search

cloudfoundry-incubator / kubecf

Cloud Foundry on Kubernetes
Apache License 2.0
115 stars 62 forks source link

There is no liveness probe for cloud_controller_ng container in `api` pod #1661

Open JimmyMa opened 3 years ago

JimmyMa commented 3 years ago

Describe the bug When the cloud_controller_ng process got hang, the cloud_controller_ng container didn't got restarted, because there is no liveness probe for cloud_controller_ng

To Reproduce Below is just a way to simulate the hang of cloud_controller_ng.

  1. get into cloud_controller_ng container and run rm /var/vcap/data/cloud_controller_ng/cloud_controller.sock
  2. then the cloud-controller-ng-cloud-controller-ng and cloud-controller-ng-nginx get into NOT READY state, but they whould not get restarted.

Expected behavior In BOSH CF, the monit will restart cloud-controller-ng process if it fails the healthy check. Expect the cloud-controller-ng-cloud-controller-ng in KubeCF should be restarted if it got hang or fails the healthy check.