I have a set of auto-scaled EC2 instances that run a daemon managed by supervisord on each instance.
I want to differentiate the clean-up behavior for the following cases:
- when shutting down instances (e.g., scaling-in)
i.e., shutting down supervisord itself
-> do an extra clean-up process - when I manually stop the daemon for maintenance (e.g., upgrades)
i.e., keeping supervisord alive but only shutting down my daemon
-> do a fast clean-up process
Currently supervisord has only stopsignal configuration which sends the same signal in both cases. It seems that the authors decided not to support extra reload/restart signals for graceful shutdown.
(In my case, I want to distinguish two shutdown signals depending on the shutdown reason, not wanting to have reload/restart signals, though.)
How could I make my daemons to distinguish those two cases (with supervisord or maybe without supervisord)?
