewared.utils import filter_list from .utils import parse_tags class AppImageService(CRUDService): class Config: cli_namespace = 'app.image' namespace = 'app.image' role_prefix = 'APPS' ENTRY = Dict( 'app_image_entry', Str('id'), List('repo_tags', items=[Str('repo_tag')]), List('repo_digests', items=[Str('repo_digest')]), Int('size'), Bool('dangling'), Bool('update_available'), Str('created'), Str('author'), Str('comment'), List( 'parsed_repo_tags', items=[Dict( 'parsed_repo_tag', Str('image'), Str('tag'), Str('registry'), Str('complete_tag'), additional_attrs=True, )] ), additional_attrs=True, ) @filterable def query(self, filters, options): """ Query all docker images with `query-filters` and `query-options`. `query-options.extra.parse_tags` is a boolean which when set will have normalized tags to be retrieved. """ if not self.middleware.call_sync('docker.state.validate', False): return filter_list([], filters, options) update_cache = self.middleware.call_sync('app.image.op.get_update_cache') parse_all_tags = options['extra'].get('parse_tags') images = [] for image in list_images(): config = { k if isinstance(k, str) else k[0]: image.get(k) if isinstance(k, str) else image.get(*k) for k in ( 'id', ('repo_tags', []), ('repo_digests', []), 'size', 'created', 'author', 'comment', ) } config.update({ 'dangling': len(config['repo_tags']) == 1 and config['repo_tags'][0] == '<none>:<none>', 'update_available': any(update_cache[r] for r in config['repo_tags']), }) if parse_all_tags: config['parsed_repo_tags'] = parse_tags(config['repo_tags']) images.append(config) return filter_list(images, filters, options) @accepts( Dict( 'image_pull', Dict( 'auth_config', Str('username', required=True), Str('password', required=True, max_length=4096), default=None, null=True, ), Str('image', required=True), ), roles=['APPS_WRITE'] ) @returns() @job() def pull(self, job, data): """ `image` is the name of the image to pull. Format for the name is "registry/repo/image:v1.2.3" where registry may be omitted and it will default to docker registry in this case. It can or cannot contain the tag - this will be passed as is to docker so this should be analogous to what `docker pull` expects. `auth_config` should be specified if image to be retrieved is under a private repository. """ def callback(entry): nonlocal job # Just having some sanity checks in place in case we come across some weird registry if not isinstance(entry, dict) or any( k not in entry for k in ('progressDetail', 'status') ) or entry['status'].lower().strip() not in ('pull complete', 'downloading'): return if entry['status'].lower().strip() == 'pull complete': job.set_progress(95, 'Image downloaded, doing post processing') return progress = entry['progressDetail'] if not isinstance(progress, dict) or any( k not in progress for k in ('current', 'total') ) or progress['current'] > progress['total']: return job.set_progress((progress['current']/progress['total']) * 90, 'Pulling image') self.middleware.call_sync('docker.state.validate') auth_config = data['auth_config'] or {} image_tag = data['image'] pull_image(image_tag, callback, auth_config.get('username'), auth_config.get('password')) job.set_progress(100, f'{image_tag!r} image pulled successfully') @accepts( Str('image_id'), Dict( 'options', Bool('force', default=False), ) ) def do_delete(self, image_id, options): """ Delete docker image `image_id`. `options.force` when set will force delete the image regardless of the state of containers and should be used cautiously. """ self.middleware.call_sync('docker.state.validate') image = self.get_instance__sync(image_id) delete_image(image_id, options['force']) self.middleware.call_sync('app.image.op.remove_from_cache', image) return True
5,137
Python
.py
119
32.478992
115
0.573485
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,071
utils.py
truenas_middleware/src/middlewared/middlewared/plugins/apps_images/utils.py
import re from collections import defaultdict from middlewared.service import CallError # Default values DEFAULT_DOCKER_REGISTRY = 'registry-1.docker.io' DEFAULT_DOCKER_REPO = 'library' DEFAULT_DOCKER_TAG = 'latest' DOCKER_CONTENT_DIGEST_HEADER = 'Docker-Content-Digest' # Taken from OCI: https://github.com/opencontainers/go-digest/blob/master/digest.go#L63 DIGEST_RE = r'[a-z0-9]+(?:[.+_-])*:[a-zA-Z0-9=_-]+' DOCKER_AUTH_HEADER = 'WWW-Authenticate' DOCKER_AUTH_URL = 'https://auth.docker.io/token' DOCKER_AUTH_SERVICE = 'registry.docker.io' DOCKER_MANIFEST_SCHEMA_V1 = 'application/vnd.docker.distribution.manifest.v1+json' DOCKER_MANIFEST_SCHEMA_V2 = 'application/vnd.docker.distribution.manifest.v2+json' DOCKER_MANIFEST_LIST_SCHEMA_V2 = 'application/vnd.docker.distribution.manifest.list.v2+json' DOCKER_RATELIMIT_URL = 'https://registry-1.docker.io/v2/ratelimitpreview/test/manifests/latest' def parse_digest_from_schema(response: dict) -> list[str]: """ Parses out the digest according to schemas specs: https://docs.docker.com/registry/spec/manifest-v2-1/ """ media_type = response['response']['mediaType'] if media_type == DOCKER_MANIFEST_SCHEMA_V2: digest_value = response['response']['config']['digest'] return [digest_value] if isinstance(digest_value, str) else digest_value elif media_type == DOCKER_MANIFEST_LIST_SCHEMA_V2: if manifests := response['response']['manifests']: return [digest['digest'] for digest in manifests] return [] def parse_auth_header(header: str) -> dict[str, str]: """ Parses header in format below: 'Bearer realm="https://ghcr.io/token",service="ghcr.io",scope="redis:pull"' Returns: { 'auth_url': 'https://ghcr.io/token', 'service': 'ghcr.io', 'scope': 'redis:pull' } """ adapter = { 'realm': 'auth_url', 'service': 'service', 'scope': 'scope', } results = {} parts = header.split() if len(parts) > 1: for part in parts[1].split(','): key_value = part.split('=') if len(key_value) == 2 and key_value[0] in adapter: results[adapter[key_value[0]]] = key_value[1].strip('"') return results def normalize_reference(reference: str) -> dict: """ Parses the reference for image, tag and repository. Most of the logic has been used from docker engine to make sure we follow the same rules/practices for normalising the image name / tag """ # This needs to be done as containerd automatically adds docker.io as a registery which can't be queried by us # when checking for update alerts as registry-1.docker.io is the one used to actually query that information reference = reference.removeprefix('docker.io/') registry_idx = reference.find('/') if registry_idx == -1 or (not any(c in reference[:registry_idx] for c in ('.', ':')) and reference[:registry_idx] != 'localhost'): registry, tagged_image = DEFAULT_DOCKER_REGISTRY, reference else: registry, tagged_image = reference[:registry_idx], reference[registry_idx + 1:] if '/' not in tagged_image: tagged_image = f'{DEFAULT_DOCKER_REPO}/{tagged_image}' # if image is not tagged, use default value. if ':' not in tagged_image: tagged_image += f':{DEFAULT_DOCKER_TAG}' # At this point, tag should be included already – we just need to see whether this # tag is named or digested and respond accordingly. ref_is_digest = False if '@' in tagged_image: matches = re.findall(DIGEST_RE, tagged_image) if not matches: raise CallError(f'Invalid reference format: {tagged_image}') tag = matches[-1] tag_pos = tagged_image.find(tag) image = tagged_image[:tag_pos - 1].rsplit(':', 1)[0] sep = '@' ref_is_digest = True elif ':' in tagged_image: image, tag = tagged_image.rsplit(':', 1) sep = ':' return { 'reference': reference, 'image': image, 'tag': tag, 'registry': registry, 'complete_tag': f'{registry}/{image}{sep}{tag}', 'reference_is_digest': ref_is_digest, } def get_chart_releases_consuming_image( image_names: list | set, chart_releases: list, get_mapping: bool = False ) -> dict | list: chart_releases_consuming_image = defaultdict(list) if get_mapping else set() images = {i['complete_tag']: i for i in map(normalize_reference, image_names)} for chart_release in chart_releases: for image in chart_release['resources']['container_images']: parsed_image = normalize_reference(image) if parsed_image['complete_tag'] in images and images[ parsed_image['complete_tag'] ]['tag'] == parsed_image['tag']: if get_mapping: chart_releases_consuming_image[chart_release['name']].append(parsed_image['reference']) else: chart_releases_consuming_image.add(chart_release['name']) return chart_releases_consuming_image if get_mapping else list(chart_releases_consuming_image) def parse_tags(references: list[str]) -> list[dict[str, str]]: return [normalize_reference(reference=reference) for reference in references] def normalize_docker_limits_header(headers: dict) -> dict: if not all(limit_key in headers for limit_key in ['ratelimit-limit', 'ratelimit-remaining']): return {'error': 'Unable to retrieve rate limit information from registry'} total_pull_limit, total_time_limit = headers['ratelimit-limit'].split(';w=') remaining_pull_limit, remaining_time_limit = headers['ratelimit-remaining'].split(';w=') return { 'total_pull_limit': int(total_pull_limit), 'total_time_limit_in_secs': int(total_time_limit), 'remaining_pull_limit': int(remaining_pull_limit), 'remaining_time_limit_in_secs': int(remaining_time_limit), 'error': None, }
6,049
Python
.py
126
41.166667
134
0.66039
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,072
dockerhub_ratelimit.py
truenas_middleware/src/middlewared/middlewared/plugins/apps_images/dockerhub_ratelimit.py
from middlewared.schema import Dict, Int, returns, Str from middlewared.service import accepts, Service from .client import ContainerRegistryClientMixin from .utils import normalize_docker_limits_header class ContainerImagesService(Service): class Config: namespace = 'app.image' @accepts(roles=['APPS_READ']) @returns(Dict( Int('total_pull_limit', null=True, description='Total pull limit for Docker Hub registry'), Int( 'total_time_limit_in_secs', null=True, description='Total time limit in seconds for Docker Hub registry before the limit renews' ), Int('remaining_pull_limit', null=True, description='Remaining pull limit for Docker Hub registry'), Int( 'remaining_time_limit_in_secs', null=True, description='Remaining time limit in seconds for Docker Hub registry for the ' 'current pull limit to be renewed' ), Str('error', null=True), )) async def dockerhub_rate_limit(self): """ Returns the current rate limit information for Docker Hub registry. Please refer to https://docs.docker.com/docker-hub/download-rate-limit/ for more information. """ limits_header = await ContainerRegistryClientMixin().get_docker_hub_rate_limit_preview() if limits_header.get('response_obj') and hasattr(limits_header['response_obj'], 'headers'): return normalize_docker_limits_header(limits_header['response_obj'].headers) return { 'error': 'Unable to retrieve rate limit information from registry', }
1,644
Python
.py
33
41.181818
107
0.676856
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,073
wait_on_disks.py
truenas_middleware/src/middlewared/middlewared/scripts/wait_on_disks.py
#!/usr/bin/python3 from collections import deque from time import time, sleep from sys import exit from pyudev import Context, Monitor, MonitorObserver DQ = deque(maxlen=2) def callback(dev): if uuid := dev.get('ID_PART_ENTRY_UUID'): DQ.append({'time': time(), 'name': dev.sys_name, 'uuid': uuid, 'action': dev.action}) def get_observer(): ctx = Context() mon = Monitor.from_netlink(ctx) mon.filter_by(subsystem='block') return MonitorObserver(mon, callback=callback) def main(max_wait=600.0, interval=5.0): """ `max_wait`: float representing the total time (in seconds) we should block and wait for disk events. `interval`: float representing the time we sleep between each iteration to allow new disk events to come in. """ obs = get_observer() obs.start() # start background thread last_event = dict() while max_wait > 0: max_wait -= round(interval, 2) sleep(interval) try: event = DQ[-1] except IndexError: # no events received break else: if event == last_event: # we've waited 5 seconds and no more events have come in break else: last_event = event obs.send_stop() # clean up background thread (non-blocking) if __name__ == '__main__': try: main() finally: exit(0) # always exit success (for now)
1,473
Python
.py
44
26.431818
93
0.616254
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,074
setup_cgroups.py
truenas_middleware/src/middlewared/middlewared/scripts/setup_cgroups.py
#!/usr/bin/python3 import contextlib import os CGROUP_ROOT_PATH = '/sys/fs/cgroup' CGROUP_AVAILABLE_CONTROLLERS_PATH = os.path.join(CGROUP_ROOT_PATH, 'cgroup.subtree_control') def get_available_controllers_for_consumption() -> set: try: with open(CGROUP_AVAILABLE_CONTROLLERS_PATH, 'r') as f: return set(f.read().split()) except FileNotFoundError: raise Exception( 'Unable to determine cgroup controllers which are available for consumption as ' f'{CGROUP_AVAILABLE_CONTROLLERS_PATH!r} does not exist' ) def update_available_controllers_for_consumption(to_add_controllers: set) -> set: # This will try to update available controllers for consumption and return the current state # regardless of the update failing with contextlib.suppress(FileNotFoundError, OSError): with open(CGROUP_AVAILABLE_CONTROLLERS_PATH, 'w') as f: f.write(f'{" ".join(map(lambda s: f"+{s}", to_add_controllers))}') return get_available_controllers_for_consumption() def main(): # Logic copied over from kubernetes # https://github.com/kubernetes/kubernetes/blob/08fbe92fa76d35048b4b4891b41fc6912e689cc7/ # pkg/kubelet/cm/cgroup_manager_linux.go#L238 # FIXME: See if this is now required for docker supported_controllers = {'cpu', 'cpuset', 'memory', 'hugetlb', 'pids'} system_supported_controllers_path = os.path.join(CGROUP_ROOT_PATH, 'cgroup.controllers') try: with open(system_supported_controllers_path, 'r') as f: available_controllers = set(f.read().split()) except FileNotFoundError: raise Exception( 'Unable to determine available cgroup controllers as ' f'{system_supported_controllers_path!r} does not exist' ) # What we are doing here is that we get controllers which are supported and required by k8s to function # then we check if they are available for consumption, if not we try to add them to subtree control # and then do a final check to confirm they have been added as desired needed_controllers = supported_controllers & available_controllers available_controllers_for_consumption = get_available_controllers_for_consumption() if missing_controllers := needed_controllers - available_controllers_for_consumption: # If we have missing controllers, lets try adding them to subtree control available_controllers_for_consumption = update_available_controllers_for_consumption(missing_controllers) missing_controllers = needed_controllers - available_controllers_for_consumption if missing_controllers: raise Exception( f'Missing {", ".join(missing_controllers)!r} cgroup controller(s) ' 'which are required for apps to function' ) if __name__ == '__main__': main()
2,859
Python
.py
52
48.173077
113
0.716792
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,075
sedhelper.py
truenas_middleware/src/middlewared/middlewared/scripts/sedhelper.py
#!/usr/bin/env python import argparse from concurrent.futures import ThreadPoolExecutor import errno from truenas_api_client import Client, ClientException # A particular Kioxia PM6 drive has shown us that it takes longer than # 60 seconds to unlock/setup SED. Specifically, the "sedutil-cli --initialSetup" # command took ~135 seconds to complete. Note the other PM6 drives took < 30 secs. # (60 secs is default websocket timeout so bump this to 300 secs (5mins) to be safe) TIMEOUT = 300 def setup(password, disk=None): def sed_setup(client, disk_name, password): rv = client.call('disk.sed_initial_setup', disk_name, password) if rv == 'SUCCESS': print(f'{disk_name}\t\t[\033[92mOK\x1B[0m]') elif rv == 'SETUP_FAILED': print(f'{disk_name}\t\t[\033[91mSETUP FAILED\x1B[0m]') elif rv == 'LOCKING_DISABLED': print(f'{disk_name}\t\t[\033[91mLOCKING DISABLED\x1B[0m]') elif rv in ('ACCESS_GRANTED', 'NO_SED'): pass return disk_name, rv with Client(call_timeout=TIMEOUT) as c: disk_filter = [] if disk: disk_filter.append(('name', '=', disk)) disks = c.call('disk.query', disk_filter, {'extra': {'passwords': True}}) boot_disks = c.call('boot.get_disks') disks = list(filter(lambda d: d['name'] not in boot_disks, disks)) if not disks: print(f'Disk {disk} not found') return global_sed_password = c.call('system.advanced.sed_global_password') if global_sed_password != password and not (disk and disks[0]['passwd'] == password): print('Given password does not match saved one') return action = False no_sed = False granted = False with ThreadPoolExecutor(max_workers=12) as e: for disk_name, rv in e.map(lambda disk: sed_setup(c, disk['name'], password), disks): if rv == 'NO_SED': no_sed = True if rv in ('SUCCESS', 'SETUP_FAILED', 'LOCKING_DISABLED'): action = True if rv == 'ACCESS_GRANTED': granted = True if not action: if no_sed and not granted: print('No SED disks were found in the system') else: print('No new SED disks detected') def unlock(): with Client(call_timeout=TIMEOUT) as c: try: c.call('disk.sed_unlock_all') except ClientException as e: if e.errno == errno.EACCES: print('SED disks failed to unlocked') else: raise else: print('All SED disks unlocked') def main(): parser = argparse.ArgumentParser() subparsers = parser.add_subparsers(help='sub-command help', dest='action') parser_setup = subparsers.add_parser('setup', help='Setup new SED disks') parser_setup.add_argument('--disk', help='Perform action only on specified disk') parser_setup.add_argument('password', help='Password to use on new disks') subparsers.add_parser('unlock', help='Unlock SED disks') args = parser.parse_args() if args.action == 'setup': setup(args.password, disk=args.disk) elif args.action == 'unlock': unlock() else: parser.print_help() if __name__ == '__main__': main()
3,412
Python
.py
79
34.075949
97
0.605072
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,076
vendor_service.py
truenas_middleware/src/middlewared/middlewared/scripts/vendor_service.py
import hashlib import subprocess import sys from middlewared.plugins.system_vendor.vendor import get_vendor from middlewared.utils.vendor import Vendors def load_envvars(envvars_file: str): try: envvars = [] with open(envvars_file, "r") as f: for line in f: if (line := line.strip()) and line[0] != "#": envvars.append(line.split("=", 1)) return dict(envvars) except (OSError, ValueError): return dict() def get_hostid() -> str | None: try: with open("/etc/hostid", "rb") as f: return hashlib.file_digest(f, "sha256").hexdigest() except Exception: pass def start_hexos_websocat(): url = "wss://api.hexos.com" envvars_file = "/etc/default/websocat" envvars = load_envvars(envvars_file) systemd_opts = ( "--unit=websocat", "--description=websocat daemon for HexOS", "--property=Restart=always", "--property=RestartSec=10", "--uid=www-data", f"--setenv=URL={url}", *[f"--setenv={name}={value}" for name, value in envvars.items()] ) wsocat_path = "/usr/local/libexec/wsocat" wsocat_opts = ( "--buffer-size 1048576", "--ping-interval 30", "--ping-timeout 60", "--exit-on-eof", "--text", ) local_server = "ws://127.0.0.1:6000/websocket" hostid_hash = get_hostid() if hostid_hash is None: return ip_output = subprocess.check_output("ip -o -4 route get 8.8.8.8", shell=True, text=True) ip_address = ip_output.partition("src")[-1].split()[0] remote_server = f"{url}/server/{hostid_hash}/{ip_address}" # Start a transient service subprocess.run([ "systemd-run", *systemd_opts, "/bin/bash", "-c", " ".join([wsocat_path, *wsocat_opts, local_server, remote_server]) ]) def main(): if get_vendor() == Vendors.HEXOS: start_hexos_websocat() if __name__ == "__main__": try: main() finally: sys.exit(0) # Never fail
2,088
Python
.py
65
25.138462
92
0.583749
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,077
configure_fips.py
truenas_middleware/src/middlewared/middlewared/scripts/configure_fips.py
#!/usr/bin/python3 import os import shutil import sqlite3 import subprocess from middlewared.utils.db import query_config_table from middlewared.utils.rootfs import ReadonlyRootfsManager FIPS_MODULE_FILE = '/usr/lib/ssl/fipsmodule.cnf' OPENSSL_CONFIG_FILE = '/etc/ssl/openssl.cnf' BASE_OPENSSL_CONFIG_FILE = '/conf/base/etc/ssl/openssl.cnf' OPENSSL_FIPS_FILE = '/etc/ssl/openssl_fips.cnf' def validate_system_state() -> None: for path in (FIPS_MODULE_FILE, OPENSSL_CONFIG_FILE, OPENSSL_FIPS_FILE, BASE_OPENSSL_CONFIG_FILE): if not os.path.exists(path): raise Exception(f'{path!r} does not exist') def modify_openssl_config(enable_fips: bool) -> None: shutil.copyfile(BASE_OPENSSL_CONFIG_FILE, OPENSSL_CONFIG_FILE) if enable_fips: with open(OPENSSL_CONFIG_FILE, 'a') as f: f.write(f'\n.include {OPENSSL_FIPS_FILE}\n') def configure_fips(enable_fips: bool) -> None: if enable_fips: subprocess.check_call([ 'openssl', 'fipsinstall', '-out', FIPS_MODULE_FILE, '-module', '/usr/lib/x86_64-linux-gnu/ossl-modules/fips.so', ], timeout=30, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL) modify_openssl_config(enable_fips) def main() -> None: validate_system_state() try: security_settings = query_config_table('system_security') except (sqlite3.OperationalError, IndexError): # This is for the case when users are upgrading and in that case table will not exist # so we should always disable fips as a default because users might not be able to ssh # into the system security_settings = {'enable_fips': False} with ReadonlyRootfsManager('/') as readonly_rootfs: readonly_rootfs.make_writeable() configure_fips(security_settings['enable_fips']) if __name__ == '__main__': main()
1,869
Python
.py
41
39.902439
101
0.701435
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,078
wait_to_hang_and_dump_core.py
truenas_middleware/src/middlewared/middlewared/scripts/wait_to_hang_and_dump_core.py
# -*- coding=utf-8 -*- import contextlib import logging import os import re import socket import subprocess import time from truenas_api_client import Client logger = logging.getLogger(__name__) """The point of this script is so that the development team can ask a user to run this script from the command line during the unfortunate event the main middleware process "hangs". (We've had past situations where asyncio main event loop hung up). The idea is to at least allow user to get us some information so we can try and deduce what might be going on.""" def main(): logging.basicConfig(level=logging.INFO, format="[%(asctime)s] %(message)s") interval = 10 logger.info("Probing middleware accessibility with %d seconds interval", interval) while True: try: with Client(): pass except socket.timeout: logger.info("Caught timeout, dumping core") dump_core() break else: time.sleep(interval) def dump_core(): middlewared_pid = int( re.match( r"MainPID=([0-9]+)", subprocess.check_output( "systemctl show --property MainPID middlewared".split(), encoding="utf-8", ), ).group(1), ) logger.info("middlewared PID: %d", middlewared_pid) core_file = f"core.{middlewared_pid}" with contextlib.suppress(FileNotFoundError): os.unlink(core_file) subprocess.run( ["gdb", "-p", str(middlewared_pid), "-batch", "-ex", "generate-core-file"], check=True, ) logger.info("Compressing core file %r", core_file) subprocess.run(["gzip", core_file], check=True) logger.info("%r is ready!", f"{core_file}.gz") if __name__ == "__main__": main()
1,802
Python
.py
52
28.269231
86
0.642693
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,079
gdb.py
truenas_middleware/src/middlewared/middlewared/scripts/gdb.py
import argparse import logging import requests import subprocess import sys import tempfile class MiddlewareGDB(object): def __init__(self): self.logger = logging.getLogger('middlewaregdb') def install_dbg(self): packages = ['python3-dbg', 'python3-dev'] for p in packages: try: subprocess.run(['dpkg', '-L', p], capture_output=True, check=True) except subprocess.CalledProcessError: subprocess.run(['apt', 'install', '-y', p], check=True) def run_gdb(self): proc = subprocess.Popen( ['pgrep', '-o', '-f', 'middlewared'], stdout=subprocess.PIPE, stderr=subprocess.STDOUT ) try: middlewared_pid = int(proc.communicate()[0].split()[0].strip()) except Exception: print('Failed to find middlewared process', file=sys.stderr) sys.exit(1) proc = subprocess.Popen([ 'gdb', '-p', str(middlewared_pid), '-batch', '-ex', 'thread apply all py-list', '-ex', 'thread apply all py-bt', ], stdout=subprocess.PIPE, stderr=subprocess.STDOUT) output = proc.communicate()[0] outfile = tempfile.NamedTemporaryFile(delete=False) outfile.write(output) rv = {'local': outfile.name} try: r = requests.post('http://ix.io', { 'f:1': output, }) rv['remote'] = r.text.strip() except Exception: pass return rv def main(self): parser = argparse.ArgumentParser() parser.parse_args() print('Making sure debug packages are installed') self.install_dbg() print('Running gdb') rv = self.run_gdb() if 'local' in rv: print(f'Local output: {rv["local"]}') if 'remote' in rv: print(f'Remote output: {rv["remote"]}') def main(): MiddlewareGDB().main() if __name__ == '__main__': main()
2,042
Python
.py
60
24.766667
82
0.555951
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,080
base.py
truenas_middleware/src/middlewared/middlewared/rclone/base.py
class BaseRcloneRemote: name = NotImplemented title = NotImplemented buckets = False bucket_title = "Bucket" can_create_bucket = False custom_list_buckets = False readonly = False fast_list = False rclone_type = NotImplemented credentials_schema = NotImplemented credentials_oauth = False credentials_oauth_name = None refresh_credentials = [] task_schema = [] extra_methods = [] restic = False def __init__(self, middleware): self.middleware = middleware async def create_bucket(self, credentials, name): raise NotImplementedError async def list_buckets(self, credentials): raise NotImplementedError async def validate_task_basic(self, task, credentials, verrors): pass async def validate_task_full(self, task, credentials, verrors): pass async def get_credentials_extra(self, credentials): return {} async def get_task_extra(self, task): return {} async def get_task_extra_args(self, task): return [] async def cleanup(self, task, config): pass def get_restic_config(self, task): raise NotImplementedError
1,210
Python
.py
37
26.243243
68
0.678788
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,081
sftp.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/sftp.py
import os import tempfile from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Int, Str class SFTPRcloneRemote(BaseRcloneRemote): name = "SFTP" title = "SFTP" rclone_type = "sftp" credentials_schema = [ Str("host", title="Host", required=True), Int("port", title="Port"), Str("user", title="Username", required=True), Str("pass", title="Password"), Int("private_key", title="Private Key ID"), ] async def get_credentials_extra(self, credentials): result = {} if "private_key" in credentials["attributes"]: with tempfile.NamedTemporaryFile(mode="w+", delete=False) as tmp_file: tmp_file.write((await self.middleware.call("keychaincredential.get_of_type", credentials["attributes"]["private_key"], "SSH_KEY_PAIR"))["attributes"]["private_key"]) result["key_file"] = tmp_file.name return result async def cleanup(self, task, config): if "private_key" in task["credentials"]["attributes"]: os.unlink(config["key_file"])
1,231
Python
.py
27
34.074074
105
0.585427
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,082
hubic.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/hubic.py
from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Str class HubicRcloneRemote(BaseRcloneRemote): name = "HUBIC" title = "Hubic" buckets = True bucket_title = "Container" fast_list = True rclone_type = "hubic" credentials_schema = [ Str("token", title="Access Token", required=True, max_length=None), ]
382
Python
.py
12
27
75
0.708791
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,083
azureblob.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/azureblob.py
import re from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Str from middlewared.validators import Match, URL class AzureBlobRcloneRemote(BaseRcloneRemote): name = "AZUREBLOB" title = "Microsoft Azure Blob Storage" buckets = True bucket_title = "Container" fast_list = True rclone_type = "azureblob" credentials_schema = [ Str("account", title="Account Name", required=True, validators=[ Match(r"^[a-z0-9\-.]+$", re.IGNORECASE, "Account Name field can only contain alphanumeric characters, - and .") ]), Str("key", title="Account Key", required=True), Str("endpoint", title="Endpoint", default="", validators=[URL(empty=True)]), ] async def get_task_extra(self, task): return {"chunk_size": "100Mi"}
848
Python
.py
21
33.952381
89
0.667888
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,084
google_drive.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/google_drive.py
import textwrap from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Bool, Password, Str class GoogleDriveRcloneRemote(BaseRcloneRemote): name = "GOOGLE_DRIVE" title = "Google Drive" fast_list = True rclone_type = "drive" credentials_schema = [ Str("client_id", title="OAuth Client ID", default=""), Password("client_secret", title="OAuth Client Secret", default=""), Password("token", title="Access Token", required=True, max_length=None), Str("team_drive", title="Team Drive ID (if connecting to Team Drive)"), ] credentials_oauth = True refresh_credentials = ["token"] task_schema = [ Bool("acknowledge_abuse", title="Allow files which return cannotDownloadAbusiveFile to be downloaded.", description=textwrap.dedent("""\ If downloading a file returns the error "This file has been identified as malware or spam and cannot be downloaded" with the error code "cannotDownloadAbusiveFile" then enable this flag to indicate you acknowledge the risks of downloading the file and TrueNAS will download it anyway. """), default=False), ] async def get_credentials_extra(self, credentials): if credentials["attributes"].get("team_drive"): return dict() return dict(root_folder_id="root")
1,417
Python
.py
29
40.862069
119
0.678028
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,085
googlephotos.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/googlephotos.py
import os from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Str class GooglePhotosRcloneRemote(BaseRcloneRemote): name = "GOOGLE_PHOTOS" title = "Google Photos" rclone_type = "googlephotos" credentials_schema = [ Str("client_id", title="OAuth Client ID", default=""), Str("client_secret", title="OAuth Client Secret", default=""), Str("token", title="Access Token", required=True, max_length=None), ] refresh_credentials = ["token"] async def validate_task_full(self, task, credentials, verrors): # `/media/by-day` contains a huge tree of empty directories for all days starting from 2000-01-01. Listing # them all will never complete due to the API rate limits. folder = task["attributes"]["folder"].strip("/") if not folder: verrors.add( "attributes.folder", "Pulling from the root directory is not allowed. Please, select a specific directory." ) return folder = os.path.normpath(folder) for prohibited in ["media", "media/by-day"]: if folder == prohibited: verrors.add( "attributes.folder", f"Pulling from the {prohibited} directory is not allowed. Please, select a specific directory." ) return
1,410
Python
.py
31
35.580645
115
0.624362
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,086
mega.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/mega.py
from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Str class MegaRcloneRemote(BaseRcloneRemote): name = "MEGA" title = "Mega" rclone_type = "mega" credentials_schema = [ Str("user", title="Username", required=True), Str("pass", title="Password", required=True), ]
337
Python
.py
10
28.7
53
0.69969
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,087
s3.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/s3.py
import textwrap import boto3 from botocore.client import Config from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Bool, Int, Password, Str from middlewared.utils.lang import undefined class S3RcloneRemote(BaseRcloneRemote): name = "S3" title = "Amazon S3" buckets = True fast_list = True rclone_type = "s3" credentials_schema = [ Str("access_key_id", title="Access Key ID", required=True), Password("secret_access_key", title="Secret Access Key", required=True), Str("endpoint", title="Endpoint URL", default=""), Str("region", title="Region", default=""), Bool("skip_region", title="Endpoint does not support regions", default=False), Bool("signatures_v2", title="Use v2 signatures", default=False), Int("max_upload_parts", title="Maximum number of parts in a multipart upload", description=textwrap.dedent("""\ This option defines the maximum number of multipart chunks to use when doing a multipart upload. This can be useful if a service does not support the AWS S3 specification of 10,000 chunks (e.g. Scaleway). """), default=10000), ] task_schema = [ Str("region", title="Region", default=""), Str("encryption", title="Server-Side Encryption", enum=[None, "AES256"], default=None, null=True), Str("storage_class", title="The storage class to use", enum=["", "STANDARD", "REDUCED_REDUNDANCY", "STANDARD_IA", "ONEZONE_IA", "INTELLIGENT_TIERING", "GLACIER", "GLACIER_IR", "DEEP_ARCHIVE"]), ] def _get_client(self, credentials): config = None if credentials["attributes"].get("signatures_v2", False): config = Config(signature_version="s3") client = boto3.client( "s3", config=config, endpoint_url=credentials["attributes"].get("endpoint", "").strip() or None, region_name=credentials["attributes"].get("region", "").strip() or None, aws_access_key_id=credentials["attributes"]["access_key_id"], aws_secret_access_key=credentials["attributes"]["secret_access_key"], ) return client async def validate_task_basic(self, task, credentials, verrors): if task["attributes"]["encryption"] not in (None, "", "AES256"): verrors.add("encryption", 'Encryption should be null or "AES256"') if not credentials["attributes"].get("skip_region", False): if not credentials["attributes"].get("region", "").strip(): response = await self.middleware.run_in_thread( self._get_client(credentials).get_bucket_location, Bucket=task["attributes"]["bucket"] ) task["attributes"]["region"] = response["LocationConstraint"] or "us-east-1" async def get_credentials_extra(self, credentials): result = {} if (credentials["attributes"].get("endpoint") or "").rstrip("/").endswith(".scw.cloud"): if credentials["attributes"].get("max_upload_parts", 10000) == 10000: result["max_upload_parts"] = 1000 return result async def get_task_extra(self, task): result = dict( encryption=undefined, server_side_encryption=task["attributes"].get("encryption") or "", skip_region=undefined, signatures_v2=undefined, provider="Other", ) if not task["credentials"]["attributes"].get("skip_region", False): if task["credentials"]["attributes"].get("region", "").strip(): if not (task["attributes"].get("region") or "").strip(): result["region"] = task["credentials"]["attributes"]["region"] else: # Some legacy tasks have region=None, it's easier to fix it here than in migration result["region"] = task["attributes"].get("region") or "us-east-1" else: if task["credentials"]["attributes"].get("signatures_v2", False): result["region"] = "other-v2-signature" else: result["region"] = "" return result def get_restic_config(self, task): url = task["credentials"]["attributes"].get("endpoint", "").rstrip("/") if not url: url = "s3.amazonaws.com" env = { "AWS_ACCESS_KEY_ID": task["credentials"]["attributes"]["access_key_id"], "AWS_SECRET_ACCESS_KEY": task["credentials"]["attributes"]["secret_access_key"], } return url, env
4,754
Python
.py
89
41.88764
120
0.594356
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,088
yandex.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/yandex.py
from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Password, Str class YandexRcloneRemote(BaseRcloneRemote): name = "YANDEX" title = "Yandex" fast_list = True rclone_type = "yandex" credentials_schema = [ Str("client_id", title="OAuth Client ID", default=""), Password("client_secret", title="OAuth Client Secret", default=""), Str("token", title="Access Token", required=True, max_length=None), ] credentials_oauth = True refresh_credentials = ["token"]
549
Python
.py
14
33.857143
75
0.692453
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,089
box.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/box.py
from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Password, Str class BoxRcloneRemote(BaseRcloneRemote): name = "BOX" title = "Box" rclone_type = "box" credentials_schema = [ Str("client_id", title="OAuth Client ID", default=""), Password("client_secret", title="OAuth Client Secret", default=""), Password("token", title="Access Token", required=True, max_length=None), ] credentials_oauth = True refresh_credentials = ["token"]
520
Python
.py
13
34.692308
80
0.691849
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,090
ftp.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/ftp.py
from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Int, Str class FTPRcloneRemote(BaseRcloneRemote): name = "FTP" title = "FTP" rclone_type = "ftp" credentials_schema = [ Str("host", title="Host", required=True), Int("port", title="Port"), Str("user", title="Username", required=True), Str("pass", title="Password", required=True), ]
423
Python
.py
12
29.583333
53
0.660934
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,091
dropbox.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/dropbox.py
import textwrap from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Int, Password, Str from middlewared.validators import Range class DropboxRcloneRemote(BaseRcloneRemote): name = "DROPBOX" title = "Dropbox" rclone_type = "dropbox" credentials_schema = [ Str("client_id", title="OAuth Client ID", default=""), Password("client_secret", title="OAuth Client Secret", default=""), Password("token", title="Access Token", required=True, max_length=None), ] credentials_oauth = True task_schema = [ Int("chunk_size", title="Upload chunk size (in megabytes)", description=textwrap.dedent("""\ Upload chunk size. Must fit in memory. Note that these chunks are buffered in memory and there might be a maximum of «--transfers» chunks in progress at once. Dropbox Business accounts can have monthly data transfer limits per team per month. By using larger chnuk sizes you will decrease the number of data transfer calls used and you'll be able to transfer more data to your Dropbox Business account. """), default=48, validators=[Range(min_=5, max_=149)]), ] async def get_task_extra(self, task): return {"chunk_size": str(task["attributes"].get("chunk_size", 48)) + "M"}
1,335
Python
.py
24
48.75
117
0.695084
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,092
swift.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/swift.py
from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Password, Str, Int class OpenStackSwiftRcloneRemote(BaseRcloneRemote): name = "OPENSTACK_SWIFT" title = "OpenStack Swift" buckets = True bucket_title = "Container" fast_list = True rclone_type = "swift" credentials_schema = [ Str("user", required=True, title="User name (OS_USERNAME)"), Password("key", required=True, title="API key or password (OS_PASSWORD)"), Str("auth", required=True, title="Authentication URL for server (OS_AUTH_URL)"), Str("user_id", default="", title="User ID to log in - most swift systems use user and leave " "this blank (v3 auth) (OS_USER_ID)"), Str("domain", default="", title="User domain - optional (v3 auth) (OS_USER_DOMAIN_NAME)"), Str("tenant", default="", title="Tenant name - optional for v1 auth, this or tenant_id " "required otherwise (OS_TENANT_NAME or OS_PROJECT_NAME)"), Str("tenant_id", default="", title="Tenant ID - optional for v1 auth, this or tenant required " "otherwise (OS_TENANT_ID)"), Str("tenant_domain", default="", title="Tenant domain - optional (v3 auth) " "(OS_PROJECT_DOMAIN_NAME)"), Str("region", default="", title="Region name (OS_REGION_NAME)"), Str("storage_url", default="", title="Storage URL (OS_STORAGE_URL)"), Password("auth_token", default="", title="Auth Token from alternate authentication (OS_AUTH_TOKEN)"), Str("application_credential_id", default="", title="Application Credential ID (OS_APPLICATION_CREDENTIAL_ID)"), Str("application_credential_name", default="", title="Application Credential Name " "(OS_APPLICATION_CREDENTIAL_NAME)"), Password("application_credential_secret", default="", title="Application Credential Secret " "(OS_APPLICATION_CREDENTIAL_SECRET)"), Int("auth_version", enum=[0, 1, 2, 3], title="AuthVersion - set it if your auth URL has no version " "(ST_AUTH_VERSION)"), Str("endpoint_type", enum=["public", "internal", "admin"], title="Endpoint type to choose from the service catalogue " "(OS_ENDPOINT_TYPE)"), ]
2,500
Python
.py
51
38.176471
78
0.591077
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,093
b2.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/b2.py
import textwrap from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Int, Str from middlewared.validators import Range class B2RcloneRemote(BaseRcloneRemote): name = "B2" title = "Backblaze B2" buckets = True fast_list = True rclone_type = "b2" credentials_schema = [ Str("account", title="Account ID or Application Key ID", description=textwrap.dedent("""\ Put your Account ID here and use your Master Application Key as Application Key or create separate Application Key and use it with its own Application Key ID """), required=True), Str("key", title="Application Key", required=True), ] task_schema = [ Int("chunk_size", title="Upload chunk size (in megabytes)", description=textwrap.dedent("""\ Upload chunk size. Must fit in memory. Note that these chunks are buffered in memory and there might be a maximum of «--transfers» chunks in progress at once. Also, your largest file must be split in no more than 10 000 chunks. """), default=96, validators=[Range(min_=5)]), ] async def get_task_extra(self, task): chunk_size = task["attributes"].get("chunk_size", 96) extra = {"chunk_size": f"{chunk_size}M"} if chunk_size > 200: extra["upload_cutoff"] = f"{chunk_size}M" return extra async def get_task_extra_args(self, task): chunk_size = task["attributes"].get("chunk_size", 96) if chunk_size > 128: return [f"--multi-thread-cutoff={chunk_size * 2 + 1}M"] return []
1,639
Python
.py
35
39.057143
117
0.651163
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,094
google_cloud_storage.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/google_cloud_storage.py
import json import textwrap from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Bool, Str class GoogleCloudStorageRcloneRemote(BaseRcloneRemote): name = "GOOGLE_CLOUD_STORAGE" title = "Google Cloud Storage" buckets = True fast_list = True rclone_type = "google cloud storage" credentials_schema = [ Str("service_account_credentials", title="Service Account", required=True, max_length=None), ] task_schema = [ Bool("bucket_policy_only", title="Bucket Policy Only", description=textwrap.dedent("""\ Access checks should use bucket-level IAM policies. If you want to upload objects to a bucket with Bucket Policy Only set then you will need to set this. """), default=False), ] async def get_credentials_extra(self, credentials): return dict( service_account_credentials=(credentials["attributes"]["service_account_credentials"]. replace("\r", ""). replace("\n", "")), project_number=json.loads(credentials["attributes"]["service_account_credentials"])["project_id"], )
1,219
Python
.py
26
37.461538
113
0.648649
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,095
webdav.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/webdav.py
from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Str class WebDavRcloneRemote(BaseRcloneRemote): name = "WEBDAV" title = "WebDAV" rclone_type = "webdav" credentials_schema = [ Str("url", title="URL", required=True), Str("vendor", title="Name of the WebDAV site/service/software", enum=["NEXTCLOUD", "OWNCLOUD", "SHAREPOINT", "OTHER"], required=True), Str("user", title="Username", required=True), Str("pass", title="Password", required=True), ] async def get_task_extra(self, task): return dict(vendor=task["credentials"]["attributes"]["vendor"].lower())
671
Python
.py
15
38.333333
82
0.669739
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,096
storjix.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/storjix.py
import errno import io import xml.etree.ElementTree as ET from aws_requests_auth.aws_auth import AWSRequestsAuth import boto3 import botocore import requests from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Password, Str from middlewared.service_exception import CallError from middlewared.utils.network import INTERNET_TIMEOUT class StorjIxError(CallError): pass class StorjIxRcloneRemote(BaseRcloneRemote): name = "STORJ_IX" title = "Storj iX" buckets = True can_create_bucket = True custom_list_buckets = True fast_list = True rclone_type = "s3" credentials_schema = [ Str("access_key_id", title="Access Key ID", required=True), Password("secret_access_key", title="Secret Access Key", required=True), ] task_schema = [] async def create_bucket(self, credentials, name): def create_bucket_sync(): s3_client = boto3.client( "s3", config=botocore.config.Config(user_agent="ix-storj-1"), endpoint_url="https://gateway.storjshare.io", aws_access_key_id=credentials["attributes"]["access_key_id"], aws_secret_access_key=credentials["attributes"]["secret_access_key"], ) # s3 bucket naming rules: https://docs.aws.amazon.com/AmazonS3/latest/userguide/bucketnamingrules.html try: s3_client.create_bucket(Bucket=name) except s3_client.exceptions.BucketAlreadyExists as e: raise CallError(str(e), errno=errno.EEXIST) except botocore.exceptions.ParamValidationError as e: raise StorjIxError("The bucket name can only contain lowercase letters, numbers, and hyphens.", errno.EINVAL, str(e)) except botocore.exceptions.ClientError as e: if "InvalidBucketName" in e.args[0]: raise StorjIxError("The bucket name must be between 3-63 characters in length and cannot contain " "uppercase.", errno.EINVAL, str(e)) raise return await self.middleware.run_in_thread(create_bucket_sync) async def list_buckets(self, credentials): def list_buckets_sync(): auth = AWSRequestsAuth(aws_access_key=credentials["attributes"]["access_key_id"], aws_secret_access_key=credentials["attributes"]["secret_access_key"], aws_host="gateway.storjshare.io", aws_region="", aws_service="s3") r = requests.get("https://gateway.storjshare.io/?attribution", auth=auth, timeout=INTERNET_TIMEOUT) r.raise_for_status() ns = "{http://s3.amazonaws.com/doc/2006-03-01/}" return [ { "name": bucket.find(f"{ns}Name").text, "time": bucket.find(f"{ns}CreationDate").text, "enabled": "ix-storj-1" in (bucket.find(f"{ns}Attribution").text or ""), } for bucket in ET.parse(io.StringIO(r.text)).iter(f"{ns}Bucket") ] return await self.middleware.run_in_thread(list_buckets_sync) async def get_credentials_extra(self, credentials): return {"endpoint": "https://gateway.storjshare.io"} async def get_task_extra(self, task): # Storj recommended these settings return { "chunk_size": "64M", "disable_http2": "true", "upload_cutoff": "64M", "provider": "Other", } def get_restic_config(self, task): url = "gateway.storjshare.io" env = { "AWS_ACCESS_KEY_ID": task["credentials"]["attributes"]["access_key_id"], "AWS_SECRET_ACCESS_KEY": task["credentials"]["attributes"]["secret_access_key"], } return url, env
4,014
Python
.py
85
35.305882
118
0.599181
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,097
http.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/http.py
from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Str class HTTPRcloneRemote(BaseRcloneRemote): name = "HTTP" title = "HTTP" readonly = True rclone_type = "http" credentials_schema = [ Str("url", title="URL", required=True), ]
298
Python
.py
10
25.1
52
0.706714
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,098
pcloud.py
truenas_middleware/src/middlewared/middlewared/rclone/remote/pcloud.py
from middlewared.rclone.base import BaseRcloneRemote from middlewared.schema import Password, Str class PcloudRcloneRemote(BaseRcloneRemote): name = "PCLOUD" title = "pCloud" rclone_type = "pcloud" credentials_schema = [ Str("client_id", title="OAuth Client ID", default=""), Password("client_secret", title="OAuth Client Secret", default=""), Password("token", title="Access Token", required=True, max_length=None), Str("hostname", title="API hostname"), ] credentials_oauth = True credentials_oauth_name = "pcloud2" refresh_credentials = ["token"]
618
Python
.py
15
35.666667
80
0.691152
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)
25,099
conf.py
truenas_middleware/docs/source/conf.py
# -*- coding: utf-8 -*- # # Configuration file for the Sphinx documentation builder. # # This file does only contain a selection of the most common options. For a # full list see the documentation: # http://www.sphinx-doc.org/en/master/config # -- Path setup -------------------------------------------------------------- # If extensions (or modules to document with autodoc) are in another directory, # add these directories to sys.path here. If the directory is relative to the # documentation root, use os.path.abspath to make it absolute, like shown here. # import os import sys sys.path.insert(0, os.path.abspath('../../src/middlewared')) # -- Project information ----------------------------------------------------- project = 'TrueNAS' copyright = '2021, iXsystems' author = 'iXsystems' # The short X.Y version version = '' # The full version, including alpha/beta/rc tags release = '22.02' # -- General configuration --------------------------------------------------- # If your documentation needs a minimal Sphinx version, state it here. # # needs_sphinx = '1.0' # Add any Sphinx extension module names here, as strings. They can be # extensions coming with Sphinx (named 'sphinx.ext.*') or your custom # ones. extensions = [ 'sphinx.ext.autodoc', 'sphinx.ext.autosectionlabel', 'sphinx.ext.viewcode', 'sphinx-prompt', 'sphinx_substitution_extensions', ] # Add any paths that contain templates here, relative to this directory. templates_path = ['_templates'] # The suffix(es) of source filenames. # You can specify multiple suffix as a list of string: # # source_suffix = ['.rst', '.md'] source_suffix = '.rst' # The master toctree document. master_doc = 'index' # The language for content autogenerated by Sphinx. Refer to documentation # for a list of supported languages. # # This is also used if you do content translation via gettext catalogs. # Usually you set "language" from the command line for these cases. language = 'en' # List of patterns, relative to source directory, that match files and # directories to ignore when looking for source files. # This pattern also affects html_static_path and html_extra_path. exclude_patterns = [] # The name of the Pygments (syntax highlighting) style to use. pygments_style = None rst_prolog = f'.. |version| replace:: {release}\n' default_role = 'code' # -- Options for HTML output ------------------------------------------------- # The theme to use for HTML and HTML Help pages. See the documentation for # a list of builtin themes. # html_theme = 'sphinx_rtd_theme' # Theme options are theme-specific and customize the look and feel of a theme # further. For a list of options available for each theme, see the # documentation. # # html_theme_options = {} # Add any paths that contain custom static files (such as style sheets) here, # relative to this directory. They are copied after the builtin static files, # so a file named "default.css" will overwrite the builtin "default.css". html_static_path = ['_static'] # Custom sidebar templates, must be a dictionary that maps document names # to template names. # # The default sidebars (for documents that don't match any pattern) are # defined by theme itself. Builtin themes are using these templates by # default: ``['localtoc.html', 'relations.html', 'sourcelink.html', # 'searchbox.html']``. # # html_sidebars = {} # -- Options for HTMLHelp output --------------------------------------------- # Output file base name for HTML help builder. htmlhelp_basename = 'TrueNASdoc' # -- Options for LaTeX output ------------------------------------------------ latex_elements = { # The paper size ('letterpaper' or 'a4paper'). # # 'papersize': 'letterpaper', # The font size ('10pt', '11pt' or '12pt'). # # 'pointsize': '10pt', # Additional stuff for the LaTeX preamble. # # 'preamble': '', # Latex figure (float) alignment # # 'figure_align': 'htbp', } # Grouping the document tree into LaTeX files. List of tuples # (source start file, target name, title, # author, documentclass [howto, manual, or own class]). latex_documents = [ (master_doc, 'TrueNAS.tex', 'TrueNAS Documentation', 'iXsystems', 'manual'), ] # -- Options for manual page output ------------------------------------------ # One entry per manual page. List of tuples # (source start file, name, description, authors, manual section). man_pages = [ (master_doc, 'truenas', 'TrueNAS Documentation', [author], 1) ] # -- Options for Texinfo output ---------------------------------------------- # Grouping the document tree into Texinfo files. List of tuples # (source start file, target name, title, author, # dir menu entry, description, category) texinfo_documents = [ (master_doc, 'TrueNAS', 'TrueNAS Documentation', author, 'TrueNAS', 'One line description of project.', 'Miscellaneous'), ] # -- Options for Epub output ------------------------------------------------- # Bibliographic Dublin Core info. epub_title = project # The unique identifier of the text. This can be a ISBN number # or the project homepage. # # epub_identifier = '' # A unique identification for the text. # # epub_uid = '' # A list of files that should not be packed into the epub file. epub_exclude_files = ['search.html'] # -- Extension configuration -------------------------------------------------
5,415
Python
.py
137
37.445255
79
0.663671
truenas/middleware
2,277
481
13
LGPL-3.0
9/5/2024, 5:13:34 PM (Europe/Amsterdam)