On Tuesday, 27 May 2025 at 15:06:14 UTC, Andrea Fontana wrote:
In a production environment, how do you handle deployment
without killing ongoing requests?
Exactly in this way. Just deploy the new executable and restart
workers sending a SIGUSR1 signal to daemon. That's it. Workers
will be restarted softly without any hiccup.
Andrea
The situation is more complicated if you also want to restart the
daemon due to a configuration change or a code update in daemon.
In that case, since serverino usually runs under nginx, it's
better to start another instance of serverino on a different port
and use nginx to redirect requests to the new process without any
loss.
In any case restart of serverino it's really quick!
Andrea
