Guys,
I'm wondering if I need to change the order of the processes in the DAEMONS
line in rc.conf to prevent a hang on reboot/shutdown if smb shares are mounted.
It is almost a 3 minute hang:
Aug 19 16:42:01 providence postfix/qmgr[1316]: A178EA64CD: removed
Aug 19 16:43:13 providence smbd[29587]: [2011/08/19 16:43:13.608240, 0]
printing/print_cups.c:110(cups_connect)
Aug 19 16:43:13 providence smbd[29587]: Unable to connect to CUPS server
192.168.7.15:631 - Connection refused
Aug 19 16:43:13 providence smbd[11057]: [2011/08/19 16:43:13.623014, 0]
printing/print_cups.c:487(cups_async_callback)
Aug 19 16:43:13 providence smbd[11057]: failed to retrieve printer list:
NT_STATUS_UNSUCCESSFUL
Aug 19 16:44:49 providence shutdown[30753]: shutting down for system reboot
Aug 19 16:44:49 providence init: Switching to runlevel: 6
Apparently, the unmount attempt happens immediately after the postfix
shutdown and the total time between the time of hang and time of shutdown is
2:48 seconds which seems like forever. (it's probably actually a 2:30 sec.
timeout, with 18 secs burned somewhere else)
The DAEMONS line I'm using is:
DAEMONS=(hwclock syslog-ng network hal @netfs @ntpd @sshd @crond !avahi-daemon
@mysqld @postfix @cups @samba @apcupsd)
From the error, it looks like smbd attempts to contact cups and gets a
connection refused (I have 1 smb printer defined). I don't know whether this is
a bug or a problem with the shutdown order. If I manually unmount the smb drives
before calling shutdown, then there is no hang. However, I shouldn't need to
manually unmount the smb drives as that should be handled automatically.
I can obviously write a simple script that is called on shutdown that does
the unmount, but I would rather see if I can identify and fix the issue
correctly. Anybody have any thoughts on whether I should move things around in
the DAEMONS line? Anybody else seeing this with mounted smb shares? The shares I
have are these:
//phoenix/config on /mnt/phx-cfg type cifs (rw)
//phoenix/samba on /mnt/phx type cifs (rw)
//phoenix/david on /mnt/phx-david type cifs (rw)
Any thoughts? This isn't something that just started today. It has existed
for a number of months, but I've just lived with it as this box is rarely
rebooted except for kernel updates...
--
David C. Rankin, J.D.,P.E.
