-
Bug
-
Resolution: Done
-
Critical
-
None
-
3
-
False
-
False
-
No
-
-
-
-
-
-
-
v1.2.0
-
No
-
No
-
Yes
-
None
-
-
MODH Sprint 33, MODH Sprint 34
Description of problem:
Dashboard fails to load on OpenShift 4.8.X.
Prerequisites (if any, like setup, operators/versions):
Build v111-46+
OpenShift 4.8.X
Steps to Reproduce
- Navigate to the RHODS dashboard
Actual results:
User cannot access dashboard. Instead receives error 500 Internal Error "oauth-serving-cert"
Expected results:
Dashboard loads. Monitoring routes load.
Reproducibility (Always/Intermittent/Only Once):
Always
Build Details: v111-46+
Workaround:
Scale down the operator and set the image tag on the dashboard deployment. This effectively shuts down the operator and leaves RHODS unable to heal/update, but it works.
Additional info:
Initial Analysis:
The oauth-proxy container the dashboard uses tracks the latest image. When latest was updated to 4.9, it broke existing installations of RHODS on OpenShift 4.8 by looking for configmap oauth-serving-cert which only exists on OpenShift 4.9 in the openshift-config-managed namespace.
Usually, it takes a few releases after the release of new versions for upgrades to be allowed to the new version, this leaves existing installs broken. (existing installations on 4.8 have to wait for awhile before they can upgrade to 4.9). For this reason, this will continue to be a problem for any OpenShift upgrade if the oauth-proxy container isn't set to track the OpenShift version.