NFS clients can no longer mount shares

[Giuseppe Vacanti]

Possibly the bug is in the server. I cannot decide.

This is on 10.04.

After a reboot NFS clients can no longer mount NFS shares. For instance:

mount -v nfs:/cosine /mnt/
mount: no type was given - I'll assume nfs because of the colon
mount.nfs: timeout set for Wed Jan 4 15:37:15 2012
mount.nfs: text-based options: 'addr=192.168.1.11'
mount.nfs: mount(2): Protocol not supported
mount.nfs: trying 192.168.1.11 prog 100003 vers 3 prot UDP port 2049
mount.nfs: trying 192.168.1.11 prog 100005 vers 3 prot UDP port 39863
mount.nfs: mount to NFS server 'nfs:/cosine' failed: RPC Error: Success

Because if this clients with entries in /etc/fstab or that make use of autofs are just stranded without their disk partitions.

If I give nfsvers=2, as below, the mount succeeds, but I cannot use NFS version 2:

mount -v -o nfsvers=2 nfs:/cosine /mnt/
mount: no type was given - I'll assume nfs because of the colon
mount.nfs: timeout set for Wed Jan 4 15:37:50 2012
mount.nfs: text-based options: 'nfsvers=2,addr=192.168.1.11'
nfs:/cosine on /mnt type nfs (rw,nfsvers=2)

I have clients on 11.10, they can mount but they seem to negotiate a lower NFS version:

mount -v nfs:/cosine /mnt
mount: no type was given - I'll assume nfs because of the colon
mount.nfs: timeout set for Wed Jan 4 15:40:32 2012
mount.nfs: trying text-based options 'vers=4,addr=192.168.1.11,clientaddr=192.168.1.2'
mount.nfs: mount(2): No such file or directory
mount.nfs: trying text-based options 'addr=192.168.1.11'
mount.nfs: prog 100003, trying vers=3, prot=6
mount.nfs: trying 192.168.1.11 prog 100003 vers 3 prot TCP port 2049
mount.nfs: prog 100005, trying vers=3, prot=17
mount.nfs: trying 192.168.1.11 prog 100005 vers 3 prot UDP port 39863
mount.nfs: portmap query retrying: RPC: Program/version mismatch
mount.nfs: prog 100005, trying vers=3, prot=6
mount.nfs: trying 192.168.1.11 prog 100005 vers 3 prot TCP port 45556
mount.nfs: portmap query failed: RPC: Program/version mismatch
mount.nfs: prog 100003, trying vers=2, prot=6
mount.nfs: trying 192.168.1.11 prog 100003 vers 2 prot TCP port 2049
mount.nfs: prog 100005, trying vers=1, prot=17
mount.nfs: trying 192.168.1.11 prog 100005 vers 1 prot UDP port 39863
nfs:/cosine on /mnt type nfs (rw)

While in the latter case the machine is at least usable (but I have problems with ACLs), in the former it is broken

ProblemType: Bug
DistroRelease: Ubuntu 10.04
Package: nfs-kernel-server 1:1.2.0-4ubuntu4.2
ProcVersionSignature: Ubuntu 2.6.32-37.81-generic-pae 2.6.32.49+drm33.21
Uname: Linux 2.6.32-37-generic-pae i686
Architecture: i386
Date: Wed Jan 4 15:30:49 2012
ProcEnviron:
 PATH=(custom, no user)
 LANG=en_US.UTF-8
 SHELL=/bin/bash
SourcePackage: nfs-utils

[Giuseppe Vacanti (gvacanti)]

The nfs-kernel-server init script runs

rpcinfo -u localhost nfs 3

if localhost cannot be resolved for some reason this command will fail, and the server is started with the option

--no-nfs-version 3

even though the kernel could actually support this version.

rpcinfo looks up localhost as specified in nsswitch.conf, and if one has the latter configured as the man page suggests, namely

hosts: dns [!UNAVAIL=return] files

the lookup of localhost will fail.

If NFS v3 is not available, and you have not configured NFS v4, then the clients will negotiate NFS v2 or v1, and these have no acl support. If the clients try to mount using version 4, the mount will fail.

I close the bug as resolved.