cheeeee commented on PR #24072:
URL: https://github.com/apache/kafka/pull/24072#issuecomment-6580149258
Test notes for 1bc314cf28 (base 8aa57006ff), JDK 17.
Unit tests (`LogCleanerTest`):
* `testCommittedTombstoneAfterAbortedTransactionBeforeCleaningStart`
fails without the fix: `expected: <[3, 4, 5, 8]> but was: <[2, 3, 4, 5,
8]>`
(the deleted key 2 is back).
* `testBuildOffsetMapSkipsAbortedTransactionSpanningCleaningStart` fails
if the filter is `firstOffset() >= start` instead of
`lastOffset() >= start`: `expected: <-1> but was: <3>`.
* `LogCleanerTest`, `LogCleanerManagerTest`, `LogCleanerIntegrationTest`:
121 tests pass.
* `./gradlew :storage:test`, run 1: 1357 tests, 1 failed:
`TransactionsWithTieredStoreTest.testClassicBumpTransactionalEpochWithTV2Enabled`,
whose test cluster did not start ("Failed to start Raft server").
Alone it passed 3 of 3 times on this PR and 3 of 3 on 8aa57006ff.
Run 2: all 1357 tests pass.
End to end, with the `repro.sh` attached to KAFKA-21410 (one KRaft
node; a pass stops at offset 78644 with a full offset map, the next one
starts there; then the tombstone's delete horizon passes):
| OS | trunk (`p1`) | this PR (`p1`) | this PR (`p2`, control) |
|---|---|---|---|
| macOS 26, arm64 | `K=v1` | K not returned | K not returned |
| Linux, Ubuntu 24.04 container | `K=v1` | K not returned | K not returned |
| OpenBSD 8.0, arm64 (QEMU) | `K=v1` | K not returned | K not returned |
The trunk runs used 94b8481467 (no changes to the cleaner files up to
8aa57006ff). With the PR the first segment holds no record with key K
after the cleaning; on trunk it holds `offset: 120002 ... key: K
payload: v1`.
--
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]
