I am building a playbook which upgrades firmware on network switches. These switches are deployed in pairs, and the general procedure goes something like this:
- Remove B switch from service
- Upgrade/Reboot B switch
- Failover to B switch and remove A switch from service
- Upgrade/Reboot A switch
- Failover to A switch and verify.
Seems pretty straightforward, but there are a lot of tasks that make up those those steps and a lot that can go wrong. If I am doing batches of 10 pairs at a time (for example), and a task fails during step 1 for one of the B switches, I need to also remove the corresponding A switch from the rest of the playbook. Is this possible?