We encountered as issue, where after updating our master Server Jenkins was no longer able to start up and crashed our Kubernetes Cluster.

After analyzing we found the cause. The temporary nodes Jenkins creates for builds on the Kubernetes Cluster stay on the master Server after a job is finished. So we had multiple thousands of these directories under /var/lib/jenkins/nodes each with a config.xml. On Jenkins startup the Kubernetes plugin tried to get ids for each of the nodes from our Kubernetes cluster, which crashed our Kubernetes cluster.

After deleting the temporary nodes from our Jenkins master, it started up normally again.

Is there an option which automatically deletes the temporary nodes from the Jenkins master?