[
https://issues.apache.org/jira/browse/KAFKA-6647?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17056509#comment-17056509
]
ASF GitHub Bot commented on KAFKA-6647:
---------------------------------------
guozhangwang commented on pull request #8267: KAFKA-6647: Do note delete the
lock file while holding the lock [WIP]
URL: https://github.com/apache/kafka/pull/8267
1. Inside StateDirectory#cleanRemovedTasks, skip deleting the lock file (and
hence the parent directory) until releasing the lock. And after the lock is
released only go ahead and delete the parent directory if `manualUserCall ==
true`. That is, this is triggered from `KafkaStreams#cleanUp` and users are
responsible to make sure that Streams instance is not started and hence there
are no other threads trying to grab that lock.
2. As a result, during scheduled cleanup the corresponding task.dir would
not be empty but be left with only the lock file, so effectively we still
achieve the goal of releasing disk spaces. For callers of `listTaskDirectories`
like KIP-441 (cc @ableegoldman to take a look) I've introduced a new
`listNonEmptyTaskDirectories` which excludes such dummy task.dirs with only the
lock file left.
### Committer Checklist (excluded from commit message)
- [ ] Verify design and implementation
- [ ] Verify test coverage and CI build status
- [ ] Verify documentation (including upgrade notes)
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
> KafkaStreams.cleanUp creates .lock file in directory its trying to clean
> (Windows OS)
> -------------------------------------------------------------------------------------
>
> Key: KAFKA-6647
> URL: https://issues.apache.org/jira/browse/KAFKA-6647
> Project: Kafka
> Issue Type: Bug
> Components: streams
> Affects Versions: 1.0.1
> Environment: windows 10.
> java version "1.8.0_162"
> Java(TM) SE Runtime Environment (build 1.8.0_162-b12)
> Java HotSpot(TM) 64-Bit Server VM (build 25.162-b12, mixed mode)
> org.apache.kafka:kafka-streams:1.0.1
> Kafka commitId : c0518aa65f25317e
> Reporter: George Bloggs
> Priority: Minor
>
> When calling kafkaStreams.cleanUp() before starting a stream the
> StateDirectory.cleanRemovedTasks() method contains this check:
> {code:java}
> ... Line 240
> if (lock(id, 0)) {
> long now = time.milliseconds();
> long lastModifiedMs = taskDir.lastModified();
> if (now > lastModifiedMs + cleanupDelayMs) {
> log.info("{} Deleting obsolete state directory {}
> for task {} as {}ms has elapsed (cleanup delay is {}ms)", logPrefix(),
> dirName, id, now - lastModifiedMs, cleanupDelayMs);
> Utils.delete(taskDir);
> }
> }
> {code}
> The check for lock(id,0) will create a .lock file in the directory that
> subsequently is going to be deleted. If the .lock file already exists from a
> previous run the attempt to delete the .lock file fails with
> AccessDeniedException.
> This leaves the .lock file in the taskDir. Calling Utils.delete(taskDir) will
> then attempt to remove the taskDir path calling Files.delete(path).
> The call to files.delete(path) in postVisitDirectory will then fail
> java.nio.file.DirectoryNotEmptyException as the failed attempt to delete the
> .lock file left the directory not empty. (o.a.k.s.p.internals.StateDirectory
> : stream-thread [restartedMain] Failed to lock the state directory due
> to an unexpected exception)
> This seems to then cause issues using streams from a topic to an inMemory
> store.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
