Application rollback with ansible

Hi!

I was hoping to canvas opinion on how people using ansible handle rollbacks. We’re migrating a large multihost multi-app project and have have two competing opinions:

  1. Utilise idempotency and simply redeploy a previous working set on a failure (which should only deploy the systems which have changed versions)
  2. Maintain historical versions of applications on hosts and use symlinks to rollback.

What is your experience or preference?

Thanks,

Jack

Look to the new v2 engine which has 'exception handling' built into task blocks.

when will v2 be stable?