[236] in Zephyr_Bugs
[jfc@ATHENA.MIT.EDU: zhm, afsd]
daemon@ATHENA.MIT.EDU (Jonathan I. Kamens)
Thu Jan 31 19:10:39 1991
Date: Thu, 31 Jan 91 18:06:16 -0500
From: "Jonathan I. Kamens" <jik@pit-manager.MIT.EDU>
To: bug-afs@ATHENA.MIT.EDU, bug-zephyr@ATHENA.MIT.EDU, rel-eng@ATHENA.MIT.EDU
afs: any chance of making afsd notice signals?
zephyr: any chance of doing the locking or something that jfc
suggests?
rel-eng: any chance of removing zhm.pid in /etc/rc?
----- Forwarded message
To: bugs@ATHENA.MIT.EDU
Subject: zhm, afsd
Date: Thu, 31 Jan 91 16:40:42 EST
From: John Carr <jfc@ATHENA.MIT.EDU>
Short version: I worked long enough in single user mode that the zhm
pid file from last reboot held the pid of an AFS daemon. This caused
zhm to hang trying to kill it.
Long version:
There are 2 bugs.
1. afsd should not ignore signals. I should be able to kill AFS daemons and
shut down AFS without crashing the machine. I understand that CMU/Transarc
is unwilling or unable to write robust code; if they don't fix this we
should.
2. zhm should not try forever to kill the old daemon. There are a few ways
to work around this. Two safe ones are (1) have /etc/rc remove
/etc/athena/zhm.pid, (2) have zhm to keep the file open and flock()ed (in
this case, zhm should try to lock the file before reading a PID).
----- End of forwarded message
