Replies: 2 comments 16 replies
-
|
Hallo @vrutkovs , |
Beta Was this translation helpful? Give feedback.
-
|
that's something for github.com/openshift/openshift-docs I suppose |
Beta Was this translation helpful? Give feedback.
-
|
Hello, After updating to latest 4.14 and after that to 4.15 many operators are failing. Errors like this install strategy failed: clusterrolebindings.rbac.authorization.k8s.io "openshift-gitops-operator-controller-manager-service-system:auth-delegator" already exists |
Beta Was this translation helpful? Give feedback.
-
|
https://issues.redhat.com/browse/OCPBUGS-28845 I suspect it will end up closed by someone who's a bit overzealous, so feel free to watch and help out @vrutkovs . I'll watch from my work account as well :) |
Beta Was this translation helpful? Give feedback.
-
|
Doesn't look as though the ticket is going anywhere. I ran into the same issue on my 4.14.0-0.okd-2023-11-14-101924 > 4.15.0-0.okd-2024-02-10-035534 upgrade today. If you're still struggling with this, I got around this by looking at which permissions OLM was struggling with, and adding the olm-managed=true label (just clusterrolebindings of the various operators within kube-system) Recently OLM started labelling all objects they own (operator-framework/operator-lifecycle-manager#3020). My assumption is the logic to capture existing objects isn't 100% accurate and has missed adding that label somehow but we'll find out once the ticket is looked at/updated. |
Beta Was this translation helpful? Give feedback.
-
|
probably this is still somehow related to https://issues.redhat.com/browse/OCPBUGS-25673 |
Beta Was this translation helpful? Give feedback.
-
|
I also suggest to inspect the CSV related rolebindings in the kube-system namespace: |
Beta Was this translation helpful? Give feedback.
-
|
This got reported again at some point when the OLM code that started using that label made it into OpenShift 4.15 so it's being tracked and triaged under a different issue now. I can confirm that serially deleting the offending resources as each InstallPlan failed eventually got me to a stable and ready state on OKD, which resulted in OLM redeploying them with the label. Most of the resources, in my case, that got orphaned were RoleBindings in the kube-system namespace. |
Beta Was this translation helpful? Give feedback.
-
Client tools for OpenShift
These archives contain the client tooling for OpenShift.
To verify the contents of this directory, use the 'gpg' and 'shasum' tools to
ensure the archives you have downloaded match those published from this location.
The openshift-install binary has been preconfigured to install the following release:
This discussion was created from the release 4.15.0-0.okd-2024-01-27-070424.
Beta Was this translation helpful? Give feedback.
All reactions