Processes stuck in kernel during NFS mount

Thierry Danis (danis@mail.dotcom.fr)
Fri, 18 Jun 1999 00:33:04 +0200


Hello,

Configuration : RH 6.0.

I have a problem using NFS which happens quite frequently these days
(with 2.2.9-ac4 and 2.2.9 with NFS V3 support. Can't say about other
versions).

Sometimes some of our NFS servers are rebooting (purposely). They
are Solaris and/or SunOS (for the case it matters).

Our Linux clients are using automount (kernel version) to mount NFS
partitions on demand.

When the mail server is rebooting, the /var/spool/mail directory
(which is automounted from the server) sometimes gets unmounted
or remounted (depending on whether my still running mutt client
is doing some request or not). At that time, the mount request remains
stuck in the kernel, and every subsequent umount/mount request gets stuck
as well, even after the server has come again to life.

At the time of the writing, I have the following situation when I
attempt to log on my client machine :

# ps auxww | grep mount
root 477 0.0 0.3 1144 244 ? S Jun15 0:00 rpc.mountd
root 800 0.0 0.3 1292 220 ? S Jun15 0:00 /usr/sbin/automount /a file /etc/auto.liste rw,soft,bg,intr,timeo=20,wsize=8192,rsize=8192,actimeo=1
root 17105 0.0 0.3 1296 224 ? S Jun17 0:00 /usr/sbin/automount /a file /etc/auto.liste rw,soft,bg,intr,timeo=20,wsize=8192,rsize=8192,actimeo=1
root 17106 0.0 0.7 1104 460 ? D Jun17 0:00 /bin/umount /a/var.spool.mail
root 17152 0.0 0.6 1296 420 ? S 00:08 0:00 /usr/sbin/automount /a file /etc/auto.liste rw,soft,bg,intr,timeo=20,wsize=8192,rsize=8192,actimeo=1
root 17153 0.0 0.8 1140 544 ? D 00:08 0:00 /bin/mount -t nfs -s -o rw,soft,bg,intr,timeo=20,wsize=8192,rsize=8192,actimeo=1 obole:/u0 /a/obole.u0
root 17166 0.0 0.6 1296 420 ? S 00:11 0:00 /usr/sbin/automount /a file /etc/auto.liste rw,soft,bg,intr,timeo=20,wsize=8192,rsize=8192,actimeo=1
root 17167 0.0 0.8 1140 544 ? D 00:11 0:00 /bin/mount -t nfs -s -o rw,soft,bg,intr,timeo=20,wsize=8192,rsize=8192,actimeo=1,noac mbox:/var/spool/mail.export /a/var.spool.mail

Here, we have 3 processes in a 'D' state : 1 umount and 1 mount of the
mailbox, and 1 mount of a home directory. There is no way to kill the
mount/umount processes. Rebooting the machines hangs forever when
it tries to umount the partition.

That is rather annoying !

A+,

-- 
	Thierry Danis

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/