Description of problem: I have an existing 4.5.5 cluster, trying to install CNV 2.4.1 following the CLI doc: https://docs.openshift.com/container-platform/4.5/virt/install/installing-virt-cli.html However the deploy makes no progress and the catalog-operator logs show the 2.4.1 release is "not found": ime="2020-09-09T13:57:06Z" level=info msg=syncing event=update reconciling="*v1alpha1.Subscription" selflink=/apis/operators.coreos.com/v1alpha1/namespaces/openshift-sriov-network-operator/subscriptions/sriov-network-operator-subscription time="2020-09-09T13:58:01Z" level=info msg=syncing event=update reconciling="*v1alpha1.Subscription" selflink=/apis/operators.coreos.com/v1alpha1/namespaces/openshift-cnv/subscriptions/hco-operatorhub time="2020-09-09T13:58:01Z" level=info msg=syncing event=update reconciling="*v1alpha1.Subscription" selflink=/apis/operators.coreos.com/v1alpha1/namespaces/openshift-cnv/subscriptions/hco-operatorhub E0909 13:58:01.793882 1 queueinformer_operator.go:290] sync {"update" "openshift-cnv"} failed: {kubevirt-hyperconverged 2.4 kubevirt-hyperconverged-operator.v2.4.1 {redhat-operators openshift-marketplace}} not found: rpc error: code = Unknown desc = no entry found for kubevirt-hyperconverged 2.4 kubevirt-hyperconverged-operator.v2.4.1 E0909 13:58:01.802058 1 queueinformer_operator.go:290] sync "openshift-cnv" failed: {kubevirt-hyperconverged 2.4 kubevirt-hyperconverged-operator.v2.4.1 {redhat-operators openshift-marketplace}} not found: rpc error: code = Unknown desc = no entry found for kubevirt-hyperconverged 2.4 kubevirt-hyperconverged-operator.v2.4.1 time="2020-09-09T13:58:01Z" level=info msg=syncing event=update reconciling="*v1alpha1.Subscription" selflink=/apis/operators.coreos.com/v1alpha1/namespaces/openshift-cnv/subscriptions/hco-operatorhub E0909 13:58:02.002589 1 queueinformer_operator.go:290] sync {"update" "openshift-cnv"} failed: {kubevirt-hyperconverged 2.4 kubevirt-hyperconverged-operator.v2.4.1 {redhat-operators openshift-marketplace}} not found: rpc error: code = Unknown desc = no entry found for kubevirt-hyperconverged 2.4 kubevirt-hyperconverged-operator.v2.4.1 [...] Version-Release number of selected component (if applicable): # oc version Client Version: 4.5.5 Server Version: 4.5.5 Kubernetes Version: v1.18.3+08c38ef
This is just OLM not correctly refreshing operator hub content from the AppRegistry, as a workaround I can suggest to run: oc patch operatorhub.config.openshift.io/cluster -p='{"spec":{"disableAllDefaultSources":true}}' --type=merge oc patch operatorhub.config.openshift.io/cluster -p='{"spec":{"disableAllDefaultSources":false}}' --type=merge to force OLM to refresh the content. Then we need to keep investigating it. Jenifer, can you please try applying the suggested workaround on your cluster?
Thanks Simone! That workaround to refresh OLM did the trick and now it is progressing: NAME DISPLAY VERSION REPLACES PHASE kubevirt-hyperconverged-operator.v2.4.1 OpenShift Virtualization 2.4.1 Installing Do you want to close this bug if the OLM problem is tracked elsewhere?
Simone, the refresh helped. How would the system have fixed the issue? How much time would this have taken? Why did Jenifer run into this issue? Would a regular user run into it as well?
OLM is tracking this bug here https://bugzilla.redhat.com/show_bug.cgi?id=1877484
this is the 4.5.z bug https://bugzilla.redhat.com/show_bug.cgi?id=1877564
(In reply to Fabian Deutsch from comment #3) > How would the system have fixed the issue? How much time would this have > taken? > Why did Jenifer run into this issue? Would a regular user run into it as > well? This is not CNV specific, I fear that without a forced refresh OLM will never notice it. The proposed workaround can be considered safe enough to be proposed to customers in a KBS article until we will get a fix in a future release of OCP.
(In reply to Simone Tiraboschi from comment #6) > (In reply to Fabian Deutsch from comment #3) > > > How would the system have fixed the issue? How much time would this have > > taken? > > Why did Jenifer run into this issue? Would a regular user run into it as > > well? > > This is not CNV specific, I fear that without a forced refresh OLM will > never notice it. > The proposed workaround can be considered safe enough to be proposed to > customers in a KBS article until we will get a fix in a future release of > OCP. How do you suggest we prepare for testing this fix? Can we emulate the circumstances of this bug manually somehow? Is this something we can even test or should we rely on that bug fix?
OCP issue fixed. moving to ON QA