Thank you for the PR, I'm happy to see a first green run in a reasonable amount of time!

One overall comment: It seems to me CLUSTER_NAME is set to keycloak_$(whoami). IMHO this would lead to conflicts if two workflows would run in parallel, for example when they try to create a SSH key which is then already present. I remember from previous attempts that the playbook will then not write the key to disk, and due to that the next ssh command will fail.

So AFAIK we'll need a an identifier which is unique to the run. Maybe we can use an (abbreviated?) GitHub commit ID for this.

See below for some comments on specific code snippets.

Hi @ryanemerson - thank you for this PR which is a huge step towards our goal!

It is good to get it merged now, and we'll get more feedback by doing this.

When reviewing this, two things should be fixed ASAP once this has been merged:

• Maven is run without the batch flag -B which creates a lot more output than needed.
• The Aurora tasks should be skipped if now AWS token is available to avoid errors on forked repositories. Otherwise it would annoy fellow team members.

I've added this to issue #26967