Fix readByLayers using wrong ReadType for reverse-key tables#101153
Open
alexey-milovidov wants to merge 7 commits intomasterfrom
Open
Fix readByLayers using wrong ReadType for reverse-key tables#101153alexey-milovidov wants to merge 7 commits intomasterfrom
readByLayers using wrong ReadType for reverse-key tables#101153alexey-milovidov wants to merge 7 commits intomasterfrom
Conversation
`readByLayers` hardcoded `ReadType::InOrder` when reading data for the sharded join optimization. For tables with descending primary keys (`allow_experimental_reverse_key = 1`), the correct type should be `ReadType::InReverseOrder` when `input_order_info->direction == -1`. This caused flaky failures in `03668_shard_join_in_reverse_order` when randomized settings enabled both `read_in_order_use_virtual_row` and disabled `enable_join_runtime_filters`, producing wrong output order. The fix mirrors the logic already used in `spreadMarkRangesAmongStreamsWithOrder`. Closes #100870 Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
Member
Author
|
Add a test. |
Contributor
|
Workflow [PR], commit [70f8e23] Summary: ❌
AI ReviewSummaryThis PR fixes PR Metadata
Findings
ClickHouse Rules
Final Verdict
|
clickhouse-gh
bot
added
the
pr-bugfix
![clickhouse-gh[bot]](https://avatars.githubusercontent.com/in/928874?s=60&v=4){kind=small}
| sort_description.emplace_back(sorting_columns[i], input_order_info->direction); | ||
| } | ||
|
|
||
| const auto read_type = input_order_info->direction == 1 ? ReadType::InOrder : ReadType::InReverseOrder; |
Contributor
There was a problem hiding this comment.
Please add the required Changelog entry in the PR description for the selected Bug Fix category (the template requires it unless category is one of the explicitly exempt ones).
Suggested text:
Fix incorrect read order in readByLayers for reverse-key MergeTree tables, which could produce unstable result ordering in shard-join queries when reading in reverse order.
Adds a deterministic test that exercises the `readByLayers` code path (via `query_plan_join_shard_by_pk_ranges`) on a table with a descending primary key (`allow_experimental_reverse_key = 1`). This verifies that the correct `ReadType` is used based on `input_order_info->direction`. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
…verse-key-read-type
The number 04068 was taken by another test on master. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>
…verse-key-read-type
…verse-key-read-type
amosbird
approved these changes
Mar 31, 2026
Contributor
LLVM Coverage Report
Changed lines: 92.86% (13/14) · Uncovered code |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
readByLayershardcodedReadType::InOrderwhen reading data for the sharded join optimization. For tables with descending primary keys (allow_experimental_reverse_key = 1), the correct type should beReadType::InReverseOrderwheninput_order_info->direction == -1.This caused flaky failures in
03668_shard_join_in_reverse_orderwhen randomized settings enabled bothread_in_order_use_virtual_rowand disabledenable_join_runtime_filters, producing wrong output order.The fix mirrors the logic already used in
spreadMarkRangesAmongStreamsWithOrder.Closes #100870
Changelog category (leave one):