owner stringclasses 14
values | repo stringclasses 14
values | id int64 116k 4.61B | issue_number int32 1 181k | author stringlengths 1 39 | body stringlengths 1 262k | created_at timestamp[us]date 2000-06-06 02:40:44 2026-06-02 20:16:18 | updated_at timestamp[us]date 2000-06-06 02:40:44 2026-06-02 20:22:56 | reactions unknown | author_association stringclasses 5
values |
|---|---|---|---|---|---|---|---|---|---|
ClickHouse | ClickHouse | 478,196,027 | 4,860 | Gladdy | @den-crane Thanks for the reference, I managed to overlook that issue in my search. The issue raised there is the memory consumption for the lookup tables (I haven't dug into the implementation yet), but if you keep the resolution of the lookup tables the same, 2^32 seconds = 136 years and 2^64 nanos = 584 years, which... | 2019-03-30T01:59:42 | 2019-03-30T01:59:42 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,211,505 | 523 | lemmsh | I don't quite get how... I only see in that commit that libunwind has been moved from contrib | 2019-03-30T06:05:02 | 2019-03-30T06:05:02 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,401,388 | 4,840 | kindred77 | libboost-dev 1.58 is installed.
apt list | grep libboost-dev
libboost-dev/xenial 1.58.0.1ubuntu1 amd64
cmake ..
-- IPO/LTO not enabled.
-- CMAKE_BUILD_TYPE is not set, set to default = RELWITHDEBINFO
-- CMAKE_BUILD_TYPE: RELWITHDEBINFO
-- Using linker: gold (selected from: LLD_PATH=LLD_PATH-NOTFOUND; GOLD_... | 2019-04-01T00:41:44 | 2019-04-01T00:41:44 | {} | NONE |
ClickHouse | ClickHouse | 478,599,050 | 4,785 | alexey-milovidov | > Polygon dictionaries
Cancelled (moved to 2020). | 2019-04-01T14:19:41 | 2019-04-01T14:19:41 | {} | MEMBER |
ClickHouse | ClickHouse | 478,792,931 | 4,631 | simPod | It is related to under-replicated partitions I believe
It started the same time the unde rreplication occurred | 2019-04-02T00:12:25 | 2019-04-02T00:12:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 479,055,849 | 4,883 | alesapin | Performance seems like flap. | 2019-04-02T15:36:56 | 2019-04-02T15:36:56 | {} | MEMBER |
ClickHouse | ClickHouse | 479,427,947 | 4,888 | proller | Already fixed https://github.com/yandex/ClickHouse/pull/4835 | 2019-04-03T10:14:21 | 2019-04-03T10:14:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,115,083 | 4,856 | nvartolomei | Nice debug :) there is still a very small chance that connection will be in close wait if this check succeeds, time drift/vm pauses. Negligible to some degree :) | 2019-03-29T19:05:30 | 2019-03-29T19:05:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,125,948 | 4,402 | filimonov | > BTW, ASan report inside `libunwind` is normal (by design).
Wierd :) In that case is there a way to skip asan checks for libunwind?
> Now I can't reproduce it on master. Probably, it's connected with [eee7835](https://github.com/yandex/ClickHouse/commit/eee783580b6681e0d462ed35fb8659b79c5860c7)
Yep, may be. ... | 2019-03-29T19:42:30 | 2019-03-29T19:42:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,213,226 | 4,860 | filimonov | IMHO it's nice to have 'someday', currently not high priority task at all. I would say that option with storing microseconds since Epoch in signed Int64 (exactly like in JavaScript Date) sounds better than nanoseconds (nanoseconds is too granular for most of database tasks, with microseconds DateTime64 can process any ... | 2019-03-30T06:33:19 | 2019-03-31T07:52:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,656,498 | 4,875 | 4ertus2 | FYI @Gladdy
There's a crash with last changes on my data. I'm trying to investigate what is wrong.
```
2019.04.01 17:52:46.896839 [ 32 ] {} <Error> BaseDaemon: ########################################
2019.04.01 17:52:46.898270 [ 32 ] {} <Error> BaseDaemon: (version 19.5.1.1) (from thread 30) Received signal Seg... | 2019-04-01T16:47:17 | 2019-04-01T16:50:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 477,931,393 | 4,837 | nvartolomei | One thing that is missing: not respecting the timeouts in this `for (;;)` loop. If for some reason the server goes rogue we won't have a clear error in logs and replication_queue table. | 2019-03-29T09:34:10 | 2019-03-29T09:34:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,100,108 | 4,837 | alesapin | Better fix https://github.com/yandex/ClickHouse/pull/4856. | 2019-03-29T18:19:00 | 2019-03-29T18:19:00 | {} | MEMBER |
ClickHouse | ClickHouse | 478,194,954 | 4,860 | den-crane | https://github.com/yandex/ClickHouse/issues/3507 | 2019-03-30T01:44:40 | 2019-03-30T01:44:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,277,603 | 4,863 | Gladdy | Thanks very much for some useful feedback, will try to submit a patch tonight. | 2019-03-30T18:55:41 | 2019-03-30T18:55:41 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,535,634 | 4,736 | aegorov05 | > I had similar problems: a small amount of messages was lost.
> It turned out that the messages were recorded in Kafka with incorrect timestamp of messages, instead of milliseconds, seconds were written.
> Maybe you also write seconds to timestamps?
No, we write ms to timestamp
But I think CH Kafka Consumer shou... | 2019-04-01T11:06:55 | 2019-04-01T11:24:40 | {} | NONE |
ClickHouse | ClickHouse | 479,529,400 | 4,875 | Gladdy | I doubt PaddedPODArray vs std::vector is going to make a huge difference in performance (and if there is any performance issue later down the line profiling will show whether it is actually in that bit of the code), so whilst I did think that swapping out std::map versus this approach was going to make a big difference... | 2019-04-03T15:04:58 | 2019-04-03T15:04:58 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,235,757 | 523 | alexey-milovidov | #4792 | 2019-03-30T11:09:38 | 2019-03-30T11:09:38 | {} | MEMBER |
ClickHouse | ClickHouse | 478,242,192 | 4,713 | Enmk | > .
Sorry for delay, it should be fixed now. | 2019-03-30T12:40:17 | 2019-03-30T12:40:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,514,420 | 4,658 | dstruck | Latest release still has the issue.
```
$ clickhouse-client
ClickHouse client version 19.4.2.7.
Connecting to localhost:9000 as user default.
Connected to ClickHouse server version 19.4.2 revision 54416.
xxx :) SELECT timezone()
SELECT timezone()
┌─timezone()────────┐
│ Europe/Berlin │
└─────────... | 2019-04-01T09:55:40 | 2019-04-01T11:10:51 | {} | NONE |
ClickHouse | ClickHouse | 478,675,329 | 4,631 | kmatt | @simPod Correct - behavior returned shortly after I posted that message. Still looking at librdkafka parameters, currently heartbeat and timeout. | 2019-04-01T17:41:42 | 2019-04-01T17:41:42 | {} | NONE |
ClickHouse | ClickHouse | 478,717,924 | 2,901 | simPod | As I learnt some time ago, materialized view is just a little unfortunately chosen name for insert trigger. Inserting into view is just inserting into it's inner table.
If you need to populate by range, I guess it would be better to use INSERT AS SELECT. But personally. I'd not use POPULATE anymore just because it s... | 2019-04-01T19:42:03 | 2019-04-01T19:42:03 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 477,993,958 | 523 | alexey-milovidov | We have implemented this "unmanageable hell" in #4828 | 2019-03-29T13:17:21 | 2019-03-29T13:17:21 | {} | MEMBER |
ClickHouse | ClickHouse | 478,299,225 | 4,863 | 4ertus2 | Current PR is OK. But it looks there's one more issue at Join.cpp:869 and Join.cpp:897. There we just copy left key column to the right one expecting that they contain the same data. If it's an ASOF time column the expectation is wrong. We need to extract real B.t data instead of copying A.t. | 2019-03-30T23:50:02 | 2019-03-30T23:53:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,299,830 | 4,863 | Gladdy | Did you manage to get ASOF INNER working on your dataset?
I noticed the issue with the wrong timestamps in my original pull request, though I haven't gone around to look at the the joining code in as much detail yet to work out the best way of extracting the real B.t instead of as you say a copy of A.t without some... | 2019-03-31T00:03:21 | 2019-03-31T00:04:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,310,083 | 4,847 | fancyqlx | @alexey-milovidov Hi. I have added a test and some additional fixes for this function. | 2019-03-31T04:02:09 | 2019-03-31T04:02:09 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,394,242 | 4,863 | Gladdy | Currently an incremental build takes about 7 minutes (from start of build until a final binary is ready to run a test) on my laptop upon a single change to Join.cpp, which isn't ideal but workable I guess. The right timestamp values are working, will submit a pull request shortly. | 2019-03-31T23:07:05 | 2019-03-31T23:07:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,595,278 | 4,865 | ShivaS123 | Seems like it's fixed and recently been committed https://github.com/yandex/ClickHouse/pull/4794 ?
| 2019-04-01T14:10:34 | 2019-04-01T14:13:13 | {} | NONE |
ClickHouse | ClickHouse | 478,662,178 | 4,875 | Gladdy | @4ertus2 The last changes being the removal of the macros or the my cleanup to make the timestamps work correctly? | 2019-04-01T17:03:51 | 2019-04-01T17:03:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,888,549 | 4,878 | alex-krash | @kindred77 , you have to start Java bridge application https://github.com/alex-krash/clickhouse-jdbc-bridge
According to error, you've provided, you are trying to query MySQL via **odbc**-bridge. JDBC-bridge requires standalone Java application with MySQL driver loaded. | 2019-04-02T08:02:26 | 2019-04-02T08:02:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 479,377,224 | 4,886 | filimonov | [Altinity](https://github.com/Altinity) ( [website](http://altinity.com) ) is currently working on clickhouse operator for Kurbenetes. That should automate most of ClickHouse-specific tasks.
Please check that https://github.com/Altinity/clickhouse-operator/ (beta stage currently), any feedback/review would be helpf... | 2019-04-03T07:43:41 | 2019-04-03T07:43:41 | {
"+1": 1
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 477,930,639 | 4,837 | nvartolomei | > We will not create new Poco sessions if session for this schema://URI:port is already in pool
We don't have to. [receiveResponse](https://github.com/ClickHouse-Extras/poco/blob/d0271f990869b7f23dac4bb2ee0c127cf815f077/Net/src/HTTPClientSession.cpp#L279) closes the connection on exception.
If we get the same htt... | 2019-03-29T09:31:44 | 2019-03-29T09:31:44 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,097,639 | 4,855 | 4ertus2 | #4681 | 2019-03-29T18:11:11 | 2019-03-29T18:11:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,798,299 | 4,840 | kindred77 | it works, thanks very much. | 2019-04-02T00:42:37 | 2019-04-02T00:42:37 | {} | NONE |
ClickHouse | ClickHouse | 479,081,805 | 4,875 | 4ertus2 | > Ah, I see now that `std::is_trivially_copyable<AsofRowRefs>::value == false`, which is a requirement for std::memmove (https://en.cppreference.com/w/cpp/string/byte/memmove) , so it shouldn't be stored in the HashTable as there is a comment about it:
Yup, I'm on the way to split AsofRowRefs::Entry and lookup struc... | 2019-04-02T16:30:51 | 2019-04-02T16:30:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 479,523,751 | 4,875 | 4ertus2 | > We should probably add a test that stresses the ASOF join beyond 4k elements with a sufficiently big hash table that the moving actually occurs and/or adding a static assert that the requirements on the hash table values are actually met?
Currently SortedLookupPODArray uses std::vector<T> as Base cause if I change... | 2019-04-03T14:51:50 | 2019-04-03T14:54:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 479,535,577 | 4,875 | 4ertus2 | > I doubt PaddedPODArray vs std::vector is going to make a huge difference in performance
Me too. The question is why implementation with PODArray crashes. It could be some error that is well visible with PODArray and may be hidden but not fixed with std::vector.
> If you do not place std::vector in HashMap anymo... | 2019-04-03T15:20:04 | 2019-04-03T15:20:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 477,995,070 | 4,700 | alexey-milovidov | Implemented in #4845 | 2019-03-29T13:21:11 | 2019-03-29T13:21:23 | {} | MEMBER |
ClickHouse | ClickHouse | 478,279,607 | 4,863 | Gladdy | ASOF INNER join should be supported already? There is a test for it anyway that gets me the results I expect: https://github.com/yandex/ClickHouse/blob/master/dbms/tests/queries/0_stateless/00927_asof_join_noninclusive.sql | 2019-03-30T19:12:26 | 2019-03-30T19:12:26 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,527,244 | 4,867 | Gladdy | Thanks, will patch those up. Do you have any better way of checking the code style other than the included scripts in utils/check-style that would actually catch these issues (they, and by extension your CI system does not report these style issues)? Also, when I actually run the clang-format tool with the included .cl... | 2019-04-01T10:36:51 | 2019-04-01T10:36:51 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,531,533 | 4,867 | 4ertus2 | > Thanks, will patch those up. Do you have any better way of checking the code style other than the included scripts in utils/check-style that would actually catch these issues (they, and by extension your CI system does not report these style issues)?
We have style checker but it is not ideal yet. Maybe one day it ... | 2019-04-01T10:52:29 | 2019-04-01T10:52:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,550,431 | 4,840 | proller | Can you try remove libboost-dev and recompile with deletion of build/CMakeCache.txt ?
Or compile with your system boost : `cmake -DUSE_INTERNAL_BOOST_LIBRARY=0` | 2019-04-01T12:00:38 | 2019-04-01T12:00:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 479,015,213 | 4,631 | simPod | But maybe it's implemented through cppkafka https://github.com/mfontanini/cppkafka/blob/4a887607b356eeef715306d442242d3a8d55c13e/src/consumer.cpp#L316 | 2019-04-02T14:13:08 | 2019-04-02T14:13:08 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 477,938,276 | 4,837 | alesapin | ```From my understanding keep-alive on server side will keep the connection open for keep-alive timeout, but close it after that. Poco implementation seems fine.```
Our default `keep_alive_timeout` is 3 seconds https://github.com/yandex/ClickHouse/blob/master/dbms/programs/server/config.xml#L88 for Interserver handl... | 2019-03-29T09:56:12 | 2019-03-29T10:01:26 | {} | MEMBER |
ClickHouse | ClickHouse | 477,947,130 | 4,774 | 4ertus2 | > Furthermore, I had a quick look through the codebase looking for binary-tree like structure that is Arena allocator enabled, but wasn't able to find any. Is it correct that clickhouse doesn't use any, eg. for lookups into indexed columns?
We've discussed it with @alexey-milovidov. The right way is to insert data i... | 2019-03-29T10:24:04 | 2019-03-29T10:24:04 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,104,773 | 4,402 | KochetovNicolai | Now I can't reproduce it on master. Probably, it's connected with https://github.com/yandex/ClickHouse/commit/eee783580b6681e0d462ed35fb8659b79c5860c7
Now I also can't create manual test to reproduce this bug. | 2019-03-29T18:33:09 | 2019-03-29T18:33:09 | {} | MEMBER |
ClickHouse | ClickHouse | 478,226,869 | 4,389 | zombiemonkey | Didn't get back to you - but thanks for addressing this. | 2019-03-30T09:38:10 | 2019-03-30T09:38:10 | {} | NONE |
ClickHouse | ClickHouse | 478,281,110 | 4,863 | 4ertus2 | > ASOF INNER join should be supported already? There is a test for it anyway that gets me the results I expect: https://github.com/yandex/ClickHouse/blob/master/dbms/tests/queries/0_stateless/00927_asof_join_noninclusive.sql
Hmm... It could be my bad (data or query). I'll test one more time later. | 2019-03-30T19:26:22 | 2019-03-30T19:26:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,693,821 | 2,901 | mcarbonneaux | when origin table are enormous, populate can be very long and can fail...
can be usefull to populate by range of data to avoid fail...
will be usefull to more secure the populate option to be abel to trottle the populate even if the populate will be more long....
or add the ability to insert into the materialized... | 2019-04-01T18:34:29 | 2019-04-01T18:34:58 | {} | NONE |
ClickHouse | ClickHouse | 478,769,642 | 4,875 | Gladdy | Not sure about the PODArray limitations (though shouldn't the storage extend beyond end as otherwise use you are in invalid memory? end_of_storage should be > end? The 4k might be hitting the next (unmapped) page or is just an artifact of the initial size and it not resizing correctly beyond that?)
Ah, I see now th... | 2019-04-01T22:25:53 | 2019-04-02T11:37:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 479,059,706 | 1,567 | healiseu | My complements to @valyala for raising this issue and of course to @alexey-milovidov, @KochetovNicolai and others that contributed to its implementation. Although this is solved I have opened a discussion on its performance and possibly an alternative more efficient implementation using bit-stuffed pointers (#4796) and... | 2019-04-02T15:44:35 | 2019-04-02T15:57:30 | {} | NONE |
ClickHouse | ClickHouse | 477,898,881 | 4,837 | alesapin | @nvartolomei Do you have any ideas how to reproduce this problem in test? | 2019-03-29T07:35:23 | 2019-03-29T07:35:23 | {} | MEMBER |
ClickHouse | ClickHouse | 477,956,586 | 4,843 | fancyqlx | #4847 | 2019-03-29T10:57:14 | 2019-03-29T10:57:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,226,160 | 4,392 | zombiemonkey | Didn't get back to you - but thanks for addressing this. Can confirm that 19.4.x has been rock solid. | 2019-03-30T09:31:25 | 2019-03-30T09:31:25 | {} | NONE |
ClickHouse | ClickHouse | 478,257,055 | 4,860 | Gladdy | There are some examples of use cases where it's useful (if not required) to have nanosecond precision, one of them is finance/trading (for pricing/trade/response time analytics).
I agree that it compresses less well, though I'd argue that if the high precision information is needed then there is not much you can do ... | 2019-03-30T15:38:14 | 2019-03-30T15:38:14 | {
"+1": 4
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,272,593 | 523 | lemmsh | awesome!
From a quick look on the code I see that only static expression can be in the filter for now. Almost all our use cases require filters like 'where some_id in (select id from allowed_ids where user = '${current_user}')'. But it looks like relatively easy to fix.
Why didn't you make it on AST level? it se... | 2019-03-30T18:06:33 | 2019-03-30T18:06:33 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,288,206 | 4,791 | filimonov | Most probably docker volume / filesystem related error. Look like it tries to create hardlinks between different docker volumes / different overlay FS layers.
Could you please run bush inside your container and check if you are able to create hardlinks inside that folder?
I.e. something like
```
docker exec ... | 2019-03-30T20:49:41 | 2019-03-30T20:50:45 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,330,193 | 4,863 | 4ertus2 | > Did you manage to get ASOF INNER working on your dataset?
Yup, there were some errors in my queries. It looks working.
> I noticed the issue with the wrong timestamps in my original pull request, though I haven't gone around to look at the the joining code in as much detail yet to work out the best way of extr... | 2019-03-31T10:40:00 | 2019-03-31T10:41:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,359,729 | 4,863 | Gladdy | @4ertus2 I was hoping to get the right timestamp value working today, though so far no luck. Between taking quite a while to perform an incremental build (any tips on the fastest building configuration? I've tried enabling building shared objects and disabling most of the dependencies but it doesn't seem to make much o... | 2019-03-31T17:09:46 | 2019-03-31T17:09:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,375,238 | 4,863 | 4ertus2 | I don't know how to improve build speed much. Here we have developer servers in our VPN. If you have fast machine I can share you my build script that copies git working copy to build server, builds it there, run tests and fetch built binary (if needed). | 2019-03-31T19:57:17 | 2019-03-31T19:57:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,388,058 | 4,631 | simPod | @kmatt 👋 you wrote this
```
Testing with message size and reconnect parameters to librdkafka is showing some improvements in my case:
<kafka>
<fetch_max_bytes>104857600</fetch_max_bytes>
<fetch_message_max_bytes>104857600</fetch_message_max_bytes>
<receive_message_max_bytes>105000000</receive_message_ma... | 2019-03-31T21:42:03 | 2019-03-31T21:42:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,663,820 | 4,875 | 4ertus2 | > @4ertus2 The last changes being the removal of the macros or the my cleanup to make the timestamps work correctly?
Cleanup & changes from std::map to PODArray. | 2019-04-01T17:08:35 | 2019-04-01T17:08:35 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,854,850 | 4,877 | ghost | Furthermore, for the case when one of the columns types is `Array` instead of `Nested`, it is enough to have it in only one of the joined tables to make the query not working:
```sql
CREATE TABLE left_table (a String, b String) ENGINE = Memory();
INSERT INTO left_table (a, b) VALUES ('a', 'b'), ('d', 'e');
CREA... | 2019-04-02T05:53:19 | 2019-04-02T06:02:55 | {} | NONE |
ClickHouse | ClickHouse | 478,927,494 | 4,879 | 4ertus2 | We need some time to investigate the issue. As workaround you may try to set alias to the column or try to JOIN with USING instead of ON. It may help. | 2019-04-02T09:56:56 | 2019-04-02T09:56:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 479,388,438 | 4,886 | bzon | @filimonov I actually tried it out and it's superb. But we are going to production in a couple of months so we really can't use it. | 2019-04-03T08:19:31 | 2019-04-03T08:19:31 | {} | NONE |
ClickHouse | ClickHouse | 479,407,572 | 4,886 | alex-zaitsev | > @filimonov I actually tried it out and it's superb. But we are going to production in a couple of months so we really can't use it.
@bzon , what features are you missing in Altinity operator? We can pay more attention on those to make it useful earlier for you. | 2019-04-03T09:13:59 | 2019-04-03T09:13:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 477,987,105 | 4,837 | alesapin | Don't merge please, I'm investigating the issue. | 2019-03-29T12:54:19 | 2019-03-29T17:57:51 | {} | MEMBER |
ClickHouse | ClickHouse | 478,291,676 | 4,863 | Gladdy | The latest commit should address all of the feedback on the pull request. I pulled all the RowRefs into a separate file RowRefs.h (RowRef, RowRefList and AsofRowRefs) and added a SortedLookupPODArray based on PaddedPODArray, alongside with fixing the minor formatting issues and changing `const Join *` into `const Join&... | 2019-03-30T21:41:46 | 2019-03-30T21:41:46 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,541,772 | 4,869 | blinkov | cc @zhang2014 @sundy-li | 2019-04-01T11:29:07 | 2019-04-01T11:29:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,983,386 | 4,631 | simPod | Shouldn't we have [rebalance cb](https://github.com/edenhill/librdkafka/blob/master/examples/rdkafka_consumer_example.cpp#L232)?
@vavrusa you have implemented Kafka engine. If I may ask, is it forgotten or there's a reason for it [not to be present](https://github.com/yandex/ClickHouse/search?q=rebalance&type=Code) ... | 2019-04-02T12:58:14 | 2019-04-02T12:58:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 479,417,162 | 3,917 | dgrr | Seems that you can rename tables if it does not have comments. When a table contains a comment it shows the following error:
```
Received exception from server (version 19.4.1):
Code: 62. DB::Exception: Received from localhost:9000, ::1. DB::Exception: Syntax error (in file /var/lib/clickhouse/metadata/db/supertable... | 2019-04-03T09:41:48 | 2019-04-03T09:45:16 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 477,942,977 | 4,843 | fancyqlx | It seems that the engine had not handled const arguments correctly. I will fix this later | 2019-03-29T10:10:48 | 2019-03-29T10:10:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,101,442 | 4,047 | alesapin | Sorry for ignoring this issue :( I didn't receive any notifications from GitHub.
**Fix:** https://github.com/yandex/ClickHouse/pull/4856.
**Work around:** set `<keep_alive_timeout>` in `config.xml` to 8 seconds. | 2019-03-29T18:23:04 | 2019-03-29T18:23:04 | {} | MEMBER |
ClickHouse | ClickHouse | 478,152,124 | 4,837 | nvartolomei | @alesapin I'd argue that this is still needed (in one form or another) and #4856 is just a hack to make it work for that specific use case. This exception would still be generated if one ClickHouse server for one reason or another has different keep alive settings, or as I mentioned in #4856 because of clock drifts. | 2019-03-29T21:15:43 | 2019-03-29T21:15:43 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,666,200 | 4,875 | Gladdy | Hmm - I'd be very curious to know what went wrong then. Note, that I haven't done much testing on "real" data yet, just focused on the writing and passing the test queries. If you need any help please let me know as I'm sure I'll run into the same errors when running asof on actual data that is more complicated than th... | 2019-04-01T17:15:37 | 2019-04-01T17:15:37 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 479,459,485 | 4,875 | Gladdy | Thanks for the work on making the AsofRowRefs movable (and spotting the race condition on the sorting that I had missed in introducing the SortedLookupPODArray). We should probably add a test that stresses the ASOF join beyond 4k elements with a sufficiently big hash table that the moving actually occurs and/or adding ... | 2019-04-03T12:01:25 | 2019-04-03T12:27:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,198,980 | 4,860 | Gladdy | Convenience is definitely a big part of it with proper formatting etc when running analytic queries but also general table cleanliness and preventing bugs in accidentally rounding the times when forgetting the fractional part. Lastly I recently added an asof join to ClickHouse which currently only performs the asof ope... | 2019-03-30T02:41:17 | 2019-03-30T02:41:17 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,206,545 | 4,847 | fancyqlx | > Thank you! Can you please also add a test (in `dbms/tests/queries/0_stateless` directory)?
No problem! There are still some problems in this function. I will take a further look and add some tests later. | 2019-03-30T04:38:48 | 2019-03-30T04:38:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,225,259 | 4,716 | zombiemonkey | @den-crane - I know :) I think you misunderstood that this was seeking clarity as to why set(0) was not working as documented. As KochetovNicolai pointed out - there was a bug fixed in #4640. 1000 was simply one of the values that was used for testing from a range of values and not the problem/issue WRT the behavior of... | 2019-03-30T09:23:39 | 2019-03-30T09:23:39 | {} | NONE |
ClickHouse | ClickHouse | 478,278,342 | 4,863 | 4ertus2 | > One more thing.
And there's another issue. If we joining with Enum that is on of very interested cases for trades it lead to an error, cause there's no default for enum. I'm happy that left 0 value unused here. So I'm able to alter update this enum.
```
Code: 49. DB::Exception: Unexpected value 0 for type Enum... | 2019-03-30T19:01:28 | 2019-03-30T19:04:20 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,280,745 | 4,744 | vitlibar | A possible reason for this segfault has been fixed (in master).
However to work correctly your queries should be edited:
1) Instead of
```
CREATE TABLE IF NOT EXISTS table (
TimeRecvd UInt32,
Device String,
SrcIP String
)
ENGINE = Kafka SETTINGS
kafka_broker_list = 'localhost:... | 2019-03-30T19:22:21 | 2019-03-30T19:23:28 | {} | MEMBER |
ClickHouse | ClickHouse | 478,640,886 | 4,791 | yu45020 | @filimonov
`some_table1/tmp_mut_all_1_1_0_2` is not created when I update values, so Docker fails to link the folder.
I guess this is a problem caused by Ubuntu because it mounts a ntfs disk as root user, so user `default` can't modify folders under `/var/lib`. I am wondering whether loginning in as root will so... | 2019-04-01T16:03:14 | 2019-04-01T16:03:14 | {} | NONE |
ClickHouse | ClickHouse | 478,691,875 | 4,518 | ghost | You can flush logs with `SYSTEM FLUSH LOGS` query to see the queries in the logs table immediately.
For your example if the last command of the example is changed to
```
echo -e "SELECT id1, sum(id2) AS id2 FROM G1 WHERE 1=0 GROUP BY id1;\n\nSYSTEM FLUSH LOGS;\n\nSELECT count(*) AS cnt_where_false FROM system.quer... | 2019-04-01T18:28:48 | 2019-04-01T18:33:51 | {} | NONE |
ClickHouse | ClickHouse | 478,736,011 | 4,710 | knoguchi | I confirmed the bug has been fixed. Thank you! | 2019-04-01T20:31:49 | 2019-04-01T20:31:49 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,112,293 | 4,736 | portmare | I had similar problems: a small amount of messages was lost.
It turned out that the messages were recorded in Kafka with incorrect timestamp of messages, instead of milliseconds, seconds were written.
Maybe you also write seconds to timestamps?
To check this, use **kafkacat**:
`> kafkacat -b broker:9092 -t topic ... | 2019-03-29T18:56:43 | 2019-03-29T18:56:43 | {} | NONE |
ClickHouse | ClickHouse | 478,277,787 | 4,863 | 4ertus2 | One more thing. When I write 'ASOF JOIN' it leads to 'ASOF INNER JOIN' and start doing something instead of exception that it's not supported yet. We need negative test for it. You may find some of them by searching { serverError XX } in stateless tests files. | 2019-03-30T18:57:06 | 2019-03-30T18:57:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,278,857 | 4,710 | vitlibar | The bug has been fixed (in master). However to work correctly your script should be edited:
1) Instead of
```
create table simple (
t UInt64,
url String
) ENGINE=Kafka('localhost:9092', 'topic', 'mygroup', 'Protobuf');
```
use either
```
create table simple (
t UInt64,
url String
) ENGINE=Kafka(... | 2019-03-30T19:06:13 | 2019-03-30T21:12:52 | {} | MEMBER |
ClickHouse | ClickHouse | 478,697,964 | 4,875 | 4ertus2 | > Hmm - I'd be very curious to know what went wrong then. Note, that I haven't done much testing on "real" data yet, just focused on the writing and passing the test queries. If you need any help please let me know as I'm sure I'll run into the same errors when running asof on actual data that is more complicated than ... | 2019-04-01T18:45:28 | 2019-04-01T19:46:14 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,929,038 | 4,880 | 4ertus2 | Need reproduce on 19.5+. It's hard to merge changes from #4681 into 19.4. | 2019-04-02T10:01:29 | 2019-04-02T10:01:29 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 477,966,473 | 4,847 | alexey-milovidov | Thank you! Can you please also add a test (in `dbms/tests/queries/0_stateless` directory)? | 2019-03-29T11:34:46 | 2019-03-29T11:35:04 | {} | MEMBER |
ClickHouse | ClickHouse | 477,707,276 | 4,796 | healiseu | @KochetovNicolai your comments are indeed helpful, it seems this positional encoded representation is not suitable for all cases. I think it can be varied according to the data column. Have you considered to use [bit-stuffed pointers][2] as keys. Such indices only have as many bits that it takes to represent the field.... | 2019-03-28T17:59:07 | 2019-03-29T09:24:38 | {} | NONE |
ClickHouse | ClickHouse | 478,675,723 | 2,045 | jenyacazacu | is there any update on this, any plans to release encryption at rest in a future version? | 2019-04-01T17:42:51 | 2019-04-01T17:42:51 | {
"+1": 6
} | NONE |
ClickHouse | ClickHouse | 478,755,016 | 3,912 | dshuvar | @alesapin That's a little bit weird having rpm branch in a repo and do not support it. | 2019-04-01T21:29:06 | 2019-04-01T21:29:06 | {
"+1": 1
} | NONE |
ClickHouse | ClickHouse | 477,907,787 | 4,837 | alesapin | Also I don't understand why this (implemented) solution will help:
```
retry this error until we succeed. Poco will create new
connections for us.
```
We will not create new Poco sessions if session for this `schema://URI:port` is already in pool https://github.com/yandex/ClickHouse/blob/master/dbms/src/IO/HTTPCom... | 2019-03-29T08:11:02 | 2019-03-29T08:16:25 | {} | MEMBER |
ClickHouse | ClickHouse | 477,972,440 | 4,840 | proller | Did you have libboost-dev installed?
Can you show full cmake and ninja/make output ? | 2019-03-29T11:59:31 | 2019-03-29T11:59:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,177,045 | 4,681 | alexey-milovidov | Fixed in master. | 2019-03-29T23:16:15 | 2019-03-29T23:16:15 | {} | MEMBER |
ClickHouse | ClickHouse | 478,197,673 | 4,860 | den-crane | I don't really understand what is the difference if you store a seconds' fraction in another column?
Just convince for in your backend? | 2019-03-30T02:22:22 | 2019-03-30T02:22:52 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 478,282,693 | 4,860 | filimonov | > There are some examples of use cases where it's useful (if not required) to have nanosecond precision, one of them is finance/trading (for pricing/trade/response time analytics).
Of course there are some special scenarios when nanoseconds or even picosecond are needed. But because of CERN need picosecond precision... | 2019-03-30T19:40:49 | 2019-03-31T07:55:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 479,428,063 | 4,886 | bzon | @alex-zaitsev the most important thing is proven stability and production scaling patterns (documentation). Think use cases for production workloads at a massive scale. Basically, the best practices in clickhouse operational work that clickhouse operator is solving and how does it do it. | 2019-04-03T10:14:41 | 2019-04-03T10:14:41 | {} | NONE |
ClickHouse | ClickHouse | 479,447,245 | 4,888 | mobilesfinks | Update to the specified version 19.4.3.11
Same problem
```
root@kube-clickhouse-stage-copier-m9wr5:/app# ./copy_start.sh
Include not found: zookeeper-servers
Include not found: clickhouse_distributed_ddl
Logging debug to /app/clickhouse-copier_20190403111609_48/log.log
Logging errors to /app/clickhouse-copier_... | 2019-04-03T11:20:22 | 2019-04-03T11:20:22 | {} | NONE |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.