[lug] Sun Client Redhat AS3 NFS Probs

[Jeffrey A. St. Pierre]

On Thu, 13 May 2004, D. Stimits wrote:

> Jeffrey A. St. Pierre wrote:
> > Ladies and Jellymen,
> > 
> > Sanity is a fleeting thing and I am certainly about to lose mine.  
> > I have encountered a most perplexing error between a Redhat AS3
> > NFS server and a Sun Solaris 8 Client.  Essentially the sun
> > accessing the fileserver works, but causes the server to spool
> > errors like such:
> > 
> > May 13 11:28:49 <server> rpc.rquotad: Can't find filesystem
> > mountpoint for directory /export/rd01/stack
> > May 13 11:28:49 <server> rpc.rquotad: No correct mountpoint
> > specified.
> > May 13 11:28:50 <server> rpc.rquotad: Can't find filesystem
> > mountpoint for directory /export/rd01/stack
> > May 13 11:28:50 <server> rpc.rquotad: No correct mountpoint
> > specified.
> > May 13 11:28:51 <server> rpc.rquotad: Can't find filesystem
> > mountpoint for directory /export/rd01/stack
> > May 13 11:28:51 <server> rpc.rquotad: No correct mountpoint
> > specified.
> > 
> > 
> > As far as I can tell, mounting the NFS share on another RH Ent3
> > box does not cause these errors.  which means there is something
> > about the mount requests from the Sun that causes it to spew this
> > error.
> > 
> > Here are the settings on the Redhat Server:
> > 
> > /etc/exports=
> > 
> >   /export/rd01 128.138.xxx.0/255.255.255.0(rw,async)
> >   /export/rd02 128.138.xxx.0/255.255.255.0(rw,async)  
> > 
> > chkconfig --list
> > 
> >   nfs             0:off   1:off   2:off   3:on    4:on    5:on
> >                   6:off
> >   nfslock         0:off   1:off   2:off   3:on    4:on    5:on
> >                   6:off
> > 
> > And the settings on the Solaris 8 client:
> > 
> > /etc/auto_master=
> > 
> >   # Master map for automounter
> >   #
> >   +auto_master
> >   /projects               /etc/auto_proj  -rw,bg,intr,soft
> >   
> > /etc/auto_proj=
> >  
> >   # Home directory map for automounter
> >   # RCS: {$Id: auto_proj,v 1.161 2004/05/11 20:10:46 root Exp $}
> >   #
> >   +auto_proj
> >   stack   -rw,rsize=8192,wsize=8192 \
> >           <server>.Colorado.EDU:/export/rd01/&
> > 
> > uname -a output:
> >   SunOS <client> 5.8 Generic_108528-27 sun4u sparc SUNW,Ultra-4
> > 
> > 
> > Generally the errors mentioned above spool continuously while
> > jobs that access the mount are running.  However, I have been
> > able to run 'du -sm *' on all the sub-dirs from the client side
> > without generating errors.  I have executed 'dd if=/dev/zero
> > of=/projects/stack/testfile bs=16k count=61384' and seen no ill
> > results.  sync or async in the /etc/exports doesn't change
> > anything... I can't seem to find what is actually placing the
> > call from the Sun client to the rquotad on the RH server.
> > 
> > Any ideas...
> ...
> 
> I see that quota is being mentioned in the errors. You might want to 
> compare on the working versus failing systems kernel options for quota, 
> quota tools, and how quota is set up. Finally, you might want to check 
> that they all use the same underlying filesystem type, as not all 
> filesystems support quota in the same way (e.g., XFS supports full ACL's 
> but only with the right kernel options and tools).
> 
> D. Stimits, stimits AT comcast DOT net
> 
> 
> 

D. Stimits, 

Thanks for the reply, but I'm not sure I quite follow you.  
There are no "working systems" at this point.  I just have the RH
AS3 Server and the Solaris 8 Client.  The only other client I
tried to use was another Redhat box, which didn't produce errors,
but that is inconclusive at this time because nothing was really
done from that system, and it is not the same type of platform
that I am having trouble with.  The errors occur when certain
jobs are being run on the Sun client that access the filesystem
on the RH server, but not everthing causes it.  The user who's
jobs are causing the errors is working with me to try to figure
out just what is causeing the errors.

My Redhat server is running the rquotad daemon, but I haven't
applied any quotas to the exported filesystems as of yet.
This server is using ext3 for it's filesystems, is there
something better for nfs shared filesystems?  and where exactly
does one look for the kernel options for quota?  I'm using the
default Redhat kernel, and I really hope I don't have to
recompile it everytime Redhat comes out with an update or this
whole "Enterprise" BS is for the birds.

thanks again,

-Jeff