This incident has been resolved. One of our server (dedicated to automation execution) was unreachable since Sunday morning (7:18am). The automations started on this machine failed. This server should have restarted automatically but it didn't. We need to fix this issue.
Meanwhile, our load-balancing system spread the work on other machines and most of the automation jobs did run. But this Monday morning, as there were too many automation jobs for a reduced number of servers, the automations may have been delayed for a few hours.