Slaves which never connect due to ECS resource constraints never get removed

XMLWordPrintable

      I've given a build of master a go in order to get mount points (among other things) happening in my ECS slaves.

      What I've noticed as part of https://github.com/jenkinsci/amazon-ecs-plugin/commit/88f43cef39dfd6bd4e0f9856dbe91eba9f22fcc2 is that any Slaves which are created but never have a matching ECS task created (such as due to resource constraints, ala. memory) never get cleaned up by Jenkins.

      At the moment I have some 2500+ ECS Slaves in Jenkins which don't appear to want to go away as I don't think that there's any Jenkins thread looking after them as a result of the RetentionStrategy.NOOP

      I'll make a PR as I believe that having a RetentionStrategy of some sort is actually needed, or there's a listener on the cloud provisioning bit to inhibit the creation of a slave if there are no ECS resources available to create a task

            Assignee:
            Nicolas De Loof
            Reporter:
            Lee Webb
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

              Created:
              Updated:
              Resolved: