hadoop-common-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jon Bringhurst (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HADOOP-12621) Allow shutdown before full YARN_STOP_TIMEOUT has elapsed
Date Mon, 07 Dec 2015 18:34:11 GMT

     [ https://issues.apache.org/jira/browse/HADOOP-12621?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Jon Bringhurst updated HADOOP-12621:
------------------------------------
    Description: 
I noticed that yarn-daemon.sh will wait the full time of YARN_STOP_TIMEOUT on shutdown. Here's
a snippet from yarn-daemon.sh in release-2.7.1 (note the sleep for the full time):

{noformat}
    if [ -f $pid ]; then
      TARGET_PID=`cat $pid`
      if kill -0 $TARGET_PID > /dev/null 2>&1; then
        echo stopping $command
        kill $TARGET_PID
        sleep $YARN_STOP_TIMEOUT
        if kill -0 $TARGET_PID > /dev/null 2>&1; then
          echo "$command did not stop gracefully after $YARN_STOP_TIMEOUT seconds: killing
with kill -9"
          kill -9 $TARGET_PID
        fi
      else
        echo no $command to stop
      fi
      rm -f $pid
    else
      echo no $command to stop
    fi
{noformat}

The script should poll the process rather than wait the entire time.

I'll also dig into trunk to see what the behavior is there.

  was:
I noticed that yarn-daemon.sh will wait the full time of YARN_STOP_TIMEOUT on shutdown. Here's
a snippet from yarn-daemon.sh in release-2.7.1:

{noformat}
    if [ -f $pid ]; then
      TARGET_PID=`cat $pid`
      if kill -0 $TARGET_PID > /dev/null 2>&1; then
        echo stopping $command
        kill $TARGET_PID
        sleep $YARN_STOP_TIMEOUT
        if kill -0 $TARGET_PID > /dev/null 2>&1; then
          echo "$command did not stop gracefully after $YARN_STOP_TIMEOUT seconds: killing
with kill -9"
          kill -9 $TARGET_PID
        fi
      else
        echo no $command to stop
      fi
      rm -f $pid
    else
      echo no $command to stop
    fi
{noformat}

The script should poll the process rather than wait the entire time.

I'll also dig into trunk to see what the behavior is there.


> Allow shutdown before full YARN_STOP_TIMEOUT has elapsed
> --------------------------------------------------------
>
>                 Key: HADOOP-12621
>                 URL: https://issues.apache.org/jira/browse/HADOOP-12621
>             Project: Hadoop Common
>          Issue Type: Improvement
>            Reporter: Jon Bringhurst
>            Assignee: Jon Bringhurst
>
> I noticed that yarn-daemon.sh will wait the full time of YARN_STOP_TIMEOUT on shutdown.
Here's a snippet from yarn-daemon.sh in release-2.7.1 (note the sleep for the full time):
> {noformat}
>     if [ -f $pid ]; then
>       TARGET_PID=`cat $pid`
>       if kill -0 $TARGET_PID > /dev/null 2>&1; then
>         echo stopping $command
>         kill $TARGET_PID
>         sleep $YARN_STOP_TIMEOUT
>         if kill -0 $TARGET_PID > /dev/null 2>&1; then
>           echo "$command did not stop gracefully after $YARN_STOP_TIMEOUT seconds: killing
with kill -9"
>           kill -9 $TARGET_PID
>         fi
>       else
>         echo no $command to stop
>       fi
>       rm -f $pid
>     else
>       echo no $command to stop
>     fi
> {noformat}
> The script should poll the process rather than wait the entire time.
> I'll also dig into trunk to see what the behavior is there.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message