controller Ingress/Service are sometimes missing after a Restart

Article ID:360047094291
1 minute readKnowledge base

Issue

  • After restarting a Managed controller / Team controller from Operations center:

    • the controller is not reachable anymore with either a 404 or a 503

    • the controller Service and/or Ingress is missing

Environment

Related Issues

  • CPLT2-6520: controller restart logic can lead to missing Ingress/Service

Explanation

Until version 2.235.2.3 of CloudBees CI, the restart logic of Managed controllers sends the order to recreate the controller resources without waiting for their deletion. There is a slight chance that some of the controller resources be deleted after they have been recreated. In most cases, the Service or the Ingress would be missing, causing the controller to be unreachable.

  • If the controller Ingress is missing, the controller URL responds with a 404.

  • If the controller Service is missing, the controller URL responds with a 503.

This is caused by a bug in the Master Provisioning Plugin.

Resolution

Solution

The issue has been fixed in Master Provisioning plugin version 2.5.3 and is available in the July release 2.235.2.3 of CloudBees CI.

The solution is to upgrade to CloudBees CI 2.235.2.3

Workaround

  • Stop the controller from Operations Center

  • Wait for resources to be deleted (a couple of seconds)

  • Start the controller from Operations Center