Re: [gentoo-user] System hangs when poweroff with NFS mounted

[Canek Peláez Valdés]

On Fri, Jan 31, 2014 at 7:04 PM, Amankwah <amankwah7@×××××.com> wrote:
> On Fri, Jan 24, 2014 at 11:24:46AM -0600, Canek Peláez Valdés wrote:
>> On Fri, Jan 24, 2014 at 9:45 AM, 刘洋 <amankwah7@×××××.com> wrote:
>> > journalctl --boot=-1
>> > -- Logs begin at Tuesday 2013-12-24 21:48:33 CST, end at Friday
>> > 2014-01-24 22:38:38 CST. --
>> > 1月 24 21:54:22 diamond systemd[3061]: Failed to open private bus
>> > connection: Failed to connect to socket
>> > /run/user/1000/dbus/user_bus_socket: No such file or directory
>> > 1月 24 21:54:22 diamond systemd[3061]: Mounted /sys/kernel/config.
>> > 1月 24 21:54:22 diamond systemd[3061]: Mounted /sys/fs/fuse/connections.
>> > 1月 24 21:54:22 diamond systemd[3061]: Stopped target Bluetooth.
>> > 1月 24 21:54:22 diamond systemd[3061]: Stopped target Sound Card.
>> > 1月 24 21:54:22 diamond systemd[3061]: Starting Default.
>> > 1月 24 21:54:22 diamond systemd[3061]: Reached target Default.
>> > 1月 24 21:54:22 diamond systemd[3061]: Startup finished in 419ms.
>> > 1月 24 21:54:31 diamond pulseaudio[3293]: [pulseaudio] pid.c: Daemon
>> > already running.
>> > 1月 24 21:54:31 diamond pulseaudio[3296]: [pulseaudio] pid.c: Daemon
>> > already running.
>> > 1月 24 21:54:31 diamond pulseaudio[3299]: [pulseaudio] pid.c: Daemon
>> > already running.
>> > 1月 24 21:54:31 diamond pulseaudio[3301]: [pulseaudio] pid.c: Daemon
>> > already running.
>> > 1月 24 21:55:01 diamond sudo[3362]: amankwah : TTY=pts/4 ;
>> > PWD=/home/amankwah ; USER=root ; COMMAND=/bin/sh
>> > 1月 24 22:34:32 diamond systemd[3061]: Stopping Default.
>> > 1月 24 22:34:32 diamond systemd[3061]: Stopped target Default.
>> > 1月 24 22:34:32 diamond systemd[3061]: Starting Shutdown.
>> > 1月 24 22:34:32 diamond systemd[3061]: Reached target Shutdown.
>> > 1月 24 22:34:49 diamond pulseaudio[3279]: [pulseaudio] core-util.c:
>> > Failed to create secure directory (/run/user/1000/pulse): No such file
>> > or directory
>> >
>> > These are the output of the command on my system after hangs. but why
>> > I saw nothing about my NFS directories?
>> >
>> > I added the script to the path, and the system shutdown correctly, but
>> > the time is too long, every NFS directory complained the timeout due
>> > to and then umounted, maybe it took about 10 minutes for choosing the
>> > shutdown under desktop to the system poweroff automatically.
>>
>> 10 minutes looks awfully long to me... and then I saw again your
>> fstab. You explicitly set a timeout of 10 milliseconds; but what if
>> there is a bug in systemd that parses 10ms (milliseconds) as 10m
>> (minutes)?
>>
>> Just to try, change your fstab to set x-systemd.device-timeout=5;
>> without a suffix, so the time specified should be in seconds.
>>
> It seems that takes my more than 10 minutes, too. I don't know why.
>
> And when I saw the shutdown screen, the WPA_** process was killed before
> the the NFS umounted. Maybe it the reason why all of the NFS filesystem
> umount process were running after the wireless connection closed.
>
> Maybe that's a BUG?

Or an order error in the related service unit files. OpenSuse has
something similar in [1] and [2].

[1] https://bugzilla.novell.com/show_bug.cgi?id=849387
[2] https://bugzilla.novell.com/show_bug.cgi?id=857031

You can override a unit file in /usr/lib by creating one in /etc with
the same name. Perhaps playing with After= and Requires= in the NFS
units will solve this.

I haven't used NFS in a long time. Are you using the service files
provided by the packages, or you got them from somewhere else?

Regards.
-- 
Canek Peláez Valdés
Posgrado en Ciencia e Ingeniería de la Computación
Universidad Nacional Autónoma de México