NFS Server Not responding error

[Conal O'Kelly]

Hi

I am getting the following error on the system:
"NFS server merc nor responding still trying"

The reason Im sure is because we pfs_mounted a cd which puts an entry in /etc/mnttab "merc:/cdrom /cdrom nfs ........"

Because of an application being temporaily migrated to another server, a DNS change was made so that the local hostname merc resolved (from DNS)to a different server (a very bad idea I know!!)

The mount command shows:
"/cdrom on merc:/cdrom ro,....etc"

I believe as a result of all this when I ran a bdf the command resolved the hostname merc to a different host and as such I got the "NFS server not responding" error. I cannot kill the bdf. Is there any way to kill this process without having to re-boot.

Sorry if its a bit long winded. Be grateful for any help.

Cheers
Con

[Michael Tully]

Hi Con,

The 'bdf' process should actually time
itself out.

At this stage I wouldn't go and reboot
the server. Try to first kill it by using
this.

# ps -ef |grep bdf | awk '{print $2}' | xargs kill -9

Is the process attached to
any particular session, like the console?

-Michael

[Conal O'Kelly]

Hi Mike
Thanks for the response. Unfortunately the bdf is not timing out (Its been running for 2 days). Originally it was attached to my shell process. Once I killed the shell the init process took control of bdf. I have tried kill -9 & a umount on /cdrom but no joy.
For the reasons I mentioned (ie DNS changes) the local system seems to be viewing the /cdrom as a nfs fs. Its also mounted with the nointr option.
Any other ideas??

Thanks.
Con

[Michael Tully]

Hi Con,

hmmm... You could try and change the DNS
so it resolves itself to the local host,
which is best anyway. This is found in
/etc/nsswitch.conf as
hosts: files [NOTFOUND=continue UNAVAIL=return TRYAGAIN=return] dns [NOTFO
UND=return UNAVAIL=continue TRYAGAIN=return]

and

# mv /etc/resolv.conf /etc/resolv.conf.save

Perhaps then you could try and unmount the CD.

Is this system 10.20 or 11?

-Michael

[Conal O'Kelly]

Hi Mike
Sorry I should have metioned that I have changed the host so it resolves from /etc/hosts first.
Anyway it seems everything just hangs, even the umount command is hung & I can't kill it. Umount reports fs is busy when trying to umount, fuser then produces the "NFS server not responding" error & also can't be killed!
So I'm at a bit of a loss. Its looking like a re-boot might be the only option.
System is 11.00.
Thanks for your help Mike, really appreciate it.

Cheers
Con

[harry d brown jr]

Con,

make sure the "merc" entry is not in /etc/fstab. Then shutdown your apps manually, do not try to run shutdown because it will hang trying to shutdown nfs. Then reboot the machine.

live free or die
harry

[Carlos Fernandez Riera]

Also check patch level:


PHKL_23409 1.0 NFS, Large Data Space, kernel memory leak
PHKL_25237 1.0 NFS nfsd deadlock,NFS SIGINT bad text region
PHNE_21376 1.0 ONC/NFS General Release/Performance Patch
PHNE_22642 1.0 ONC/NFS General Release/Performance Patch
PHNE_24034 1.0 ONC/NFS General Release/Performance Patch