You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hive.apache.org by "Peter Vary (JIRA)" <ji...@apache.org> on 2017/12/08 09:12:00 UTC
[jira] [Commented] (HIVE-18212) Make sure Yetus check always has a
full log
[ https://issues.apache.org/jira/browse/HIVE-18212?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16283238#comment-16283238 ]
Peter Vary commented on HIVE-18212:
-----------------------------------
+1
> Make sure Yetus check always has a full log
> -------------------------------------------
>
> Key: HIVE-18212
> URL: https://issues.apache.org/jira/browse/HIVE-18212
> Project: Hive
> Issue Type: Sub-task
> Reporter: Adam Szita
> Assignee: Adam Szita
> Attachments: HIVE-18212.0.patch
>
>
> Some yetus log files are left incomplete, and in these same runs {{tee}} subprocesses are left running and dangling on the ptest server.
> This is because of a bug in the yetus runner velocity template script where we make a redirection of stdout:
> {code}
> ./dev-support/test-patch.sh ${patchFile} ..... 2>&1 | tee ${logFile}
> {code}
> If the yetus output is big enough (>62K) tee will stop writing the log file and is left running even after {{test-patch.sh}} finished successfully. This because we don't make anything consume the stdout and most probably some buffers get full on Linux side.
> We should also make sure that yetus runs(since they are executed parallel to ptest test phase) are not interfering with each other in case they run very long and overlap.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)