-
Improvement
-
Resolution: Unresolved
-
Minor
-
None
We're using Jenkins in a master-slave setup, where slaves are created on demand as Docker containers. The workspace is shared with the master so that one can still access it after the slave has gone away. There is a special WorkspaceBrowser which handles this, see JENKINS-41035.
Unfortunately one cannot wipe out the workspace, because AbstractProject#doDoWipeOutWorkspace() fails to find the node on which the job was built (because the node is already gone).
So basically we need a mechanism to tell Jenkins that the slave has gone, but the workspace is still available locally.