[Fix](cloud-mow) Check partition's version to avoid wrongly update visible versions' delete bitmaps #49710
Conversation
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
Contributor
|
Thank you for your contribution to Apache Doris. Please clearly describe your PR:
|
bobhan1
force-pushed
the
check-partition-ver-when-update-delete-bitmap-o-ms
branch
3 times, most recently
from
63e9286 to
7b7c1f6
Compare
bobhan1
force-pushed
the
check-partition-ver-when-update-delete-bitmap-o-ms
branch
from
72706d6 to
e748596
Compare
Contributor
Author
|
run buildall |
1 similar comment
Contributor
Author
|
run buildall |
|
TeamCity cloud ut coverage result: |
TPC-H: Total hot run time: 35348 ms |
Contributor
Author
|
run buildall |
|
TeamCity cloud ut coverage result: |
Contributor
Author
|
run cloud_p0 |
BE UT Coverage ReportIncrement line coverage Increment coverage report
|
bobhan1
force-pushed
the
check-partition-ver-when-update-delete-bitmap-o-ms
branch
3 times, most recently
from
0fd617e to
c196ca2
Compare
Contributor
Author
|
run buildall |
|
TeamCity cloud ut coverage result: |
TPC-H: Total hot run time: 34223 ms |
TPC-DS: Total hot run time: 194016 ms |
ClickBench: Total hot run time: 31.8 s |
Contributor
Author
|
run buildall |
1 similar comment
Contributor
Author
|
run buildall |
|
TeamCity cloud ut coverage result: |
|
TeamCity cloud ut coverage result: |
TPC-H: Total hot run time: 34201 ms |
TPC-DS: Total hot run time: 193840 ms |
ClickBench: Total hot run time: 31.64 s |
BE UT Coverage ReportIncrement line coverage Increment coverage report
|
Contributor
Author
|
run feut |
github-actions
bot
added
the
approved
Contributor
|
PR approved by at least one committer and no changes requested. |
Contributor
|
PR approved by anyone and no changes requested. |
gavinchou
approved these changes
Apr 3, 2025
bobhan1
added a commit
to bobhan1/doris
that referenced
this pull request
Apr 3, 2025
…sible versions' delete bitmaps (apache#49710) considering the following problem: 1. Transaction X acquires the lock and attempts to publish with version a. This task is sent to the BE. At this point, the tablet's maximum version is a-1, and task (1) starts computation. 2. Transaction X fails on FE due to timeout and releases the lock. 3. Transaction Y acquires the lock, attempts to publish with version a, and succeeds. 4. Transaction X retries and acquires the lock again, and attempts to publish with version b. 5. Meanwhile, task (1) from Transaction X completes its computation on BE and writes the generated delete bitmap to the MS with version a. **Since Transaction X currently holds the lock, this write operation succeeds, overwriting the delete bitmaps written of actual version a by Transaction Y.** 6. Subsequent transactions on the tablet will use the pending delete bitmap to delete the version a delete bitmap written by task (1) in the MS. The root cause is that when a load txn retries in publish phase, the locks it gains are different, but they are the same in the current implementation because they have the same lock_id and initiator. This PR checks target partition's version when update delete bitmaps to avoid this problem.
bobhan1
added a commit
to bobhan1/doris
that referenced
this pull request
Apr 3, 2025
…sible versions' delete bitmaps (apache#49710) considering the following problem: 1. Transaction X acquires the lock and attempts to publish with version a. This task is sent to the BE. At this point, the tablet's maximum version is a-1, and task (1) starts computation. 2. Transaction X fails on FE due to timeout and releases the lock. 3. Transaction Y acquires the lock, attempts to publish with version a, and succeeds. 4. Transaction X retries and acquires the lock again, and attempts to publish with version b. 5. Meanwhile, task (1) from Transaction X completes its computation on BE and writes the generated delete bitmap to the MS with version a. **Since Transaction X currently holds the lock, this write operation succeeds, overwriting the delete bitmaps written of actual version a by Transaction Y.** 6. Subsequent transactions on the tablet will use the pending delete bitmap to delete the version a delete bitmap written by task (1) in the MS. The root cause is that when a load txn retries in publish phase, the locks it gains are different, but they are the same in the current implementation because they have the same lock_id and initiator. This PR checks target partition's version when update delete bitmaps to avoid this problem.
bobhan1
added a commit
to bobhan1/doris
that referenced
this pull request
Apr 3, 2025
…sible versions' delete bitmaps (apache#49710) considering the following problem: 1. Transaction X acquires the lock and attempts to publish with version a. This task is sent to the BE. At this point, the tablet's maximum version is a-1, and task (1) starts computation. 2. Transaction X fails on FE due to timeout and releases the lock. 3. Transaction Y acquires the lock, attempts to publish with version a, and succeeds. 4. Transaction X retries and acquires the lock again, and attempts to publish with version b. 5. Meanwhile, task (1) from Transaction X completes its computation on BE and writes the generated delete bitmap to the MS with version a. **Since Transaction X currently holds the lock, this write operation succeeds, overwriting the delete bitmaps written of actual version a by Transaction Y.** 6. Subsequent transactions on the tablet will use the pending delete bitmap to delete the version a delete bitmap written by task (1) in the MS. The root cause is that when a load txn retries in publish phase, the locks it gains are different, but they are the same in the current implementation because they have the same lock_id and initiator. This PR checks target partition's version when update delete bitmaps to avoid this problem.
bobhan1
added a commit
to bobhan1/doris
that referenced
this pull request
Apr 3, 2025
…sible versions' delete bitmaps (apache#49710) considering the following problem: 1. Transaction X acquires the lock and attempts to publish with version a. This task is sent to the BE. At this point, the tablet's maximum version is a-1, and task (1) starts computation. 2. Transaction X fails on FE due to timeout and releases the lock. 3. Transaction Y acquires the lock, attempts to publish with version a, and succeeds. 4. Transaction X retries and acquires the lock again, and attempts to publish with version b. 5. Meanwhile, task (1) from Transaction X completes its computation on BE and writes the generated delete bitmap to the MS with version a. **Since Transaction X currently holds the lock, this write operation succeeds, overwriting the delete bitmaps written of actual version a by Transaction Y.** 6. Subsequent transactions on the tablet will use the pending delete bitmap to delete the version a delete bitmap written by task (1) in the MS. The root cause is that when a load txn retries in publish phase, the locks it gains are different, but they are the same in the current implementation because they have the same lock_id and initiator. This PR checks target partition's version when update delete bitmaps to avoid this problem.
Closed
koarz
pushed a commit
to koarz/doris
that referenced
this pull request
Jun 4, 2025
…sible versions' delete bitmaps (apache#49710) ### What problem does this PR solve? considering the following problem: 1. Transaction X acquires the lock and attempts to publish with version a. This task is sent to the BE. At this point, the tablet's maximum version is a-1, and task (1) starts computation. 2. Transaction X fails on FE due to timeout and releases the lock. 3. Transaction Y acquires the lock, attempts to publish with version a, and succeeds. 4. Transaction X retries and acquires the lock again, and attempts to publish with version b. 5. Meanwhile, task (1) from Transaction X completes its computation on BE and writes the generated delete bitmap to the MS with version a. **Since Transaction X currently holds the lock, this write operation succeeds, overwriting the delete bitmaps written of actual version a by Transaction Y.** 6. Subsequent transactions on the tablet will use the pending delete bitmap to delete the version a delete bitmap written by task (1) in the MS. The root cause is that when a load txn retries in publish phase, the locks it gains are different, but they are the same in the current implementation because they have the same lock_id and initiator. This PR checks target partition's version when update delete bitmaps to avoid this problem.
16 tasks
dataroaring
pushed a commit
that referenced
this pull request
Jul 7, 2025
… bitmap response regardless of status code (#52547) ### What problem does this PR solve? #49710 add a check in MS to forbid stale calc delete bitmap task to wrongly update delete bitmaps in MS. But this may lead to load fail due to the check on FE. This PR let FE retry to commit the txn when encounter stale calc delete bitmap response regardless of task's status code to avoid the problem.
bobhan1
added a commit
to bobhan1/doris
that referenced
this pull request
Jul 7, 2025
… bitmap response regardless of status code (apache#52547) apache#49710 add a check in MS to forbid stale calc delete bitmap task to wrongly update delete bitmaps in MS. But this may lead to load fail due to the check on FE. This PR let FE retry to commit the txn when encounter stale calc delete bitmap response regardless of task's status code to avoid the problem.
bobhan1
added a commit
to bobhan1/doris
that referenced
this pull request
Jul 7, 2025
… bitmap response regardless of status code (apache#52547) apache#49710 add a check in MS to forbid stale calc delete bitmap task to wrongly update delete bitmaps in MS. But this may lead to load fail due to the check on FE. This PR let FE retry to commit the txn when encounter stale calc delete bitmap response regardless of task's status code to avoid the problem.
bobhan1
added a commit
to bobhan1/doris
that referenced
this pull request
Jul 8, 2025
… bitmap response regardless of status code (apache#52547) apache#49710 add a check in MS to forbid stale calc delete bitmap task to wrongly update delete bitmaps in MS. But this may lead to load fail due to the check on FE. This PR let FE retry to commit the txn when encounter stale calc delete bitmap response regardless of task's status code to avoid the problem.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
6 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.
What problem does this PR solve?
considering the following problem:
The root cause is that when a load txn retries in publish phase, the locks it gains are different, but they are the same in the current implementation because they have the same lock_id and initiator.
This PR checks target partition's version when update delete bitmaps to avoid this problem.
Release note
None
Check List (For Author)
Test
Behavior changed:
Does this need documentation?
Check List (For Reviewer who merge this PR)