Date: Sat, 14 Feb 2026 22:05:56 +0000 From: bugzilla-noreply@freebsd.org To: bugs@FreeBSD.org Subject: [Bug 293183] pwait returns before process actually terminates Message-ID: <bug-293183-227@https.bugs.freebsd.org/bugzilla/>
index | next in thread | raw e-mail
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=293183 Bug ID: 293183 Summary: pwait returns before process actually terminates Product: Base System Version: 15.0-RELEASE Hardware: amd64 OS: Any Status: New Severity: Affects Some People Priority: --- Component: bin Assignee: bugs@FreeBSD.org Reporter: cullum@sacredheartsc.com Perhaps I'm "holding it wrong," but it looks like pwait exits a split-second before the target process has actually terminated. I have a few custom rc.d scripts to handle various daemons. These scripts use the built-in stop/restart functionality of rc.subr, which uses wait_for_pids to wait until the daemon process has fully stopped before trying to start it again. Occasionally (maybe a bit less than half the time?) running `service foo restart` returns an error (during the "start" phase) that my daemon is already running (and shows the old PID, which should be stopped. After some investigation, it looks like pwait is actually returning a split second before the target process actually terminates and is removed from the process table. This appears to be some kind of race condition. I can reproduce fairly reliably. I have a process running under pid 97891: # ps -p 97891 PID TT STAT TIME COMMAND 97891 - SJ 0:01.19 /usr/local/invidious/invidious.git/invidious Now, let's terminate it and use pwait to wait until it's terminated. Right after the pwait returns, we'll run `ps` to show that the process is, in fact, STILL THERE! # kill 97891; pwait 97891; ps -p 97891 PID TT STAT TIME COMMAND 97891 - REJ 0:01.20 /usr/local/invidious/invidious.git/invidious One second later, the process is indeed gone: # kill -0 97891 kill: 97891: No such process Unfortunately this breaks rc.subr's "restart" functionality, because "service foo stop" returns before the daemon is actually terminated, causing "service foo start" to fail. I have hacked around the problem with the following poststop function in my rc.d script: invidious_companion_poststop(){ for i in $(seq 5); do pid=$(check_pidfile "$pidfile" "$procname") if [ -z "$pid" ]; then return else echo "pwait bug..." sleep 1 fi done } -- You are receiving this mail because: You are the assignee for the bug.home | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-293183-227>