-
Bug
-
Resolution: Done
-
Undefined
-
None
-
4.13.z
-
No
-
CNF RAN Sprint 239, CNF RAN Sprint 240
-
2
-
False
-
-
Description of problem:
Deployment of a 4.13.z SNO spoke from a hub cluster with ACM 2.8 fails. The failure is caused by warning messages generated by the recommended siteconfig for pre-caching as described here: https://docs.openshift.com/container-platform/4.13/scalability_and_performance/ztp_far_edge/ztp-precaching-tool.html#ztp-pre-caching-config-con_pre-caching Specifically the ignitionConfigOverride key/value The recommended pre-caching configuration likely needs to be updated to correct this issue, and least for OCP 4.13+
Version-Release number of selected component (if applicable):
OCP 4.13.z (hub and spoke) ACM 2.8 (hub)
How reproducible:
Always
Steps to Reproduce:
1. Depoloy 4.13.z hub cluster with ACM 2.8 2. Attempt to deploy spoke with OCP 4.13.z with the DU profile, including the recommended configuration for pre-caching as linked above. 3.
Actual results:
Spoke deployment fails and generates the following warning messages on the console (orange text). Image also linked in a comment below
Expected results:
Spoke deployment succeeds
Additional info:
Removing the ignitionConfigOverride entry from the siteconfig allows install to succeed.
An alternate workaround is to disable the ACM 2.8 "converged flow" installation
path, which is the default, with code like this:
hub$ cat <<EOF | kubectl create -f - > apiVersion: v1 > kind: ConfigMap > metadata: > name: my-assisted-service-config > namespace: multicluster-engine > data: > ALLOW_CONVERGED_FLOW: "false" > EOF configmap/my-assisted-service-config created hub$ oc annotate --overwrite AgentServiceConfig agent unsupported.agent-install.openshift.io/assisted-service-configmap=my-assisted-service-config