Red Hat Bugzilla – Bug 1302656
Openshift diagnostic: ClusterRegistry is reporting fake error
Last modified: 2016-01-28 10:12:33 EST
When I have a running docker registry pod and a terminated one (by some reason) , openshift diagnostic: ClusterRegistry is mistaking these as 2 working ones, and reporting fake error on image store consistency. This error message is unwanted.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Login openshift
2. Deploy the registry pod with wrong deployer image given in dc, e.g.: attempt to deploy with OSE image openshift3/ose-docker-registry:v184.108.40.206 on origin, and this will result in docker registry pod as terminating state
3. Edit dc with correct deployer image for origin: openshift/origin-docker-registry
4. oc get po -n default, and I now have a running docker registry pod and a terminated one:
NAME READY STATUS RESTARTS AGE
docker-registry-1-kshi4 0/1 Terminating 0 1h
docker-registry-2-o3094 1/1 Running 0 1h
5. Run "openshift ex diagnostics"
diagnostic: ClusterRegistry is reporting fake error on image store consistency since it mistake the 2 pods in repro step4 as 2 working ones:
[Note] Running diagnostic: ClusterRegistry
Description: Check that there is a working Docker registry
ERROR: [DClu1007 from diagnostic ClusterRegistry@openshift/origin/pkg/diagnostics/cluster/registry.go:209]
The "docker-registry" service has multiple associated pods each using
ephemeral storage. These are likely to have inconsistent stores of
images. Builds and deployments that use images from the registry may
fail sporadically. Use a single registry or add a shared storage volume
to the registries.
The fake error message reported by openshift diagnostic is unwanted
I have rc of the docker registry = 1 in the mean time
For the purposes of this check (DClu1007) it should be changed to exclude pods that are not Ready.