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 | 743,256,419 | 18,006 | den-crane | ```
CREATE TABLE example (k Int32, v Int32) ENGINE=MergeTree ORDER BY(k);
INSERT INTO example SELECT number,number FROM numbers(100000000)
SELECT k, max(v) AS max_v FROM example GROUP BY k HAVING max_v == 100;
Elapsed: 4.539 sec.
set optimize_aggregation_in_order=1;
SELECT k, max(v) AS max_v FROM example GR... | 2020-12-11T15:22:40 | 2020-12-11T15:22:40 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,303,199 | 17,500 | filimonov | May be it's related / other form of https://github.com/ClickHouse/ClickHouse/issues/14582 ? | 2020-12-11T16:47:19 | 2020-12-11T16:47:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,361,786 | 18,011 | dpzmick | wow! I updated my clickhouse image:
```
$ podman images | grep clickhouse
docker.io/yandex/clickhouse-server latest 8dc37897ab10 2 days ago 818 MB
```
and the performance is now _much much_ better:
```
setup 2.219663143157959s
fast: 0.0668754577636718... | 2020-12-11T18:47:12 | 2020-12-11T18:47:12 | {} | NONE |
ClickHouse | ClickHouse | 742,636,906 | 17,960 | den-crane | Probably you can share this table with us because it's an open data ?
clickhouse-client -q "select * from coinpath.txs_ethereum_outbound_4c4dce9598b8e2addcc99 format Native" | gzip > test.native.gz | 2020-12-10T16:37:35 | 2020-12-10T16:37:59 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,821,650 | 17,641 | akuzm | > > I think we need to patch Poco (mark displayText as virtual).
>
> `DB::Exception::what()` is virtual, isn't this enough? Patching Poco is something I'd like to avoid...
>
> UPD Oops, it's a different method, it shows the exception name.
Well, maybe if we only have to add `virtual` keyword to `Poco::Exceptio... | 2020-12-10T21:49:22 | 2020-12-10T21:49:22 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,065,364 | 17,913 | l1t1 | thanks | 2020-12-11T08:58:57 | 2020-12-11T08:58:57 | {} | NONE |
ClickHouse | ClickHouse | 742,680,473 | 17,960 | den-crane | I've reproduced but only with dict.
```sql
CREATE TABLE test3 ( `depth` UInt32,
`transfer_from_bin` String,
`transfer_to_bin` String,
`amount` Float64) ENGINE = TinyLog;
insert into test3 values (1,'1','2',3.99999958e21),(1,'1','3',100000000000000000);
create table currency_t (id UInt64, divid... | 2020-12-10T17:43:38 | 2020-12-10T18:17:06 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,707,042 | 17,960 | den-crane | even without a table
```sql
SELECT
d,
SUM(am) / dictGet('default.currency', 'divider', toUInt64(1)) AS a
FROM (select toUInt32(1) d,'1' f ,'2' t,3e21 am union all select 1,'1','3',1e18)
GROUP BY d,f,t
ORDER BY d , a
┌─d─┬────a─┐
│ 1 │ 3000 │
│ 1 │ 1 │
└───┴──────┘
```
| 2020-12-10T18:25:49 | 2020-12-10T18:26:21 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,814,274 | 17,984 | nikitamikhaylov | Not sure of what I am doing and hope for the best. | 2020-12-10T21:34:31 | 2020-12-10T21:34:31 | {} | MEMBER |
ClickHouse | ClickHouse | 742,935,496 | 17,991 | spongedu | In my opinion, we should either retry `zookeeper` initialization or make clickhosue server exit. The problem we met is neither option is done and the server can not any ddl job | 2020-12-11T03:05:18 | 2020-12-11T03:05:18 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,990,384 | 17,646 | filimonov | I don't like the proposed setting.
For me it sounds like flush from buffer should happen in context/profile of another user (system_profile?). And that flush should be registered in query_log separately from insert into buffer table.
So we actually dealing with 2 inserts. Tiny and cheap insert into buffer, and h... | 2020-12-11T06:01:07 | 2020-12-11T06:01:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,001,783 | 17,995 | hexiaoting | maybe need to upgrade your glibc.
show the result of
```
ldd ./programs/clickhouse
ldd /lib/aarch64-linux-gnu/libpthread.so.0
```
it depends on libc.so
and show
```
strings libc.so |grep GLIBC
``` | 2020-12-11T06:29:03 | 2020-12-11T06:31:53 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,048,672 | 17,962 | chenk818 | > Thanks for reply
> There is still more than 70% free space there and also there is low cpu io wait
> /dev/nvme1n1 825G 211G 573G 27% /var/lib/clickhouse
> There is enough free space here. it seems like mutate process was blocked by merge process
> We're doing batch update 4 seconds a time and we already changed t... | 2020-12-11T08:22:44 | 2020-12-11T08:22:44 | {} | NONE |
ClickHouse | ClickHouse | 743,062,308 | 17,955 | chenk818 | there are 2shards with same replica.
when the first shard complete mutation work,then clearOldMutations thread will delete the node from zk.
so the second shard can't get node from zk.
i'm not sure，you can have a try | 2020-12-11T08:52:39 | 2020-12-11T08:52:39 | {} | NONE |
ClickHouse | ClickHouse | 743,226,182 | 18,006 | hrissan | Thanks, I added SETTINGS to both requests
SELECT k, max(v) AS max_v FROM example GROUP BY k HAVING max_v = 100 SETTINGS optimize_aggregation_in_order = 1
SELECT k, max_v FROM (SELECT k, max(v) AS max_v FROM example GROUP BY k) WHERE max_v == 100 SETTINGS optimize_aggregation_in_order = 1
And I confirm they bot... | 2020-12-11T14:33:07 | 2020-12-11T14:33:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,281,705 | 17,951 | SaltTan | https://github.com/ClickHouse/ClickHouse/issues/17989 | 2020-12-11T16:08:03 | 2020-12-11T16:08:03 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,441,381 | 18,015 | den-crane | FROM r0.**".inner.minutely_test_events"**
```sql
CREATE MATERIALIZED VIEW r0.hourly_test_events
(
`hour` DateTime,
`name` String,
`sumState` AggregateFunction(sum, UInt64)
)
ENGINE = AggregatingMergeTree()
PARTITION BY toStartOfDay(hour)
PRIMARY KEY (hour, name)
ORDER BY (hour, name) AS
SELE... | 2020-12-11T21:41:24 | 2020-12-11T21:41:38 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,480,564 | 18,000 | alexey-milovidov | Does not look like a reasoanable fix, but nevertheless, let's give it a chance. | 2020-12-11T23:30:43 | 2020-12-11T23:30:43 | {} | MEMBER |
ClickHouse | ClickHouse | 742,710,216 | 11,298 | azat | This PR also removes `limit_jobs.cmake` usage, @abyss7 was it done on purpose? (if so, why `limit_jobs.cmake` had been kept?) | 2020-12-10T18:30:50 | 2020-12-10T18:30:50 | {} | MEMBER |
ClickHouse | ClickHouse | 742,847,688 | 17,813 | alexey-milovidov | @abyss7 did nothing. | 2020-12-10T22:45:32 | 2020-12-10T22:45:32 | {} | MEMBER |
ClickHouse | ClickHouse | 743,213,899 | 16,935 | Avogar | Fail in integration tests is not related. | 2020-12-11T14:11:29 | 2020-12-11T14:11:29 | {} | MEMBER |
ClickHouse | ClickHouse | 743,285,502 | 18,001 | den-crane | toStartOfDay(b.log_day - interval 1 day) =toStartOfDay(a.log_day)
```sql
SELECT DISTINCT log_day,a.user as user_day0,b.user as user_day1
FROM (
SELECT min(log_day) as log_day, user
FROM t_user_login
GROUP BY user
) a
LEFT JOIN t_user_login b
ON toStartOfDay(b.log_day - interval 1 day) =toStartOfDay(a.l... | 2020-12-11T16:15:06 | 2020-12-11T16:16:15 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,299,465 | 18,006 | hrissan | You have too little data in your example. Try 1 billion and 4.5 seconds becomes infinity, because of OOM).
But I see your point. Parallel processing allows out-of-order aggregating of table chunks, so can be faster.
Thanks for expedite answers! | 2020-12-11T16:40:31 | 2020-12-11T16:40:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,445,514 | 18,006 | den-crane | CH can spill groupby
```
┌───k─┬─max_v─┐
│ 100 │ 100 │
└─────┴───────┘
1 rows in set. Elapsed: 276.189 sec. Processed 3.38 billion rows, 27.07 GB (12.25 million rows/s., 98.01 MB/s.)
set optimize_aggregation_in_order=1
1 rows in set. Elapsed: 331.623 sec. Processed 3.38 billion rows, 27.07 GB (10.20 mi... | 2020-12-11T21:51:19 | 2020-12-11T21:51:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,664,674 | 17,909 | Avogar | Let's wait for the checks and merge it | 2020-12-10T17:18:18 | 2020-12-10T17:18:18 | {} | MEMBER |
ClickHouse | ClickHouse | 742,819,438 | 17,982 | azat | 01561_Date_and_DateTime64_comparision - already reported https://github.com/ClickHouse/ClickHouse/pull/17895#issuecomment-742101261
It does not work at this time of day... | 2020-12-10T21:44:23 | 2020-12-10T21:44:23 | {} | MEMBER |
ClickHouse | ClickHouse | 742,819,925 | 17,977 | azat | Does checks will be continued w/o sanitizers build success? (i.e. integration release and similar, the one that does not require sanitizers builds) | 2020-12-10T21:45:29 | 2020-12-10T21:45:54 | {} | MEMBER |
ClickHouse | ClickHouse | 743,039,552 | 17,990 | filimonov | https://github.com/ClickHouse/ClickHouse/blob/674d8d44b9e2542e142339c3002da3fb4f3eeb34/programs/main.cpp#L312-L340 | 2020-12-11T08:02:47 | 2020-12-11T08:02:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,177,341 | 18,003 | filimonov | Please share log fragments. Nodes don't go readonly when you shutdown another node. Do you also run zookeeper on the same nodes? | 2020-12-11T12:55:55 | 2020-12-11T12:55:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,483,124 | 17,977 | alexey-milovidov | @azat I see strange fail of the `01581_to_int256_uint256_inf_nan` test. | 2020-12-11T23:40:01 | 2020-12-11T23:40:01 | {} | MEMBER |
ClickHouse | ClickHouse | 742,815,329 | 17,641 | nikitamikhaylov | I think we need to patch Poco (mark displayText as virtual). | 2020-12-10T21:36:15 | 2020-12-10T21:36:15 | {} | MEMBER |
ClickHouse | ClickHouse | 742,819,581 | 17,641 | akuzm | > I think we need to patch Poco (mark displayText as virtual).
`DB::Exception::what()` is virtual, isn't this enough? Patching Poco is something I'd like to avoid...
UPD Oops, it's a different method, it shows the exception name. | 2020-12-10T21:44:43 | 2020-12-10T21:46:56 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,040,679 | 17,955 | filimonov | ```
<shard1>Server1_IP</shard1>
```
Isn't it possible for your servers to change the IP due to some network reconfiguration?
p.s. Why circular replication? On 2 node the disadvantages of that are much higher than potential gains. Just do 2 replicas.
| 2020-12-11T08:05:22 | 2020-12-11T08:08:07 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,100,277 | 16,621 | ilejn | Please, take care about ilejn/kerberized-hadoop image comes from docker/test/integration/kerberized_hadoop/Dockerfile | 2020-12-11T10:03:47 | 2020-12-11T10:03:47 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,186,931 | 18,002 | xtreding | @filimonov it seems to work. Thank you! | 2020-12-11T13:16:29 | 2020-12-11T13:16:29 | {} | NONE |
ClickHouse | ClickHouse | 742,831,570 | 17,928 | alexey-milovidov | Bad test. | 2020-12-10T22:10:04 | 2020-12-10T22:10:04 | {} | MEMBER |
ClickHouse | ClickHouse | 743,062,503 | 15,806 | hexiaoting | This PR is moved to https://github.com/ClickHouse/ClickHouse/pull/17829 | 2020-12-11T08:53:10 | 2020-12-11T08:53:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,803,581 | 17,963 | nikitamikhaylov | Won't wait for all tests, because everything seems to be Ok. | 2020-12-10T21:14:59 | 2020-12-10T21:14:59 | {} | MEMBER |
ClickHouse | ClickHouse | 742,819,568 | 17,982 | azat | Can someone please mark with force-tests label? | 2020-12-10T21:44:41 | 2020-12-10T21:44:50 | {} | MEMBER |
ClickHouse | ClickHouse | 742,913,464 | 17,962 | allen12921 | Thanks for reply
There is still more than 70% free space there and also there is low cpu io wait
/dev/nvme1n1 825G 211G 573G 27% /var/lib/clickhouse
There is enough free space here. it seems like mutate process was blocked by merge process
We're doing batch update 4 seconds a time and we already changed the... | 2020-12-11T01:54:30 | 2020-12-11T01:54:30 | {} | NONE |
ClickHouse | ClickHouse | 743,039,805 | 17,989 | filimonov | regression | 2020-12-11T08:03:19 | 2020-12-11T08:03:19 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,195,548 | 9,404 | KrishnaPG | Looks like Clickhouse already supports `UInt256` - is it possible to support a new datetime type, such as `DateTime256` for these large range use-cases? Given that some data needs these ranges, it should be hopefully enough to fit all the long ranges needed. | 2020-12-11T13:34:50 | 2020-12-11T13:34:50 | {} | NONE |
ClickHouse | ClickHouse | 743,479,678 | 18,004 | alexey-milovidov | Nevermind, I have added it. | 2020-12-11T23:27:24 | 2020-12-11T23:27:24 | {
"+1": 1,
"heart": 1
} | MEMBER |
ClickHouse | ClickHouse | 742,607,944 | 17,960 | den-crane | What is the type of `depth` column?
What is the table `order by` ? | 2020-12-10T15:52:55 | 2020-12-10T15:52:55 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,731,941 | 17,967 | nikitamikhaylov | No test were run, but I hope everything will be Ok. | 2020-12-10T19:07:33 | 2020-12-10T19:07:33 | {} | MEMBER |
ClickHouse | ClickHouse | 742,732,509 | 17,960 | astudnev | Yes
set optimize_redundant_functions_in_order_by=0;
solved this for my query as well | 2020-12-10T19:08:39 | 2020-12-10T19:08:39 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,799,641 | 5,225 | alexey-milovidov | @sliontc Check the logs.
If it will not help, run clickhouse-server manually with the command:
```
sudo -u clickhouse /usr/bin/clickhouse server --config-file /etc/clickhouse-server/config.xml
``` | 2020-12-10T21:07:27 | 2020-12-10T21:07:27 | {} | MEMBER |
ClickHouse | ClickHouse | 742,803,903 | 17,911 | alexey-milovidov | Remove the following line:
```
cmake/warnings.cmake
27: add_warning(frame-larger-than=65536)
``` | 2020-12-10T21:15:36 | 2020-12-10T21:15:36 | {} | MEMBER |
ClickHouse | ClickHouse | 742,840,573 | 17,646 | azat | Fixed few typos and rebased | 2020-12-10T22:29:21 | 2020-12-10T22:29:21 | {} | MEMBER |
ClickHouse | ClickHouse | 742,849,063 | 17,738 | alexey-milovidov | @filimonov Could you please add this info to the docs?
PS. There are no "obvious" things that everyone knows. We can add both examples: with port mapping, with network=host and with linked containers.
Documentation should work even for users who are not familiar with Docker. | 2020-12-10T22:48:46 | 2020-12-10T22:49:22 | {} | MEMBER |
ClickHouse | ClickHouse | 742,975,268 | 16,860 | dikmu | I met the same problem,any progress？ | 2020-12-11T05:18:26 | 2020-12-11T05:18:26 | {} | NONE |
ClickHouse | ClickHouse | 743,092,544 | 17,957 | chenkaiyin1201 | > This exception says that the replica has unprocessed alters. If the error doesn't disappear, then provide server logs table schema and query.
The problem has been dealt with. There should be an abnormal outage state that causes inconsistency between the version of ZK node metadata and the metadata_version node dat... | 2020-12-11T09:48:43 | 2020-12-11T09:48:43 | {} | NONE |
ClickHouse | ClickHouse | 743,108,084 | 17,623 | qoega | Infinite deduplication. We can store in separate column(s) data about deduplication ids. We should not store them in ZooKeeper. There should be no risk of duplicate inserts disregard qps and time between insert retries. | 2020-12-11T10:18:46 | 2020-12-11T10:18:46 | {} | MEMBER |
ClickHouse | ClickHouse | 743,232,333 | 18,006 | den-crane | >I presume this option is new/experimental and will be ON by default in the future?
>Can this option be turned ON for the whole server, or will it break some other requests?
Two reasons.
1. this option is new/experimental
2. some queries work much slower with this setting enabled | 2020-12-11T14:43:56 | 2020-12-11T14:44:34 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,832,182 | 17,847 | fibersel | @alexey-milovidov can you suggest any example, that I can use to implement testing framework?
I saw `utils/compressor/decompress_perf.cpp`, but it seems to be too complicated for my case. | 2020-12-10T22:11:25 | 2020-12-10T22:11:25 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,847,462 | 17,813 | alexey-milovidov | @qoega The proposed modification is wrong. https://github.com/ClickHouse/ClickHouse/pull/14453/files#diff-76d6bd025bb3c3e3e103da2943499b9469f7e79ee47814125cbfed7356017a55R12 | 2020-12-10T22:45:01 | 2020-12-10T22:45:01 | {} | MEMBER |
ClickHouse | ClickHouse | 743,234,398 | 9,404 | den-crane | @KrishnaPG DateTime64 is enough to support large ranges even with nanosecond precision. But this PR is not merged yet. | 2020-12-11T14:47:33 | 2020-12-11T14:47:33 | {
"+1": 2
} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,240,036 | 18,006 | hrissan | Can you give an example of query which will work much slower with this setting enabled, so we do not accidentally design tables in a way we later need a query which is prohibitively slow both WITH and WITHOUT this option) | 2020-12-11T14:57:05 | 2020-12-11T14:57:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,484,974 | 17,881 | alexey-milovidov | Changes LGTM but there is "Docs check" error (most likely a broken link). | 2020-12-11T23:47:26 | 2020-12-11T23:47:26 | {} | MEMBER |
ClickHouse | ClickHouse | 742,599,814 | 17,960 | astudnev | ```
SELECT
depth,
SUM(coinpath.txs_ethereum_outbound_4c4dce9598b8e2addcc99.amount) / dictGet('currency', 'divider', toUInt64(1)) AS `_aq.amount`
FROM coinpath.txs_ethereum_outbound_4c4dce9598b8e2addcc99
WHERE (coinpath.txs_ethereum_outbound_4c4dce9598b8e2addcc99.depth <= 3) AND (coinpath.txs_ethereum_outb... | 2020-12-10T15:39:36 | 2020-12-10T15:39:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,799,117 | 14,042 | alexey-milovidov | @sliontc Check the logs.
If it will not help, run clickhouse-server manually with the command:
```
sudo -u clickhouse /usr/bin/clickhouse server --config-file /etc/clickhouse-server/config.xml
``` | 2020-12-10T21:06:22 | 2020-12-10T21:06:22 | {} | MEMBER |
ClickHouse | ClickHouse | 742,831,593 | 17,930 | alexey-milovidov | Bad test. | 2020-12-10T22:10:08 | 2020-12-10T22:10:08 | {} | MEMBER |
ClickHouse | ClickHouse | 743,063,279 | 17,913 | KochetovNicolai | `max_threads` setting uses the number of physical cores by default, and it is usually an optimal value.
`max_final_threads` does not use the number of physical cores because it is not as optimal. When I tested parallel final on 32 cores, values between 8 and 16 were the most optimal for many cases.
In case if `max_fi... | 2020-12-11T08:54:39 | 2020-12-11T08:54:39 | {} | MEMBER |
ClickHouse | ClickHouse | 743,109,366 | 17,237 | filimonov | workaround to have old (a bit incorrect behavior) is to strip nullness from the type, like that.
```
select toTypeName( sumState(id) ) from x; /* old version return AggregateFunction(sum, UInt8), new version return AggregateFunction(sum, Nullable(UInt8)) */
select toTypeName( sumState(assumeNotNull(id)) ) from x; ... | 2020-12-11T10:21:28 | 2020-12-11T10:21:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,178,743 | 18,002 | filimonov | https://clickhouse.tech/docs/en/interfaces/http/#response-buffering
Try using `wait_end_of_query=1` | 2020-12-11T12:59:11 | 2020-12-11T12:59:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,388,423 | 17,641 | nikitamikhaylov | @akuzm Everything works fine how I think. Maybe we can merge after integration tests passed and style fixed? | 2020-12-11T19:38:40 | 2020-12-11T19:38:40 | {} | MEMBER |
ClickHouse | ClickHouse | 742,834,537 | 17,847 | alexey-milovidov | @fibersel You can use `clickhouse-compressor` tool. | 2020-12-10T22:16:18 | 2020-12-10T22:16:18 | {} | MEMBER |
ClickHouse | ClickHouse | 742,965,794 | 17,748 | azat | >It's weird that we don't have the Partial Queries table with queries with new functions. I'll look into it later.
@akuzm Finally performance tests passed and has new functions in Partial Queries table. | 2020-12-11T04:46:17 | 2020-12-11T04:46:17 | {} | MEMBER |
ClickHouse | ClickHouse | 743,060,421 | 17,646 | azat | >For me it sounds like flush from buffer should happen in context/profile of another user (system_profile?).
There is direct flush and background, direct is done in context of current query, yes this is tricky to track all those flushes, but this may be useful, and at least if buffer is full, then the query should b... | 2020-12-11T08:48:19 | 2020-12-11T08:48:19 | {} | MEMBER |
ClickHouse | ClickHouse | 743,084,771 | 17,962 | chenk818 | how many datas will be insert&updat per 4 seconds?
if not so many,you can do it within one insert/update or else use dictionary table | 2020-12-11T09:33:23 | 2020-12-11T09:33:23 | {} | NONE |
ClickHouse | ClickHouse | 743,211,209 | 18,006 | den-crane | 1. optimize_aggregation_in_order -- disabled by default
2. no optimization for this case. SELECT k, max(v) AS max_v FROM example GROUP BY k HAVING max_v == 100
Ch builds all result k, max(v) GROUP BY k and after that filters max_v == 100 | 2020-12-11T14:06:27 | 2020-12-11T14:06:27 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,287,527 | 17,999 | den-crane | Yes. What API do you use? JDBC ?
| 2020-12-11T16:18:58 | 2020-12-11T16:19:36 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,620,218 | 17,960 | astudnev | CREATE TABLE coinpath.txs_ethereum_outbound_4c4dce9598b8e2addcc99
(
`depth` UInt32,
`currency_id` UInt32,
`block` UInt32,
`tx_time` DateTime,
`tx_hash_bin` String,
`transfer_from_bin` String,
`transfer_to_bin` String,
`tx_value` Float64,
`amount` Float64
)
ENGINE = TinyLo... | 2020-12-10T16:12:48 | 2020-12-10T16:12:48 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,720,286 | 17,960 | CurtizJ | Should be fixed in #17471. Just need to backport to 20.8.
Workaround for now: `set optimize_redundant_functions_in_order_by = 0`
```sql
EXPLAIN SYNTAX
SELECT
d,
SUM(am) / dictGet('default.currency', 'divider', toUInt64(1)) AS a
FROM
(
SELECT
toUInt32(1) AS d,
'1' AS f,
... | 2020-12-10T18:48:20 | 2020-12-10T18:48:20 | {} | MEMBER |
ClickHouse | ClickHouse | 742,970,053 | 17,215 | coreyralph | Here's a minimal setup to reproduce the issue. The problem only happens when using the (undocumented) UPDATE_FIELD option in CREATE DICTIONARY.
My test setup is Debian 10, postgres 12.5-1.pgdg100+1, clickhouse 20.10.4.1. Also tested on 20.13.1.5343-testing, same issue. Data in the source table seems irrelevant, the ... | 2020-12-11T05:01:27 | 2020-12-11T05:01:27 | {} | NONE |
ClickHouse | ClickHouse | 742,991,828 | 17,623 | chang0391 | Suggest this function which is when a new node is added, all distributed tables can be rebalanced through one command,this will make the resource utilization of each node very balanced and user operation will be more convenient | 2020-12-11T06:04:33 | 2020-12-11T06:04:33 | {
"+1": 2
} | NONE |
ClickHouse | ClickHouse | 743,214,521 | 17,905 | akuzm | I don't want to backport it, because backporting changes in Poco is cumbersome, and there is a workaround -- set size using `M` suffix. | 2020-12-11T14:12:24 | 2020-12-11T14:12:24 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,278,180 | 18,005 | filimonov | See also #15816 | 2020-12-11T16:01:11 | 2020-12-11T16:01:11 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,297,748 | 9,404 | ToucheSir | @KrishnaPG I'm pretty certain you're trying to load MIMIC data into ClickHouse, so try this transform:
```sql
concat(toString(toInt16OrNull(substring(<date column>, 1, 4))-120), substring(<date column>, 5, 15))
```
Since all the dates are shifted by a fixed offset (120 years), it's trivial to restore original dates... | 2020-12-11T16:37:05 | 2020-12-11T16:37:05 | {} | NONE |
ClickHouse | ClickHouse | 743,440,839 | 18,015 | den-crane | Now you should use .inner.MV name in cascades.
https://den-crane.github.io/Everything_you_should_know_about_materialized_views_commented.pdf page18
https://youtu.be/ckChUkC3Pns?list=PLO3lfQbpDVI-hyw4MyqxEk3rDHw95SzxJ&t=10890 | 2020-12-11T21:40:05 | 2020-12-11T21:40:05 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,477,716 | 18,009 | alexey-milovidov | LGTM
PS. I don't know any ways to add a test for it - Ok. | 2020-12-11T23:22:16 | 2020-12-11T23:22:16 | {} | MEMBER |
ClickHouse | ClickHouse | 742,600,179 | 17,960 | astudnev | ```
SELECT
depth,
SUM(coinpath.txs_ethereum_outbound_4c4dce9598b8e2addcc99.amount) / dictGet('currency', 'divider', toUInt64(1)) AS `_aq.amount`
FROM coinpath.txs_ethereum_outbound_4c4dce9598b8e2addcc99
WHERE (coinpath.txs_ethereum_outbound_4c4dce9598b8e2addcc99.depth <= 3) AND (coinpath.txs_ethereum_outb... | 2020-12-10T15:40:10 | 2020-12-10T15:40:10 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,639,227 | 17,960 | astudnev | [test.native.gz](https://github.com/ClickHouse/ClickHouse/files/5674284/test.native.gz)
| 2020-12-10T16:41:31 | 2020-12-10T16:41:31 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,804,766 | 17,884 | alexey-milovidov | @l1t1 ClickHouse can be run without sudo. See the comment from den-crane above. | 2020-12-10T21:17:17 | 2020-12-10T21:17:17 | {} | MEMBER |
ClickHouse | ClickHouse | 742,831,317 | 17,895 | alexey-milovidov | @Enmk @KochetovNicolai I have reverted. Please make another try. | 2020-12-10T22:09:33 | 2020-12-10T22:09:33 | {} | MEMBER |
ClickHouse | ClickHouse | 742,831,697 | 17,931 | alexey-milovidov | Bad test. | 2020-12-10T22:10:21 | 2020-12-10T22:10:21 | {} | MEMBER |
ClickHouse | ClickHouse | 742,842,402 | 17,982 | azat | Since the culprit was reverted, rebased against upstream/master | 2020-12-10T22:33:23 | 2020-12-10T22:33:23 | {} | MEMBER |
ClickHouse | ClickHouse | 743,063,546 | 17,962 | allen12921 | Probably yes, we are doing each insert&update per 4 seconds
Neither insert nor update number is fixed
It's working fine with clickhouse server V19.13.3.26
I have tried to increase lot of parameters such as: merge_max_block_size background_pool_size max_threads
Now we don't have this issue but the most popular ... | 2020-12-11T08:55:13 | 2020-12-11T09:11:56 | {} | NONE |
ClickHouse | ClickHouse | 743,164,652 | 17,960 | CurtizJ | It turned out, that it's already backported and fixed since 20.8.8. | 2020-12-11T12:26:24 | 2020-12-11T12:26:24 | {} | MEMBER |
ClickHouse | ClickHouse | 742,805,010 | 17,884 | alexey-milovidov | Example here: https://clickhouse.tech/docs/en/operations/performance-test/ | 2020-12-10T21:17:41 | 2020-12-10T21:17:41 | {} | MEMBER |
ClickHouse | ClickHouse | 742,947,614 | 17,972 | amosbird | Superceded by Query Pipeline rewrite. | 2020-12-11T03:42:46 | 2020-12-11T03:42:46 | {} | COLLABORATOR |
ClickHouse | ClickHouse | 743,302,447 | 16,860 | filimonov | Duplicate https://github.com/ClickHouse/ClickHouse/issues/14582 | 2020-12-11T16:46:01 | 2020-12-11T16:46:01 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 742,806,539 | 147 | alexey-milovidov | @shadowDy Yes. We selected the №3 way of implementation:
> Allow to modify primary key for new parts, but leave primary key in existing parts as is. Allowing to have different primary keys in different parts of table is theoretically possible, but introduce many difficulties in query execution.
Currently this f... | 2020-12-10T21:20:19 | 2020-12-10T21:20:27 | {} | MEMBER |
ClickHouse | ClickHouse | 743,076,243 | 17,160 | zhangjmruc | @alexey-milovidov Would you please asign this to me? | 2020-12-11T09:18:28 | 2020-12-11T09:18:28 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,108,113 | 17,962 | chenk818 | with V20.6，following parameters were related
max_bytes_to_merge_at_min_space_in_pool,default value is bigger enough( 150GB)
background_processing_pool_thread_sleep_seconds default value is 10s
I can't find these with V19.x | 2020-12-11T10:18:49 | 2020-12-11T10:18:49 | {} | NONE |
ClickHouse | ClickHouse | 743,113,629 | 17,993 | olgarev | @qoega Can you review this one, pls? | 2020-12-11T10:30:30 | 2020-12-11T10:30:30 | {} | CONTRIBUTOR |
ClickHouse | ClickHouse | 743,185,396 | 9,404 | KrishnaPG | Came across this while facing a problem with future dates. Health datasets use anonymization to remove PII and offset dates into future to ensure anonymity of patients. Hence it is important to be able to insert large future dates. It is surprising that 64bit is not enough just for year 2193, 2210 etc. Here are the dat... | 2020-12-11T13:13:20 | 2020-12-11T13:13:20 | {} | NONE |
ClickHouse | ClickHouse | 750,332,138 | 18,312 | alesapin | Coverage report https://clickhouse-test-reports.s3.yandex.net/18312/cc0c43ace53fe94c599a3ba9ec9417e184f7f15c/coverage_report/test_output/index.html,
But takes about 05h:21m:52s to prepare. Trying to build contrib in release mode. | 2020-12-23T14:47:08 | 2020-12-23T14:47:08 | {} | MEMBER |
ClickHouse | ClickHouse | 750,466,137 | 18,443 | azat | >Upd. No we cannot do it, because there is Arcadia build.
Ok, let's guard displayText with POCO_CLICKHOUSE_PATCH then | 2020-12-23T20:42:34 | 2020-12-23T20:42:34 | {} | MEMBER |
ClickHouse | ClickHouse | 750,609,501 | 15,811 | nikitamikhaylov | I checkouted 20.10 locally, built with ASan and run `test_multiple_disks/test.py` multiple times. And it passed in each execution! I think this test is flaky. Maybe @excitoon can help. | 2020-12-24T00:07:12 | 2020-12-24T00:07:12 | {} | MEMBER |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.