Skip to content

Fix accuracy of uniqTheta when using UInt8 aggregation keys in parallel#94095

Merged
azat merged 2 commits intoClickHouse:masterfrom
azat:uniqTheta-u8-accuracy-fix
Jan 14, 2026
Merged

Fix accuracy of uniqTheta when using UInt8 aggregation keys in parallel#94095
azat merged 2 commits intoClickHouse:masterfrom
azat:uniqTheta-u8-accuracy-fix

Conversation

@azat
Copy link
Copy Markdown
Member

@azat azat commented Jan 13, 2026 •
edited
Loading

Changelog category (leave one):

  • Bug Fix (user-visible misbehavior in an official stable release)

Changelog entry (a user-readable short description of the changes that goes into CHANGELOG.md):

Fix accuracy of uniqTheta when using UInt8 aggregation keys in parallel (max_threads > 1 - default)

Test

SELECT
    number % 3 AS key,
    uniqTheta(generateUUIDv4(number)) AS theta,
    uniqCombined64(generateUUIDv4(number)) AS hll
FROM numbers_mt(9160000)
GROUP BY key
SETTINGS max_threads = 8

Before

   ┌─key─┬───theta─┬─────hll─┐
1. │   0 │ 3006052 │ 3051475 │
2. │   1 │ 3077663 │ 3057512 │
3. │   2 │  730659 │ 3049931 │
   └─────┴─────────┴─────────┘

After

   ┌─key─┬───theta─┬─────hll─┐
1. │   0 │ 3068206 │ 3054168 │ -- 3.05 million
2. │   1 │ 3029186 │ 3049882 │ -- 3.05 million
3. │   2 │ 3055119 │ 3047556 │ -- 3.05 million
   └─────┴─────────┴─────────┘

Fixes: #45292

@azat azat requested a review from Avogar January 13, 2026 19:53
@clickhouse-gh
Copy link
Copy Markdown
Contributor

clickhouse-gh bot commented Jan 13, 2026 •
edited
Loading

Workflow [PR], commit [0f1d705]

Summary: ✅

@clickhouse-gh clickhouse-gh bot added the pr-bugfix Pull request with bugfix, not backported by default label Jan 13, 2026
@Avogar Avogar self-assigned this Jan 13, 2026
@azat azat mentioned this pull request Jan 13, 2026
Closed
@azat azat force-pushed the uniqTheta-u8-accuracy-fix branch from 125136f to f3afd15 Compare January 13, 2026 20:25
azat
azat commented Jan 14, 2026
View reviewed changes
tests/queries/0_stateless/01798_uniq_theta_sketch.reference
17 10 10 100 100 610 610 766
52 10 10 100 100 608 608 766
5 10 10 100 100 608 608 765
5 10 10 100 100 609 609 765
Copy link
Copy Markdown
Member Author

@azat azat Jan 14, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Note, this is more accurate value, it matches with uniqExact

@azat azat added the pr-must-backport Pull request should be backported intentionally. Use this label with great care! label Jan 14, 2026
@azat azat enabled auto-merge January 14, 2026 09:53
@azat
tests: update 01798_uniq_theta_sketch reference
0f1d705 
This is the correct value, since uniqExact gives the same (last values
are for uniqExact):

    5      10      10      100     100     609     609     10      10      100     100     609     609     765
@azat azat force-pushed the uniqTheta-u8-accuracy-fix branch from 5310fa8 to 0f1d705 Compare January 14, 2026 12:47
@azat azat added this pull request to the merge queue Jan 14, 2026
Merged via the queue into ClickHouse:master with commit a99f0c1 Jan 14, 2026
132 checks passed
@azat azat deleted the uniqTheta-u8-accuracy-fix branch January 14, 2026 15:57
@robot-ch-test-poll3 robot-ch-test-poll3 added the pr-must-backport-synced The `*-must-backport` labels are synced into the cloud Sync PR label Jan 14, 2026
@robot-ch-test-poll robot-ch-test-poll added the pr-synced-to-cloud The PR is synced to the cloud repo label Jan 14, 2026
This was referenced Jan 14, 2026
Closed
Merged
Merged
Merged
Merged
robot-clickhouse-ci-2 added a commit that referenced this pull request Jan 14, 2026
@robot-clickhouse-ci-2
Merge pull request #94205 from ClickHouse/cherrypick/25.12/94095
58c411f 
Cherry pick #94095 to 25.12: Fix accuracy of uniqTheta when using UInt8 aggregation keys in parallel
robot-clickhouse added a commit that referenced this pull request Jan 14, 2026
@robot-clickhouse
Backport #94095 to 25.12: Fix accuracy of uniqTheta when using UInt8 …
857ae92 
…aggregation keys in parallel
@robot-clickhouse-ci-2 robot-clickhouse-ci-2 mentioned this pull request Jan 14, 2026
Merged
robot-ch-test-poll2 added a commit that referenced this pull request Jan 14, 2026
@robot-ch-test-poll2
Merge pull request #94202 from ClickHouse/cherrypick/25.8/94095
702dae4 
Cherry pick #94095 to 25.8: Fix accuracy of uniqTheta when using UInt8 aggregation keys in parallel
robot-clickhouse added a commit that referenced this pull request Jan 14, 2026
@robot-clickhouse
Backport #94095 to 25.8: Fix accuracy of uniqTheta when using UInt8 a…
2bbb605 
…ggregation keys in parallel
@robot-ch-test-poll2 robot-ch-test-poll2 mentioned this pull request Jan 14, 2026
Merged
robot-ch-test-poll2 added a commit that referenced this pull request Jan 14, 2026
@robot-ch-test-poll2
Merge pull request #94203 from ClickHouse/cherrypick/25.10/94095
ec4cf5b 
Cherry pick #94095 to 25.10: Fix accuracy of uniqTheta when using UInt8 aggregation keys in parallel
robot-clickhouse added a commit that referenced this pull request Jan 14, 2026
@robot-clickhouse
Backport #94095 to 25.10: Fix accuracy of uniqTheta when using UInt8 …
9d97b32 
…aggregation keys in parallel
@robot-ch-test-poll2 robot-ch-test-poll2 mentioned this pull request Jan 14, 2026
Merged
robot-ch-test-poll2 added a commit that referenced this pull request Jan 14, 2026
@robot-ch-test-poll2
Merge pull request #94204 from ClickHouse/cherrypick/25.11/94095
79f2d2a 
Cherry pick #94095 to 25.11: Fix accuracy of uniqTheta when using UInt8 aggregation keys in parallel
robot-clickhouse added a commit that referenced this pull request Jan 14, 2026
@robot-clickhouse
Backport #94095 to 25.11: Fix accuracy of uniqTheta when using UInt8 …
010f06d 
…aggregation keys in parallel
@robot-ch-test-poll2 robot-ch-test-poll2 mentioned this pull request Jan 14, 2026
Merged
@robot-ch-test-poll2 robot-ch-test-poll2 added the pr-backports-created Backport PRs are successfully created, it won't be processed by CI script anymore label Jan 14, 2026
clickhouse-gh bot added a commit that referenced this pull request Jan 14, 2026
@clickhouse-gh
Merge pull request #94231 from ClickHouse/backport/25.10/94095
0bfec98 
Backport #94095 to 25.10: Fix accuracy of uniqTheta when using UInt8 aggregation keys in parallel
azat added a commit that referenced this pull request Jan 14, 2026
@azat
Merge pull request #94232 from ClickHouse/backport/25.11/94095
235b6ac 
Backport #94095 to 25.11: Fix accuracy of uniqTheta when using UInt8 aggregation keys in parallel
azat added a commit that referenced this pull request Jan 14, 2026
@azat
Merge pull request #94206 from ClickHouse/backport/25.12/94095
fdf280f 
Backport #94095 to 25.12: Fix accuracy of uniqTheta when using UInt8 aggregation keys in parallel
clickhouse-gh bot added a commit that referenced this pull request Jan 14, 2026
@clickhouse-gh
Merge pull request #94230 from ClickHouse/backport/25.8/94095
3390752 
Backport #94095 to 25.8: Fix accuracy of uniqTheta when using UInt8 aggregation keys in parallel
@CarlosFelipeOR CarlosFelipeOR mentioned this pull request Jan 30, 2026
Closed
@alsugiliazova
Copy link
Copy Markdown
Contributor

alsugiliazova commented Feb 2, 2026

This pr also fixed this simpler example:

SELECT number % 2 AS even, uniqTheta(number) FROM numbers(10) GROUP BY even;

After this pr:
https://fiddle.clickhouse.com/5aae58e4-a03d-4706-a52a-e2b8c61e5c5a

0	5
1	5

Before:
https://fiddle.clickhouse.com/2ce3da8c-b9ce-4d02-8484-351dc81c2f7d

0	4
1	4

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@Avogar Avogar Avogar approved these changes

Assignees

@Avogar Avogar

Labels

pr-backports-created Backport PRs are successfully created, it won't be processed by CI script anymore pr-bugfix Pull request with bugfix, not backported by default pr-must-backport Pull request should be backported intentionally. Use this label with great care! pr-must-backport-synced The `*-must-backport` labels are synced into the cloud Sync PR pr-synced-to-cloud The PR is synced to the cloud repo

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

uniqTheta produces random results with multithreaded streams

6 participants

@azat @alsugiliazova @Avogar @robot-ch-test-poll @robot-ch-test-poll2 @robot-ch-test-poll3