Description of problem:
The oc get cs command should accurately display the component status however it fails to complete the check which results in an Unhealthy status being reported.
Version-Release number of selected component (if applicable):
OpenShift Enterprise version 3.0.2
How reproducible:
Very
Steps to Reproduce:
1. Setup environment according to directions
2. run `oc get cs`
Actual results:
See that the status of various parts of the system; scheduler, controller-manager, and etcd are in an Unhealthy status.
Expected results:
If system was Just set up, they should all be in a healthy status.
Additional info:
Identified on github as well, https://github.com/openshift/origin/issues/3667
Comment 2Brenton Leanhardt
2015-10-06 20:24:21 UTC
From reading the upstream bug it sounds like we'll have to fix this in Kubernetes.
Paul,
I understand it is no small endeavor, however, is this something that could make a future release? 3.1 is already out but potentially 3.2 or further?
The kubernetes componentstatuses API have hard-coded assumptions about cluster topology that are incorrect and insecure (it assumes components are running on localhost, assumes they are running insecurely, and assumes they are using hardcoded ports).
It is not supported or expected to work in Openshift.