joseluisll commented on PR #8903:
URL: https://github.com/apache/hadoop/pull/8903#issuecomment-6561577542
Rebased onto trunk (`57f78d283ae`) and republished the branch, so both
checks below have been re-run on a current base. Recording what the previous
run showed, so the red votes can be read in context. Neither failure is in a
file this patch touches: the diff is three `pom.xml` files, +14 lines,
declaring `org.slf4j:slf4j-api` at the version managed in
`hadoop-project/pom.xml`.
### Yetus `test4tests` -1
Structural for a pom-only patch: `plugins.d/test4tests.sh` counts changed
files matching `(^|/)tests?/`, gets zero, and votes -1 asking for a
justification. The justification is in the description: this patch adds no code
and no behavior, so there is nothing to assert about Java sources; the modules'
existing tests that would notice a broken `slf4j-api` classpath already run in
the precommit. For the record, the same vote plus a -1 overall did not prevent
#8848 (HADOOP-20032, a Jackson bump) from merging on 2026-10-24.
### Yetus `unit` -1 in `hadoop-hdfs`
One hard failure out of 7323 in the previous run:
`TestNameNodeMXBean.testDecommissioningNodes` at line 435, a
`decommissionDuration` timing assertion (expected 2 got 1; expected 1 got 0),
failing all three runs. It is tracked as HDFS-17225 ("Fix
TestNameNodeMXBean#testDecommissioningNodes") and HDFS-16607
("TestNameNodeMXBean.testDecommissioningNodes is flaky"), both open. The other
five failures in that run passed on retry: `TestViewFileSystemHdfs`,
`TestDecommission`, `TestBalancer`, `TestPipelinesFailover`, `TestFileCreation`.
Worth noting the patch is a no-op for these modules' own classpaths:
`hadoop-common` is `provided` in all three, so `slf4j-api` 1.7.36 already
reaches each module's compile and test classpath transitively. The added
element only makes the declaration explicit for downstream consumers, so it
cannot change what these modules load or how their tests behave.
### `Build` (GitHub Actions)
The failing job was `Run / Test yarn-server-rm (Java 17) ubuntu_24-trunk`,
on a module this patch does not modify. Its only failing test,
`TestCapacitySchedulerSchedulingRequestUpdate.testBasicPendingResourceUpdate`,
passed on retry; the job failed because the forked JVM running
`TestLeaderElectorService` crashed (`Process Exit Code: 14`, "terminated
without properly saying goodbye"). The same module's unit run is +1 on this
patch's base commit in nightly build 199 (`hadoop-yarn-server-resourcemanager
unit +1`, 92m21s).
The previous "This commit cannot be built" status mirrors Yetus's -1 overall
rather than being an additional failure; the same message appears on #8848,
#8699 and #8704, all of which carry a Yetus -1.
### Merge order
This needs to land before #8704 (HADOOP-19972) or together with it. It also
shares `hadoop-hdfs/pom.xml` lines with #8699 and `hadoop-yarn-client/pom.xml`
lines with #8892, so whichever merges second will need a rebase.
--
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]
