I have NFS server that is working as a general filer in a LAN.

Recently due to disk malfunction, I lost everything there, so I got new disks, reinstalled newest Debian (jessie), copied back config files - and NFS no longer works?!

my /etc/exports contains:

/media/media *(rw,sync,no_subtree_check,all_squash,anonuid=31244,anongid=31451)
/media/private *(rw,sync,no_subtree_check,all_squash,anonuid=16830,anongid=10818)

I did couple of reboots, /etc/init.d/nfs-common restart, /etc/init.d/nfs-kernel-server restarts.

On client side, I have, in /etc/fstab:                  /media/MEDIA      nfs     rsize=8192,wsize=8192,timeo=14,soft,intr,rw,user,noatime,nodiratime  0       0                /mnt/nfs/private  nfs     rsize=8192,wsize=8192,timeo=14,soft,intr,rw,user,noatime,nodiratime  0       0

When mounting the filesystem, mount command takes long time, and ends without printing anything. Running "df" later on shows:

df: ‘/media/MEDIA’: Stale NFS file handle

In syslog on client I see:

2013-12-18T15:27:08.319605+01:00 h3po4 kernel: [  363.398977] RPC: AUTH_GSS upcall timed out.
2013-12-18T15:27:08.319619+01:00 h3po4 kernel: [  363.398977] Please check user daemon is running.

But I have no idea what daemon it talks about, why it even needs some "user daemon", and why is there AUTH_GSS call - I want access to be 100% open to the nfs share within LAN.

I verified that it's not a firewall issue (by disabling firewall for a test).

I don't think I ever reconfigured anything else on the server, before disk crash, but I'm not sure - the machine has been working for > 4 years, without a problem, so I could have changed something and forgot about it.

So, I checked some more things.

Before mounting on client, mount directory is:

drwxr-xr-x  2 root root 4096 maj  1  2013 MEDIA/

But after "mount /media/MEDIA", waiting for it to finish, I get:

ls: cannot access /media/MEDIA: Stale NFS file handle
d?????????? ? ?    ?       ?            ? MEDIA/

The directory on server side is always "normal":

drwxrwxrwx 2 media media 6 Dec 18 14:05 /media/media/

also, exportfs works fine:

=# exportfs -v
/media/media        <world>(rw,wdelay,root_squash,all_squash,no_subtree_check,anonuid=31244,anongid=31451,sec=sys,rw,root_squash,all_squash)
/media/private      <world>(rw,wdelay,root_squash,all_squash,no_subtree_check,anonuid=16830,anongid=10818,sec=sys,rw,root_squash,all_squash)

and I can see the exported filesystems from client:

$ showmount -e blob
Export list for blob:
/media/private *
/media/media   *
  • You accidentally the whole question. – EEAA Dec 18 '13 at 14:39
  • Sorry - problem with firefox, unrelated :) – user195086 Dec 18 '13 at 14:43
  • What is the contents of /etc/default/nfs-kernel-server ? – sciurus Dec 18 '13 at 15:54
  • Can you confirm rpcbind is running and that your iptables allow connections to rpcbind? Does your client kernel support the nfs version being used by your nfs server? – Jason Zhu Dec 18 '13 at 16:11
  • All seemed to be working. Got a guy on IRC to help me by adding ",fsid=2" to options for /media/media, and ",fsid=3" to options for /media/private. Afterwards, I can actually mount the filesystems, though it takes ~ 15 seconds for the first mount to work. – user195086 Dec 18 '13 at 17:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.