You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-dev@hadoop.apache.org by "Vinod K V (JIRA)" <ji...@apache.org> on 2009/04/16 12:08:15 UTC
[jira] Commented: (HADOOP-5614) SIGKILL should be sent only after a
while after SIGTERM is sent in sigkillInCurrentThread() similar to what
sigKillThread is doing
[ https://issues.apache.org/jira/browse/HADOOP-5614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12699615#action_12699615 ]
Vinod K V commented on HADOOP-5614:
-----------------------------------
mapred.tasktracker.tasks.sleeptime-before-sigkill is currently a job configuration which I think was not the original intention. This should be picked up from TT's local conf and needs to be done as part of this patch.
> SIGKILL should be sent only after a while after SIGTERM is sent in sigkillInCurrentThread() similar to what sigKillThread is doing
> ----------------------------------------------------------------------------------------------------------------------------------
>
> Key: HADOOP-5614
> URL: https://issues.apache.org/jira/browse/HADOOP-5614
> Project: Hadoop Core
> Issue Type: Bug
> Components: mapred
> Reporter: Ravi Gummadi
> Assignee: Ravi Gummadi
> Attachments: HADOOP-5614.patch
>
>
> sigkillInCurrentThread() doesn't give time for graceful exit of taskJvm after SIGTERM is sent. Currently mapred.tasktracker.tasks.sleeptime-before-sigkill is used by SigKillThread only. It should be used by sigKillInCurrentThread() also.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.