NFS client hangs="NFS server not responding still trying"

[Santhosh.H]

"NFS server not responding" how do u get rid of it. It does not even allow to issue the cmd to kill it

[Frederic Sevestre]

Hi,

May you try to use an other connexion to stop the nfs deamon on your server.

# /sbin/init.d/nfs.client stop

I guess your NFS server is down.

Fr??d??ric

[sven verhaegen]

HI

The problem you are seeing is related to the deamon NFSD on the nfs server not being able to reply to a clients request , this doesn't actually need to mean it is down however , several issues can cause the problem ...

first the network connection between the NFS client and the NFS server could be ruptured , this can be easily checked by performing a ping test

another possible issue is the time delay on the network, this could be this big that we get timeouts on the requests , as the connection of NFS is UDP oriented this means it upper layer application will try some time to resend the request but if that fails it will simply stop and generate the 'server not responding' reply , the only sure way to know this is by network tracing the problem whith nettl or with a lan analyser

third it could be the system is so busy with request that the NFSD's on the server cannot handle all the requests , that results also in not replying to some and if it happens often enough then we get server not respondings, when having this type of problem it could be you have multiple mounts and several mountpoints which are loading the standard number of NFSD's on the system , usually only 4 NFSD's run on a nfs server , by setting the server file /etc/rc.config.d/nfsconf parameter NUM_NFSD=4 to a higher number for instance 16 and relaunching this server you could get better results , however this is only a solution when it is the NFSD's themselfs being out of breath and not able to answer it could also be related to the system resources being to log e.g. disc access times or CPU usage , then this operation would be useless

it could also be that you indeed have a real problem with the NFS server itself , the deamons being in trouble , as several problems are know within NFS at HP a series of patches are advised , starting with the ONC/NFs megapatch and NFS kernel patches , these patches have several requirement patches , a client and server in NFs environments should be equiped with latest versions of these patches to avoid known hang situations

there are some other possible causes like for instance change of names in DNS or NIS causing the actual NFS server name to have changed , stale filehandles and some other... but they are less likely to happen in en environment that is stable.

as I don't know from which situation you started I agree with the previous entry that the best action course would be to stop the mount on the client and to check if any other client has the same problem to the same server , if that is the case you should try to restart the NFS server which hold the mount origin for these clients , if the mount cannot be halted a reboot of the client is in order but always check first the server does reply to other systems requesting to mount the same filesystem....

[Sridhar Bhaskarla]

Santosh,

If you are not using NFS, then it is better to disable NFS. Even it is good for security purposes.

First do a /sbin/init.d/nfs.server stop
/sbin/init.d/nfs.client stop

to make it permanent

Edit your /etc/rc.config.d/nfsconf and make NFS_CLIENT and NFS_SERVER = 0

If you are using NFS, then you need to look at your NFS server to see if nfsd has been stopped or if the file systems were unexported there.

-Sri