| Subject: | SIGQUIT vs SIGKILL |
You absolutely definitely 100% want ...->stop() to kill the forked server, so I think it would make sense to send the unblockable SIGKILL instead of SIGQUIT.
I ran into this because for some reason our CI infrastructure at work appears to set sigprocmask and block SIGQUIT.
Can I suggest changing the signal? Or adding a 'stop_hard' or similar method that sends SIGKILL?