Yukang-Lian opened a new pull request, #67723:
URL: https://github.com/apache/doris/pull/67723

   ### What problem does this PR solve?
   
   Issue Number: N/A
   
   Related PR: N/A
   
   Problem Summary:
   
   `fault_injection_p0/test_fix_tablet_stat_fault_injection` triggers full 
compaction on 1000 tablets of one MoW table and compares `SHOW DATA` against a 
golden that assumes every compaction ran exactly once (each run injects a fixed 
size delta).
   
   In cloud mode the HTTP compaction trigger is asynchronous, and 
`trigger_and_wait_compaction` returns once every tablet reports a completion 
timestamp, no matter whether the compaction succeeded or failed. Under 1000 
concurrent full compactions on one table, all of them contend on the same 
delete bitmap lock key in meta-service; occasionally one request exhausts the 
meta-service txn conflict retry budget and the compaction fails with 
`[DELETE_BITMAP_LOCK_ERROR]txn conflict when get delete bitmap update lock`. 
That tablet silently keeps its 6 rowsets and misses one injected delta, so the 
case fails at the golden check:
   
   ```
   Check tag 'select_2' failed
   ExpectRow: [test_fix_tablet_stat_fault_injection, ..., 9.314 GB, 1000, 100, 
0.000 ]
   RealRow  : [test_fix_tablet_stat_fault_injection, ..., 9.304 GB, 1000, 100, 
0.000 ]
   ```
   
   Changes:
   
   - `trigger_and_wait_compaction`: add an optional `tablet_ids` argument 
(default empty, existing callers unchanged) so a suite can trigger and wait on 
a subset of tablets.
   - `test_fix_tablet_stat_fault_injection`: after the first round, re-trigger 
only the tablets that still hold more than 2 rowsets, at most 5 rounds, then 
assert that every tablet is compacted (the previously commented `rowsetCount` 
assertion is restored) before comparing `SHOW DATA`. A remaining failure now 
reports the tablet ids and their `last full status` instead of an opaque size 
mismatch.
   
   The lock conflict itself is a transient, recoverable failure (production 
compaction is rescheduled by the compaction scheduler); this PR only makes the 
case tolerate it without touching the golden.
   
   ### Release note
   
   None
   
   ### Check List (For Author)
   
   - Test <!-- At least one of them must be included. -->
       - [x] Regression test
       - [ ] Unit Test
       - [ ] Manual test (add detailed scripts or steps below)
       - [ ] No need to test or manual test. Explain why:
           - [ ] This is a refactor/code format and no logic has been changed.
           - [ ] Previous test can cover this change.
           - [ ] No code files have been changed.
           - [ ] Other reason <!-- Add your reason?  -->
   
   - Behavior changed:
       - [x] No.
       - [ ] Yes. <!-- Explain the behavior change -->
   
   - Does this need documentation?
       - [x] No.
       - [ ] Yes. <!-- Add document PR link here. eg: 
https://github.com/apache/doris-website/pull/1214 -->
   
   ### Check List (For Reviewer who merge this PR)
   
   - [ ] Confirm the release note
   - [ ] Confirm test cases
   - [ ] Confirm document
   - [ ] Add branch pick label <!-- Add branch pick label that this PR should 
merge into -->
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to