On Tue, Feb 6, 2018 at 4:23 PM, Matthias Sohn <[email protected]>
wrote:

> On Tue, Feb 6, 2018 at 4:19 PM, Denis Roy <denis.roy@eclipse-foundation.
> org> wrote:
>
>> I've had to restart the jGit JIPP -- it was spawning many Java processes
>> and the load average of the host (hipp3) was exceeding 117.
>>
>> Since the restart, it is behaving much better. If your Jenkins instance
>> is on the same host, it is possible that some of your build performance
>> was negatively affected during that time.
>>
>
> any idea why it was doing that ?
>
> Maybe if this happens again try to capture some thread dumps of those java
> processes.
> This could help to find out what could be the reason.
>

looks like you were witnessing several concurrent JGit verification builds
each running JGit unit
tests. For several of the changes in the series
https://git.eclipse.org/r/#/c/116788 a build was started
but didn't come back reporting a result. You probably killed them when
seeing the overloaded server.

Looks like we should reduce the number of executors, I reduced it from 3 to
2 for the JGit JIPP.

-Matthias
_______________________________________________
cross-project-issues-dev mailing list
[email protected]
To change your delivery options, retrieve your password, or unsubscribe from 
this list, visit
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev

Reply via email to