-
Bug
-
Resolution: Done
-
Major
-
None
-
5
-
False
-
False
-
Undefined
-
-
Sprint 57, Sprint 58 - week 1
When cold starting a cluster (all nodes down, then powered on), automatic route creation does not work. istiod reports:
2021-03-11T08:54:24.803501Z info ior Registering IOR component
2021-03-11T08:54:32.654781Z error ior OpenShift routes are not supported in this cluster. IOR is not enabled.
Reason seems to be that the route api is not available yet, when the istiod service starts
Route API Available:
[stack@osp16amd ocp-test3]$ while :; do sleep 1; echo $(date) $(timeout 1 oc api-resources | grep Route); done ... Thu Mar 11 08:54:56 UTC 2021 Thu Mar 11 08:54:58 UTC 2021 routes route.openshift.io true Route
Istio Pod started:
[stack@osp16amd ocp-test3]$ oc get pods -n istio-system istiod-basic-7d8ccd96f7-rndh6 -o json | jq -r .status.containerStatuses[0].state.running.startedAt 2021-03-11T08:54:24Z
As the istiod is started BEFORE the route api is available, the check for the availability of the "Route" API fails. It seems this is never rechecked.
Restarting the istiod (oc delete pod) fixes this issue.
Acceptance criteria:
- IOR check never fails, whether the cluster is Openshift or not, IOR should repeat the check with the API server until it succeeds.
- mentioned on