When trying to add ARM nodes to my x86 AWS cluster, I found multiple documentation gaps and bugs that I had to overcome or require assistance from the OCP multi-arch team. 
    

4.16
    

Always
    

    1. Provision OpenShift Cluster
    2. Run commands as documented to provision an ARM MachineSet on an x86 cluster
    

Issue 1: Invalid command to get infrastructure ID from cluster infrastructure object
$ oc get -o jsonpath='{.status.infrastructureName}{"\n"}' infrastructure cluster
error: error parsing jsonpath {.status.infrastructureName}{"\n"}, unrecognized character in action: U+201C '"'

Issue 2: For footnote 3, "Specify the role node label to add.", it is not clear if this can be freeform, or if there are specific names for node roles that should be set (or conversely, avoided). "worker" felt like the safest option for my use case, but I chose this from my experience using Kubernetes for 6+ years.

Issue 3: For footnote 4 "Specify a Red Hat Enterprise Linux CoreOS (RHCOS) Amazon Machine Image (AMI) for your AWS zone for the nodes. The RHCOS AMI must be compatible with the machine architecture" (followed by oc command to get AMI image). I think we mean "region" instead of "zone" for AWS here, no? Also worth adding a note that the region ID must be within the quotations. If you don't do this, the jq command fails.

Issue 4: The AWS filters for `securityGroups` and `subnet` need to be updated for OCP 4.16. `securityGroups` filter now has the format "<infrastructure_id>-node", and the `subnet` filter has the format "<infrastructure_id>-subnet-private-<zone>

    

MachineSet can be created, which eventually provisions node(s) for the cluster.
    

Correct command to get infrastructure ID (on my terminal):
$ oc get -o jsonpath="{.status.infrastructureName}{'\n'}" infrastructure cluster