branch-4.0: [feature](statistics) Skip collecting stats for long string columns #62686#63302
Open
yujun777 wants to merge 1 commit into
Open
branch-4.0: [feature](statistics) Skip collecting stats for long string columns #62686#63302yujun777 wants to merge 1 commit into
yujun777 wants to merge 1 commit into
Conversation
…pache#62686) Issue Number: close #0 Problem Summary: When a string column contains extremely long values, running ANALYZE on it can consume excessive memory and may even OOM the BE. This PR introduces an inline per-row guard that short-circuits stats collection for any string column whose rows exceed `Config.statistics_max_string_column_length` (default 1024 bytes). The guard is injected into every statistics-collection SQL template (FULL_ANALYZE_TEMPLATE, LINEAR_ANALYZE_TEMPLATE, DUJ1_ANALYZE_TEMPLATE, HMS full/sample) as a `NoneMovableFunction`: assert_true(col IS NULL OR LENGTH(col) <= N, 'ANALYZE_SKIP_LONG_STRING_COLUMN') AS __lc When any row violates the predicate, `assert_true` triggers a controlled error carrying the `ANALYZE_SKIP_LONG_STRING_COLUMN` marker. `BaseAnalysisTask` walks the exception cause chain for this marker and, when detected, converts the failure into a FINISHED task with a skip message so users see why the column was dropped (via SHOW ANALYZE and via OK-packet info for sync analyze). Other columns in the same job are unaffected. Setting `statistics_max_string_column_length = 0` disables the guard entirely (upstream behavior). Added FE config `statistics_max_string_column_length` (default 1024). String columns whose row byte length exceeds this value are skipped during ANALYZE with a visible skip message, instead of risking OOM. Set to 0 to disable. --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> (cherry picked from commit 34b5c6c)
Contributor
Author
|
run buildall |
Contributor
|
Thank you for your contribution to Apache Doris. Please clearly describe your PR:
|
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.
cherry-pick: #62686