Implement a way to trigger rollbacks when VM fails on provision phase and reconfigure phase.
Currently when a VM instance provision or reconfigure fails, Morpheus stops any further executions of workflows that might be attached to the instance - using a provisioning type workflow. This means that the action performed by Morpheus (provision or reconfigure) must successfully finish before the respective workflow phase can start.
This behaviour is preventing us from having automated rollbacks.
We would like to see a feature that allows the execution of rollback workflows if the VM instance fails on provision or reconfigure phases.
So this is essentially handled with the teardown phase. If a server is deleted it’ll still attempt the teardown scripts. Now the failure could prevent specific task execution targets from running. I.E domain user on a task but machine failed to join domain.
Additionally, 6.0.0 released with task retry so it gives you the ability to continue on once the issue is resolved.
Also note, if a machine fails you can still set the status as provisioned to force as “successful” to run additional workflows as needed.
I don’t know if this was the original posters intent, however with provisioning, since the teardown handles the rollback, if you could have the ability to automatically trigger a delete on failure, I’m thinking that would be close to this. The reconfigure phase version would probably be better serviced by the retry unless it was impossible.