Umount a NFS file system...

[Jonathan Caplette_1]

Hi guys,

I need to umount a NFS filesystem on ServerB. ServerB have mounted a NFS share from ServerA. My ServerA have been shutdown and returned to HP. I forgot to umount the NFS filesystem on ServerB. The thing is now I can't umount it. It says to me :nfs umount: nfs_umount: /hydr : is busy.

I tried "fuser -c /hydr" and I got this error: NFS server not responding still trying. That's normal cause it doesn't exist anymore...

How can I umount a that NFS filesystem??? Without rebooting the server...

Thanks
Jonathan

[Steven E. Protter · Accepted solution]

The NFS server is not responding. Sometimes the only way to clear this up is to boot the client.

You might try /sbin/init.d/nfs.client stop

It might fail.

You might have better luck running exportfs -av on the NFS server, assuming its Unix.

This problem can also be caused by networking problems, bad cables, inability to ping, subnet changes and firewalls between SErver A and Server B

SEP

[Pete Randall]

Jonathan,

In my experience, it's been very rare that I've successfully unmounted in this sort of situation. There ought to be a "-f" switch to the umount command to force it, but there isn't. The only thing you can do is reboot ServerB or bring up another ServerA to temporarily supply the file system in question.


Pete

[Jeff Schussele]

Hi Jonathan,

It's hard to do once the NFS server has gone down. You can try to stop the NFS client on this box
/sbin/init.d/nfs.cliebt stop
but I doubt that will do it.

Surest way is to get the server back up & on-line & then you can umount on the client.

Rgds,
Jeff

[Helen French]

If you can share that file system again frim Server A, that would be best. You can then safely umount the file system.

Otherwise, you may need to reboot the system. Apply the latest NFS patches next time, it might resolve some of these issues.

[John Carr_2]

I dont think you can do anything with a stale NFS mount without the other server.

[Geoff Wild]

Try stopping all NFS services:

/sbin/init.d/nfs.client stop
/sbin/init.d/nfs.server stop
/sbin/init.d/nfs.core stop

If it doesn't work - reboot.

Rgds...Geoff

[Marco Santerre]

My experience in this kind of situation is very similar to everything that has been mentionned already.

1- Try stopping all NFS services on your client.
2- I've also, just for the sake of clearing the problem, rebooted the ServerA just to get the NFS mountpoint up and running again, only to unmount it prior to shutting down the ServerA again.
3- Reboot ServerB

[Sascha Krueger]

you can use an option when mounting (or in fstab) nfs file systems called INT.this options lets you kill those stale mounts with some special signals. (don´t remember them now, but you should find them in the man-pages. it is not -9 or -15).another interesting option is SOFT, HARD. this controls whether the systems keeps trying to mount the nfs filesystem or if it shall stop after doing three unsuccesful trie

[Elmar P. Kolkman]

Though not the nicest way, and I haven't tried it, it should work:
1) add serverA's IP address to an existing server running NFS server
2) export the filesystem that was mounted
3) unmount on serverB (it can now reach the NFS server, though it is another server...
4) unexport the filesystem
5) remove the IP alias

You could even do all steps on serverB if you want...

[Robert-Jan Goossens]

Hi Jonathan,

http://docs.hp.com/hpux/onlinedocs/3929/ForciblyUnmountingNFSFilesystems.pdf

Regards,
Robert-Jan

[Mark Grant]

Might be worth doing soft NFS mounts in the future. It's just a mount option and should time out a broken NFS mount.

It is worth pointing out that an NFS mount should timeout anyway, it might take a few days but it is supposed to.