Sometimes after Jenkins restarts, jobs tied to specific label end up in queue indefinitely, marked as 'no node labeled ... available'.
My situation is as follows: I have 4 nodes marked (by platformlabeler) as RedHatEnterpriseServer. When Jenkins gets up, it initiates a build for three jobs, they all end up in queue with this 'no node labeled 'RedHatEnterpriseServer' available'. However, if I check any of the nodes which should be marked with this label, they all do have their label applied. The only solution for this situation is to enter configuration for any of those nodes and press Save button. All labels are then suddenly available to Jenkins and the jobs get started on every proper node.
I don't know if this is related to platformlabeler or the core itself, but it would be nice if somebody could look at it, as this behavior requires Jenkins monitoring to see whether this happens or not.