[
https://issues.apache.org/jira/browse/DAEMON-468?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17921741#comment-17921741
]
Jean-Frederic Clere commented on DAEMON-468:
--------------------------------------------
Well not waiting or not waiting enough is bad: you have to kill the service
that was started or reboot the box.
The timeout being used twice basically double the maximum wait time to
stop+kill the service, I don't thing we need to fix it but it should be
documented.
> The timeout handling in procrun needs to be fixed.
> ---------------------------------------------------
>
> Key: DAEMON-468
> URL: https://issues.apache.org/jira/browse/DAEMON-468
> Project: Commons Daemon
> Issue Type: Task
> Components: prunsrv
> Affects Versions: 1.4.1
> Reporter: Jean-Frederic Clere
> Assignee: Jean-Frederic Clere
> Priority: Major
>
> The bug in commons-daemon is that the hint on the max timeout given to
> windows is 3 seconds when no timeout parameter is given, so windows will kill
> procrun after 3 seconds, using a timeout parameter gives an hint but the hint
> is half the time procrun might wait before stopping.
> If procrun is killed by windows, only the task manager can but used to stop
> the started process, note it is a hard kill, so that is something we want to
> avoid.
>
> I am planning to change commons-daemon logic for the timeout:
> - No timeout , or Timeout<0 will wait for ever (until the shutdown timeout
> is configured in windows).
> - The actual value timeout applies to the command for the stop
> (StopImage/StopPath/StopParams) and the wait for the started processes by the
> start to stop, in the worse case after both timeout the child processes are
> killed by procrun and procrun will exit.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
