As we get more into the development of Multi-Arch Compute  and as more teams become aware of it, there are questions which come up around usability. While we do need additional customer input for a complete list, we have identified some areas to focus on:

• Determine a cluster's multi-arch readiness:
  Demo for upgrading a single arch cluster to multi: https://drive.google.com/file/d/1YID2OP2CJwkSFRL3f2hI5UXzDSkytwVX/view?ts=64552773
  Cannot tell from the console if the cluster is capable of having multiple architectures.
  Console doesn't display what the nodes' arch is.

• Inconsistent behavior w/multi payload vs single arch payloads in the following binaries:
  oc adm release info: can only search amd64 release graph. non-x86 releases must be inspected (and mirrored) by specifying the exact payload.
  oc-mirror: Defaults to amd64 https://issues.redhat.com/browse/ARMOCP-554
  openshift-installer: Will default to installing the arch it was built with, cannot 'select' an arch at install time. Users much edit the install_config.yaml if they want to install non-x86 with the multi payload https://issues.redhat.com/browse/MIXEDARCH-231 

• Scheduling workloads, operators or builds:
  OLM and console reflecting available operators. semi-related operator scheduling: https://issues.redhat.com/browse/MIXEDARCH-140
  Default samples/image streams have to be manually configured to import all images in a manifest list.
  Unclear how to purge nodes if a container of the wrong arch is pulled.
  Builds needs clearer instructions on how to schedule across nodes of different archs. What should the default behavior be?
  Is it appropriate to auto-taint mixed arch worker nodes? if so, when, and how do we determine which arch to taint.

• Telemetry:
  what metrics are exposed for heterogeneous clusters, and will existing telemetry data reflect heterogeneous cluster data.
  https://issues.redhat.com/browse/MIXEDARCH-136

AC:

• Work with the UX team and to evaluate priority of the above items, and create additional items to be addressed.
• Create a list of epics addressing the UX issues.
• Determine the timeline in which the identified issues can be addressed.