Server hangs while running bdf command

[Adithyan]

Hi,

When ever issuing a bdf command the server hangs. The scenario i that, There was a NFS mount file system in the server and the server where NFS file system was exported is no more.

The entries are there in the fstab and mnnttab. We have tried commenting the entry from fstab and tried again but it didn't help. We have renamed the mnnttab and ran mount -a. But still the entries are there in the mnnttab.

Pls help

[Wim Rombauts]

You have a stale NFS filesystem. bdf will continue, but it will take lots of time.

If I am not mistaken, man mount_nfs wil show a "force" option to the umount command, but for the umount to work correctly, you need to complete /etc/mnttab with the NFS filesystem in it.

[Wim Rombauts]

What OS-version are you on ?
My HP-UX 11i v2 system has :
umount -f
This can be a rather new option, so if you are running HP-UX 11i v1, this option could not be available.

[Rasheed Tamton]

Hi,

bdf -l should work for local file systems.

What do the below cmds show:
nfsstat
mount -v
showmount -a servername
rpcinfo -p servername
or
try restarting the nfs services

- things to try if you want
remove the old server entry from /etc/hosts (if the default name service is files)
-remove the arp table entry of the old server if it is there (arp -a then arp -d)
- rpcinfo -d prog ver

You said the old server is no more - if it is there you can remove the entry from /etc/xtab, /etc/exportfs and /etc/rmtab from the server.

I am not sure about any latest patches relates to this problem.

The last option you know - wait until the next reboot!!!

[Jollyjet]

HI better go to /usr/sbin/init.d and stop the nfsclient and try the bdf command.

./nfs.client stop

if you want to start

./nfs.client start

if it gives you the error again stop the server also.

./nfs.server stop

./nfs.server start.

[Bill Hassell]

It's not just bdf that will hang...you'll see new logins hang and anything that scans all the filesystems or tries to look at the stale mountpoint. This is a 'feature' of NFS which can be a minor or major problem on a production system. If you use NFS in production, the NFS server can never go down or lose communication. If this isn't possible to control, don't use NFS. Tools such as scp and rsync can transfer data to/from other systems without causing hangs.

[Jollyjet]

hey its not /usr/sbin

its just sbin/init.d

[Adithyan]

Greetings !

We have tried stopping and starting the nfs client service and removed the server entry from the /etc/hosts. Still the problem is there.

[Adithyan]

Hi,

We have taken a scheduled outage and restarted the server and the problem seems to be sorted out now.

Thanks.

[Wim Rombauts]

For a possible next occurance : Check then manpage of umount_nfs for support of the -f option.