Hean-Chhinling commented on PR #8798: URL: https://github.com/apache/hadoop/pull/8798#issuecomment-6316083506
Hi guys, I have tested the latest patch on a real cluster with the followings scenarios: - Localization and cache cleanup: files under filecache were created, renamed and deleted while the health check ran. The jar before the patch, it marked the node UNHEALTHY with NoSuchFileException. With the patch the node stayed healthy. - MapReduce jobs as a non-NodeManager user: with the old code, the user-owned usercache/<user>/filecache failed the disk. With the patch all jobs succeeded and the node stayed healthy, including with the depth raised so the walk reaches that directory. - Negative check: an unreadable directory owned by the NodeManager user still marks the disk as failed with a clear Cannot access <path> message. The node recovers once the directory is fixed. Are there any scenario that you would like to me test? It would be nice to get this bug resolve as soon as possible as it is falsely mark NM unhealthy and could potentially affects lots of job run. -- 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]
