Description of problem:

When upgrading ACM from 2.11 to ACM 2.12 on a non-amd64 system (such as arm64), alertmanager crashloops and the upgrade is not fully completed.

Version-Release number of selected component (if applicable):

ACM 2.12.0

How reproducible:

• Always

Steps to Reproduce:

1. Install ACM 2.11 on a non amd64 OpenShift system
2. Setup the observability stack
3. Upgrade ACM to version 2.12.0

Actual results:

• observability-alertmanager-2 container: alertmanager-proxy is crashlooping. It is using image: `quay.io/stolostron/origin-oauth-proxy:4.5` which is a amd64 image only
• The other alertmanager pods do not update to the correct 2.12 proxy image

  Expected results:

• Alertmanager is correctly updated and not crashlooping

  Additional info:

Workaound:

oc scale statefulset observability-alertmanager -n open-cluster-management-observability --replicas=2

note: set replicas to n-1 if number of replicas for the statefulset has been changed from the default (default: 3)