bdf: NFS server XXX not responding still trying

SOLVED

Hi All

I'm geting the above message running a bdf on a HP-UX 11.00 system.
I know the reason as the NFS server it's trying to reach is no longer on the network. When the NFS server was taken off the network the client still had an NFS mount from the server!!
I can see this in /etc/mnttab as there is an entry for the NFS server, hence the issues with the bdf command.

Is there any way to fix this, apart from a reboot? I have tried the moving the /etc/mnttab, shutting down NFS & fuser -ku /tmp_mnt. I can't do anything the on the NFS server as its no longer on the network.

Thanks for any help.

Cheers
Con

11 REPLIES 11

I believe you can just remove the line (or comment it) from /etc/fstab and /etc/xtab, umount -a and mount -a.

Rob

Thanks for the reply Rob.

There's no entry in /etc/fstab (it's a temporary mount that was forgotten about), though in this instance it would not solve the problem even if there were. The NFS server is no longer on the network (/etc/xtab is a server side file) so there's no possibility of un-exporting the filesystem etc. Any fix has to be done on the client.

Cheers
Con

Sorry - xtab is for exports.

/etc/mnttab and /etc/fstab

I tried to get to my UX machine, but could not.... Our above 2 msgs crossed paths, so I think you answered my reply just before I sent it.

Does work? (dont recall case on the c...)

How about , I believe, for 'force'.

If you've already tried these, sorry - anyhow, I believe I've resolved this before without rebooting...it's just been quite a while.

Good luck,
Rob

NA

Arghhh - just zero point my last 2 replies.

There's no 'c' option for fuser, and there's no 'f' on umount.....a mind is a terrible thing....sorry!

Rob

NA

Do not try using commands
bdf/df

Use mount -v and check the remote mount point with the hostname.

check for the proper hostname and mount point from mount -v.
Then just do a
#fuser -k <>
#umount <>

Doing a fuser on the local mount-point and umount on that will not work!

Hi Con,
# mv /etc/mnttab /tmp/.
# mount -v
should do the trick

you can disable the mounted NFSs and exported file systems via sam.

-d

Which is worse, smoking or picking your nose in a public place?

Thanks for the replies.

Unfortunately none of the suggestions have worked. As I said in my original post I tried recreating /etc/mnttab, stopping NFS and killing the processes (fuser -ku) on the mounted FS.
The main issue is that the NFS server is no longer on the network yet my client believes it still has an FS mounted from there (evidenced by the entry in /etc/mnttab) as it was never unmounted prior to the NFS server being pulled from the network.

Cheers
Con

Hi con

The client NFS mount is a stale NFS

set the flags of NFS_SERVER and NFS_CLIENT to 0 in /etc/rc.config.d/nfsconf
#/sbin/init.d/nfs.server stop
#mount -a

may solve the problem

never give up

Hi Con,

Few times I could get rid of this situation by fooling the system by temporarily enabling a local interface with the NFS Server's IP address. That will make the mount point stale and you can unmount the filesystem then.

For ex., if you have another interface, add the IP address of your NFS server, unmount the stale mount point and unplumb the interface. This may or may not work.

-Sri

You may be disappointed if you fail, but you are doomed if you don't try

Con,

Sri is on the right track. Put another box on the network as an NFS Server exporting the same directory using the same IP as the old NFS Server. The mount point will go stale on the NFS client, but you can umount it at that point.

In fact HP's NFS Guru Dave Olker is creating a doc on how to resolve this very issue, a title will be something like "Forcibly Unmounting NFS Filesystems White Paper".

Hope that gets the job done for you,

-> Brian Hackley

Ask me about telecommuting!