Tim Armstrong has posted comments on this change. ( http://gerrit.cloudera.org:8080/11874 )
Change subject: IMPALA-7738: Implement timeouts for HDFS open calls ...................................................................... Patch Set 4: (1 comment) http://gerrit.cloudera.org:8080/#/c/11874/4//COMMIT_MSG Commit Message: http://gerrit.cloudera.org:8080/#/c/11874/4//COMMIT_MSG@14 PS4, Line 14: and submitting it to a thread pool. The original caller waits > Answers: We could start a graceful shutdown - that would prevent more things getting scheduled and give an opportunity for fragments to drain before it does a hard exit. Although maybe we'd want to abort() instead of exit(1) at the end of the shutdown so we get a core. It might also be worth tracking the proportion of threads that have exceeded some lower threshold. Then we could warn or alert if that got to the point where performance would degrade. It seems like currently it wouldn't be obvious if things were running in certain degraded modes. Anyway, I don't think those are blockers for getting this in, since this is obviously better, but it could improve the end result a lot. -- To view, visit http://gerrit.cloudera.org:8080/11874 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: Ia14403ca5f3f19c6d5f61b9ab2306b0ad3267454 Gerrit-Change-Number: 11874 Gerrit-PatchSet: 4 Gerrit-Owner: Joe McDonnell <[email protected]> Gerrit-Reviewer: Impala Public Jenkins <[email protected]> Gerrit-Reviewer: Joe McDonnell <[email protected]> Gerrit-Reviewer: Michael Ho <[email protected]> Gerrit-Reviewer: Philip Zeyliger <[email protected]> Gerrit-Reviewer: Tim Armstrong <[email protected]> Gerrit-Comment-Date: Thu, 08 Nov 2018 18:23:12 +0000 Gerrit-HasComments: Yes
