NFS4 home mounts owned by nfsnobody

[Braden McDaniel]

On Wed, 2009-06-24 at 09:19 +0100, John Austin wrote:
> On Wed, 2009-06-24 at 00:10 -0400, Braden McDaniel wrote:
> > On Tue, 2009-06-23 at 10:13 +0100, John Austin wrote:
> > > On Mon, 2009-06-22 at 18:38 -0400, Braden McDaniel wrote:
> > > > I'm trying to set up automounting of home directories using NFS4 and 
> > > > autofs.  I seem to have it working except that all of the files in the 
> > > > mounted directories have their owner and group set to "nfsnobody".
> > > > 
> > > > /etc/exports on the server looks like this:
> > > > 
> > > > /exports *(ro,fsid=0)
> > > > /exports/share *(rw,sync,nohide)
> > > > /exports/home *(rw,insecure,sync,nohide)
> > > > 
> > > > /etc/auto.home on the client looks like this:
> > > > 
> > > > braden		-fstype=nfs4	hinge:/home/braden
> > > 
> > > I had the same problem (F11)
> > > The cure for me was to ensure that the name of the local machine (naxos) is
> > > available to "mount"
> > > Putting the name in /etc/hosts on naxos does the job
> > > 
> > > naxos ~ 2# cat /etc/hosts
> > > 127.0.0.1   localhost localhost.localdomain localhost4 localhost4.localdomain4 naxos
> > > ::1         localhost localhost.localdomain localhost6 localhost6.localdomain6
> > > 
> > > It is not good enough that hostname is set
> > 
> > Are you quite certain that's what did it for you? This doesn't appear to
> > be working for me. :-/

[snip]

> Just run a couple of checks with one of my machines
> 
> 1. Removed name from /etc/hosts and rebooted
>    nfsnobody appears everywhere on an nfs mount
> 
> Error messages such as these in /var/log/messages
> 
> Jun 24 08:56:47 monk rpc.idmapd[1501]: nss_getpwnam: name 'ja at localdomain' does not map into domain 'jaa.org.uk'

I have a similar error message:

        Jun 24 13:49:57 bolt rpc.idmapd[2481]: nss_getpwnam: name 'braden at hinge.endoframe.net' does not map into domain 'endoframe.net'

hinge is my NIS server machine (as well as the NFS server); this error
message occurs on the client.  Adding an entry for hinge to /etc/hosts
does not appear to have changed this.

        # cat /etc/hosts
        127.0.0.1   localhost localhost.localdomain localhost4 localhost4.localdomain4 bolt bolt.endoframe.net
        ::1         localhost localhost.localdomain localhost6 localhost6.localdomain6
        192.168.1.21 hinge hinge.endoframe.net

> 2. Replaced name in /etc/hosts and rebooted - problems disappear
> 
> I assume that ja at localdomain rather than ja at jaa.org.uk is being passed to rpc.idmapd which it is unhappy with
> The entry in /etc/hosts definitely fixes this for me
> I guess your problem must be different

I guess so. :-/

Thanks for going to the trouble to confirm this.

> ------------------------------------------------------
> Is  rpcidmapd/rpc.idmapd  running correctly ?

Looks like it:

        # service rpcidmapd status
        rpc.idmapd (pid 2551) is running...
        # ps -ef | grep -i rpc
        rpc       2485     1  0 14:09 ?        00:00:00 rpcbind
        rpcuser   2498     1  0 14:09 ?        00:00:00 rpc.statd
        root      2533     2  0 14:09 ?        00:00:00 [rpciod/0]
        root      2534     2  0 14:09 ?        00:00:00 [rpciod/1]
        root      2535     2  0 14:09 ?        00:00:00 [rpciod/2]
        root      2536     2  0 14:09 ?        00:00:00 [rpciod/3]
        root      2537     2  0 14:09 ?        00:00:00 [rpciod/4]
        root      2538     2  0 14:09 ?        00:00:00 [rpciod/5]
        root      2539     2  0 14:09 ?        00:00:00 [rpciod/6]
        root      2540     2  0 14:09 ?        00:00:00 [rpciod/7]
        root      2551     1  0 14:09 ?        00:00:00 rpc.idmapd
        root      3814  3762  0 14:12 pts/0    00:00:00 grep -i rpc

Please let me know if you have any further insight.

-- 
Braden McDaniel <braden at endoframe.com>