On Tue, 29 Jan 2013 18:01:59 +0100, Paul Marwick <paul.marwick@xxxxxxxxx>
wrote:
Martín Cigorraga wrote:
On Sat, Jan 19, 2013 at 5:55 AM, Paul Marwick
<paul.marwick@xxxxxxxxx>wrote:
I've got two machines in my local network that provide NFS shares. One
is
an old Raidsonic NAS (running a current version of OpenWRT), the other
a
dedicated server running Slackware). I normally mount the shares
manually
(the machine that is giving me a problem is a laptop, so its often not
connected to my own network).
The problem comes when I try to shut the laptop down (or restart it).
Unless I've manually unmounted the NFS shares, the machine will hang
and
fail to shut down. In some instances, I've been able to use CTRL-C to
break
out of the stall, but frequently, I can only turn the machine off,
which is
not exactly ideal.
I've been playing with the nfs exports, currently have the following:
/mnt/sda2/stor 192.168.1.0/255.255.255.0(rw,**
sync,no_wdelay,nohide,no_root_**squash)<http://192.168.1.0/255.255.255.0(rw,sync,no_wdelay,nohide,no_root_squash)>(this
is on the Raidsonic)
and
/mnt/sda2/stor 192.168.1.0/255.255.255.0(rw,**
sync,no_wdelay,nohide,no_root_**squash)<http://192.168.1.0/255.255.255.0(rw,sync,no_wdelay,nohide,no_root_squash)>on
the Slackware server.
The problem is specific to Arch - I've got two other distros installed
on
the laptop (though Arch is my usual OS). From SalineOS or Salix, the
laptop
will shut down or restart without having to manually unmount the NFS
shares
prior to shutdown.
I suspect I've not got something set up correctly in systemd, but I'm
currently at a loss to work out what. If I manually unmount the shares
before shutdown, there is no problem. I've also tried adding the
unmount to
/etc/rc.local.shutdown, but that doesn't work - the system still hangs
during shutdown. I'm sure that rc.local.shutdown is being executed -
if I
shut down without the shares mounted, I can see messages that they are
not
mounted.
I'm not sure if I should be looking at the NFS export parameters on the
hosts, or systemd on the laptop, though I've tried some variations on
the
exports without changing the behaviour.
Does anyone know what I should be looking at to solve this?
Paul.
Same issue here with NFS4 (both laptop and NAS are Arch), if I find
what's
happenings I post back.
I've had a suggestion that the problem may be due to NetworkManger - if
the network goes down before the NFS shares are unmounted, the sort of
hang that I'm seeing will happen.
I am using NetworkManager, so it does seem to be a possibility. I'm not
sure how to check at what point in the shutdown cycle the network is
disabled. I have a couple of ways of testing it (one possibility may be
to try a wired network, which should connect and disconnect
independently of NetworkManager), so I'll post back if I can prove or
disprove the theory.
The other option is to replace NetworkManager with wicd (which, so far
as I can remember does not disconnect the network when the desktop is
shut down).
I will do some testing and post back the results. It would be
interesting if anyone knows how the shut down sequence works, especially
when the network is disconnected.
Paul.
If you look in the <journalctl -b | grep -i systemd> log I think you will
see that the NSF will be mounted early in the boot process and
network.target gets started late in the boot because renaming occurs so
late. When the system goes down the process is reversed and network.target
stops early in the shutdown and the network is down long before nfs is
unmounted.
I think Sudaraka's solution may work.
--
ron9
