Pipeline jobs getting stuck on Jenkins master

[Chris Packham]

Hi,

We've got a setup with a mix of Pipeline, Multi-configuration and
Freestyle jobs. We have a Jenkins master and many slave nodes for
executing the builds.

Most of the time the pipeline jobs move onto one of the slave nodes
(chosen by the agent label) but occasionally we see issues where one
of the Pipeline jobs gets stuck on the master and causes the web
interface and scheduler to become unresponsive. If we can still access
the web interface we can't abort the stuck job.

Does this sound familiar to anyone?

Any suggestions on how to debug this?

Are my expectations correct that the pipelines initially start on the
master and move to a slave once they hit the agent directive?

Thanks,
Chris

[Robert Hales]

I would bet your java process is running into excess Garbage Collection,
causing the entire thing to hang. It probably isn't because of the pipeline
itself. Check your garbage collection when this happens to see if you are
hung with full collections.

Post by Chris Packham
Hi,
We've got a setup with a mix of Pipeline, Multi-configuration and
Freestyle jobs. We have a Jenkins master and many slave nodes for
executing the builds.
Most of the time the pipeline jobs move onto one of the slave nodes
(chosen by the agent label) but occasionally we see issues where one
of the Pipeline jobs gets stuck on the master and causes the web
interface and scheduler to become unresponsive. If we can still access
the web interface we can't abort the stuck job.
Does this sound familiar to anyone?
Any suggestions on how to debug this?
Are my expectations correct that the pipelines initially start on the
master and move to a slave once they hit the agent directive?
Thanks,
Chris