-
Bug
-
Resolution: Done
-
Critical
-
ACM 2.15.0
-
Quality / Stability / Reliability
-
1
-
False
-
-
False
-
-
-
Workload Mgmt Train 33 - 2, Workload Mgmt Train 34 - 1
-
Important
-
None
Description of problem:
After a openshift gitops instance is launched by ACM Gitops Addon, the upstream gitops service and gitops console plugin images are applied to the gitopsService cluster and its gitops-plugin pods respectively.
$ oc describe pods cluster-7d8f55f996-ns5ct -n openshift-gitops | grep Image Image: quay.io/redhat-developer/gitops-backend:v0.0.1 $ oc describe pods gitops-plugin-5df5f4f8cc-n7fzw -n openshift-gitops | grep Image Image: quay.io/redhat-developer/gitops-console-plugin:v0.1.0
It turns out the binaries bundled in the upstream images are not compatible in the Openshift with ppc64le arch
% oc logs -n openshift-gitops cluster-7d8f55f996-ns5ct exec container process `/root/./backend-http`: Exec format error % oc logs -n openshift-gitops gitops-plugin-5df5f4f8cc-n7fzw exec container process `/usr/bin/container-entrypoint`: Exec format error
We should apply the downstream images to the two deployments, which are compatible in the openshift with ppc64le arch.
$ oc describe pods gitops-plugin-577dfbc76-rdd7v -n openshift-gitops | grep Image Image: registry.redhat.io/openshift-gitops-1/console-plugin-rhel8@sha256:fe57b69fc1570f7ea28530a67ceb44a58537dec419e7d5aa77e8e13fa61cccfc $ oc describe pods cluster-5b5c99d9b9-lrt2j -n openshift-gitops | grep Image Image: registry.redhat.io/openshift-gitops-1/gitops-rhel8@sha256:01883ebbe93f97a8196204adcd1749280b697e4c81f80658fdba507301f033aa
Version-Release number of selected component (if applicable):
ACM 2.15
How reproducible:
Steps to Reproduce:
1. import an OCP with ppc64le arch to a ACM hub as a managed cluster
2. deploy a gitops addon to the managed cluster.
Actual results:
The cluster and gitops-plugin fails to start on the managed cluster with ppc64le arch
% oc get pods -n openshift-gitops cluster-7d8f55f996-ns5ct 0/1 CrashLoopBackOff 482 (70s ago) 40h gitops-plugin-5df5f4f8cc-n7fzw 0/1 CrashLoopBackOff 450 (4m18s ago) 40h
Expected results:
The two pods are running status and the downstream images are applied.
// log into the managed cluster % oc get pods -n openshift-gitops NAME READY STATUS RESTARTS AGE cluster-64977d4df8-t79gr 1/1 Running 0 22m gitops-plugin-7cb9588889-k2k2s 1/1 Running 0 59s // check the image used in each pod % oc get pods -n openshift-gitops cluster-64977d4df8-t79gr -o yaml |grep image image: registry.redhat.io/openshift-gitops-1/gitops-rhel8@sha256:01883ebbe93f97a8196204adcd1749280b697e4c81f80658fdba507301f033aa % oc get pods -n openshift-gitops gitops-plugin-7cb9588889-k2k2s -o yaml |grep image - image: registry.redhat.io/openshift-gitops-1/console-plugin-rhel8@sha256:fe57b69fc1570f7ea28530a67ceb44a58537dec419e7d5aa77e8e13fa61cccfc
Additional info:
Note: Both images - GitOpsServiceImage and GitOpsConsolePluginImage are for creating the gitops service and its gitops console plugin. Since they are not used by our ArgoCD pull model, currently we just define them as internal variables with default value, they are not exposed to the gitopsCluster api for end user customization. Both image tags will be upgrade along with every major ACM GA.