Hey there -

I think it might be worth posting the pipeline of your orchestration job 
that runs the 12000 builds for anyone to comment in more detail.

>From what I understand, if your pipeline is not orchestrating each build in 
a node section it will use something called a 'flyweight executor' on the 
Jenkins master.

If this is happening, even though you have 100 agents your master may be 
heavily loaded running the pipeline - I guess each build allocates some 
Java objects that have to be GCed.

If I can't find any documentation, I tend to go onto Github to look at the 
source code to understand what the Jenkins pipeline is doing behind the 
scenes: https://github.com/jenkinsci/workflow-cps-plugin.

I ran a test on a system where I had an orchestrator kicking off 1000 
builds every 4 minutes on a schedule. After a few days the Jenkins service 
stopped responding (Java memory issues). One thing I have found I needed to 
pay attention to was file and process limits as documented here:
https://support.cloudbees.com/hc/en-us/articles/204231510-Memory-problem-unable-to-create-new-native-thread-


All the best,
--Bill



On Thursday, 23 February 2017 21:33:42 UTC, Chris Overend wrote:
>
> So not sure if this is a Jenkins limitation or pipeline.
> The jobs never exceeded available resources.
> The garbage collection was stable.
>
> So why did it lock-up?
>
> It did say I used 
>
>    - 2950 million active threads
>    - 350 threads
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"Jenkins Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/jenkinsci-users/a504182f-ebf8-46a6-b343-03f639fefc8a%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to