Following the migration of the server, postgrey didn't restart and this wasn't managed by ansible, nor monitored. This did block mail delivery (no mail should have been lost however). Postgrey should be monitored, and managed.
So, notification was added, and I think it is also managed properly now.