Uh oh!
There was an error while loading. Please reload this page.
[fix](cloud) Fill schema change version holes before running - #63443
Merged
gavinchou merged 1 commit intoMay 21, 2026
Conversation
hello-stephen
commented
May 20, 2026
Contributor
Thank you for your contribution to Apache Doris. Please clearly describe your PR:
|
liaoxin01force-pushed
the
fix/cloud-schema-change-fill-holes
branch
5 times, most recently
from
May 20, 2026 15:02
5a3915d to
78af3fbCompareliaoxin01
commented
May 20, 2026
ContributorAuthor
run buildall |
### What problem does this PR solve? Issue Number: None Related PR: None Problem Summary: During cloud schema change, empty rowsets created while calculating delete bitmap were only added to a temporary tablet. The real new tablet could become RUNNING with a local version graph hole after the alter version. Subsequent delete bitmap sync on the RUNNING tablet captures old rowset ids before filling local holes, so it can fail to find a continuous version path. This change fills version holes on the real new tablet after adding schema change output rowsets and before switching it to RUNNING, preserving the existing schema change rule that skips holes at or below alter_version. Unit tests cover both the local graph repair helper and the schema change finalization path, verifying capture_consistent_versions can traverse through the filled empty rowset. ### Release note None ### Check List (For Author) - Test: Unit Test - ./run-be-ut.sh --run --filter=CloudSchemaChangeJobTest.FillVersionHolesBeforeNewTabletRunning:CloudTabletDeleteRowsetsForSchemaChangeTest.TestFillVersionHolesBeforeSchemaChangeRunning -j 8 - Behavior changed: Yes. Cloud schema change now repairs local version holes on the real new tablet before it becomes RUNNING. - Does this need documentation: No
liaoxin01force-pushed
the
fix/cloud-schema-change-fill-holes
branch
from
May 20, 2026 15:24
78af3fb to
db71d76Compareliaoxin01
commented
May 20, 2026
ContributorAuthor
run buildall |
hello-stephen
commented
May 20, 2026
Contributor
TPC-H: Total hot run time: 31200 ms |
hello-stephen
commented
May 20, 2026
Contributor
TPC-DS: Total hot run time: 170019 ms |
hello-stephen
commented
May 20, 2026
Contributor
BE Regression && UT Coverage ReportIncrement line coverage Increment coverage report
|
liaoxin01
commented
May 21, 2026
ContributorAuthor
/review |
Contributor
There was a problem hiding this comment.
Review result: no blocking issues found.
Critical checkpoint conclusions:
- Goal and tests: The change addresses the stated cloud schema-change hole by filling local versions after output rowsets are installed and before TABLET_RUNNING; the added tests cover the helper behavior and the finalization path.
- Scope: The implementation is small and focused on the existing fill_version_holes helper.
- Concurrency and locks: The new call runs while holding sync_meta_lock and the tablet header write lock, matching the surrounding finalization path and preserving the documented lock order.
- Lifecycle: No new static/global lifecycle concerns were introduced.
- Configuration and compatibility: No new config, protocol, storage-format, or mixed-version compatibility changes were introduced.
- Parallel paths: Existing sync_tablet_rowsets hole filling remains unchanged; this adds the missing schema-change finalization path on the real new tablet.
- Conditional behavior: The existing TABLET_NOTREADY/alter_version skip rule is preserved, so holes at or below alter_version are still not filled during schema change.
- Tests and results: Added BE unit tests are targeted and deterministic; I did not run them in this review environment.
- Observability: Existing fill_version_holes logging and metric remain reused.
- Transaction/persistence/data correctness: The change is local cache/meta repair after meta-service commit; empty deterministic hole rowsets are created before exposing the tablet as RUNNING, improving version-path correctness without changing committed visibility.
- Performance: The added scan/fill is bounded to schema-change finalization and reuses existing logic; no hot-path regression found.
User focus: No additional user-provided review focus was specified.
gavinchou
approved these changes
May 21, 2026
Uh oh!
There was an error while loading. Please reload this page.
Contributor
PR approved by at least one committer and no changes requested. |
Contributor
PR approved by anyone and no changes requested. |
github-actionsBot
pushed a commit
that referenced
this pull request
May 21, 2026
Problem Summary: During cloud schema change, empty rowsets created while calculating delete bitmap were only added to a temporary tablet. The real new tablet could become RUNNING with a local version graph hole after the alter version. Subsequent delete bitmap sync on the RUNNING MOW tablet captures old rowset ids before filling local holes, so it can fail to find a continuous version path. This change fills version holes on the real new tablet after adding schema change output rowsets and before switching it to RUNNING. It preserves the existing schema change rule that skips holes at or below alter_version. A unit test covers the local graph repair and verifies capture_consistent_versions can traverse through the filled empty rowset.
github-actionsBot
pushed a commit
that referenced
this pull request
May 21, 2026
Problem Summary: During cloud schema change, empty rowsets created while calculating delete bitmap were only added to a temporary tablet. The real new tablet could become RUNNING with a local version graph hole after the alter version. Subsequent delete bitmap sync on the RUNNING MOW tablet captures old rowset ids before filling local holes, so it can fail to find a continuous version path. This change fills version holes on the real new tablet after adding schema change output rowsets and before switching it to RUNNING. It preserves the existing schema change rule that skips holes at or below alter_version. A unit test covers the local graph repair and verifies capture_consistent_versions can traverse through the filled empty rowset.
Closed
zhaorongsheng pushed a commit
to zhaorongsheng/doris
that referenced
this pull request
Jun 4, 2026
…63443) Problem Summary: During cloud schema change, empty rowsets created while calculating delete bitmap were only added to a temporary tablet. The real new tablet could become RUNNING with a local version graph hole after the alter version. Subsequent delete bitmap sync on the RUNNING MOW tablet captures old rowset ids before filling local holes, so it can fail to find a continuous version path. This change fills version holes on the real new tablet after adding schema change output rowsets and before switching it to RUNNING. It preserves the existing schema change rule that skips holes at or below alter_version. A unit test covers the local graph repair and verifies capture_consistent_versions can traverse through the filled empty rowset.
Closed
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 freeto join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
What problem does this PR solve?
Issue Number: None
Related PR: None
Problem Summary: During cloud schema change, empty rowsets created while calculating delete bitmap were only added to a temporary tablet. The real new tablet could become RUNNING with a local version graph hole after the alter version. Subsequent delete bitmap sync on the RUNNING MOW tablet captures old rowset ids before filling local holes, so it can fail to find a continuous version path.
This change fills version holes on the real new tablet after adding schema change output rowsets and before switching it to RUNNING. It preserves the existing schema change rule that skips holes at or below alter_version. A unit test covers the local graph repair and verifies capture_consistent_versions can traverse through the filled empty rowset.
Release note
None
Check List (For Author)