Bug#562737: closed by Luk Claes <luk@debian.org> (Bug#562737: fixed in nfs-utils 1:1.2.2-5)

[Luk Claes <luk@debian.org>]

On 03/29/2011 12:13 AM, Bob Proulx wrote:
> Luk Claes wrote:
>>>   # service nfs-common restart
>>>   Stopping NFS common utilities: idmapd statd.
>>>   Starting NFS common utilities: statd failed!
>>
>> Can you please add 'set -x' on the second line of /etc/init.d/nfs-common
>> and give the output from 'service nfs-common restart' so we get an idea
>> in what step it fails, TIA?
> 
> It gets hidden by start-stop-daemon.
> 
>   + start-stop-daemon --start --oknodo --quiet --pidfile /var/run/rpc.statd.pid --exec /sbin/rpc.statd --
>   + RET=1
> 
> Attached is the full log trace from:
> 
>   # bash -x /etc/init.d/nfs-common start 2>/tmp/nfs-common.bash-x.out
>   Starting NFS common utilities: statd failed!
> 
> Running just the rpc.statd manually gives the same result.
> 
>   # /sbin/rpc.statd
>   # echo $?
>   1
> 
> The /var/log/syslog shows:
> 
>   Mar 28 16:01:45 hysteria rpc.statd[27554]: Version 1.2.3 starting
>   Mar 28 16:01:45 hysteria rpc.statd[27554]: failed to create RPC listeners, exiting

Can you please run rpc.statd with debugging: 'rpc.statd -Fd'? It might
be related to IPv6 (with portmap, switching to rpcbind would help in
that case), though the debug output should tell us for sure.

> Note: I don't use nfs on this machine.  There is no reason for nfs
> statd to start on it.  The /etc/default/nfs-common is unmodified from
> the default package version.  AFAIK nfs-common is getting installed by
> default at system install time to handle possible use of nfs mounts in
> /etc/fstab.  I haven't made any local changes to enable nfs
> subsequently.  I don't have an /etc/exports file.  But in the
> /etc/init.d/nfs-common file the NEED_STATD defaults to yes.

statd is also needed when you mount NFS shares to prevent locking
issues. I don't think it's a wrong default.

Cheers

Luk