[ https://issues.apache.org/jira/browse/HDFS-12711?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16221500#comment-16221500 ]
Allen Wittenauer commented on HDFS-12711: ----------------------------------------- Some status, after taking part of the afternoon to do something besides look at this. a) It's clear that surefire and therefore both Jenkins and Yetus, doesn't by default report on tests that didn't launch during low memory conditions. This is a bit problematic: we could have had this problem for a very long time but we never would have known it. It'd be great if there was a way to ask surefire what is expected to run and then do a pre-/post- comparison. Something else to consider: there are likely lots and lots of projects that may not know they have tests failing in this way... b) That said, there are two things that we might be able to add to Yetus that might help: some sort of reporting of JVMs that are still running after maven exits and killing leftover Java processes in between module unit testing. Obviously, this would only be available when running under Docker. c) This patch is working well enough that I've reconfigured Precommit-HADOOP-build to use my private Yetus branch with this patch. This is primarily to give me more general usability data. I've set the proclimit to be 5000 and mem limit to be 20g (that's slightly less than half of the node's full memory size). proclimit is likely too high but should give us a lot of wiggle room to tune downwards. d) As part of deploying this version of Yetus, I also took the opportunity to reconfigure the job to be laid out in a more sane manner as well as setting it to run on all of the Hadoop build nodes. > deadly hdfs test > ---------------- > > Key: HDFS-12711 > URL: https://issues.apache.org/jira/browse/HDFS-12711 > Project: Hadoop HDFS > Issue Type: Test > Affects Versions: 2.9.0, 2.8.2 > Reporter: Allen Wittenauer > Priority: Critical > Attachments: HDFS-12711.branch-2.00.patch > > -- This message was sent by Atlassian JIRA (v6.4.14#64029) --------------------------------------------------------------------- To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org