Hi,
I’m looking at a project where changes have to be rolled out to a number of hosts, tested and if something is wrong - the previous configuration is restored. The trick is that testing can only happen after the change has been rolled out to all machines (as they provide a single service). I’m trying to get this into a single playbook, but I’m not sure how to approach this. The testing has to be done from another machine, and if it fails - all the affected machines must be changed, so the flow of changes looks like this:
- change:
machine A
machine B
machine C - run tests using machine D (a few different tests)
- finish if all good
- if failed roll back
machine A
machine B
machine C - run tests using machine D (to confirm rollback worked)
Any suggestions on the approach here? A single playbook with multiple plays in it?
Thx,
kind regards
Pshem