Hi, I am using kubernetes plugin to provision dynamically Jenkins slaves in our Openshift cluster. We configured our pods to use a persistent storage to store Jenkins tools and caches (i.e. maven local repository). Workspaces are stored in EmptyDir volume. We also configured the plugin to destroy pods after 30 minutes idle. We are having problems because there are jobs the generate big binaries, so when a pod runs several jobs it may take a lot of ephemeral storage in the Openshift node and our Openshift administrators are complaining.

Is it possible to limit the local storage that a pod can take ? I suggest as an improvement that, after each build, the plugin checks storage consumption of the pod. If it is higher than certain threshold, then the slave will be deleted.