Sean O'Neill wrote:
> Snoop ... I was messing with NFS mounts on a Mac client and the  
> mounts were taking quite a bit of time so I pulled snoop out.
> 
> So even after I tuned /etc/default/nfs to max out at version 3 for  
> the client and the server and then restarted the NFS server SMF  
> service, the nfsmapid daemon was still probing DNS for the  
> _nfsv4idmapdomain TXT record.  I found this odd.  (My /etc/default/ 
> nfs file has the NFSMAPID_DOMAIN variable commented out - not sure if  
> this stops nfsmapid from probing DNS or not but this isn't the point  
> of my question).  I disabled the nfsmapid SMF service and tried  
> again.  The DNS probes no longer occurred and my Mac continued to  
> mount NFS mounts points fine.

If you have the nfs/mapid service enabled, it will try to make
sure it can provide the service.  I understand that you have set
the maximum NFS version to 3, but that's not something that we
have mapid check for, because you could still do a manual mount
and specify "-o vers=4" and need mapid to be there.  You were
right to disable mapid to stop the DNS query; you could also have
explicitly set a domain in /etc/default/nfs to stop the DNS queries.
Of course, I would not have expected that DNS query to have been a
problem, either (more below).

> Actually technically speaking a Mac client only supports up to NFS  
> version 3 currently (I'm running OS X 10.4.9) so even if I hadn't  
> tuned /etc/default/nfs to max out at version 3 it would still seem  
> logical that nfsmapid shouldn't be doing anything when I attempt  
> NFSv3 mounts.

I wouldn't expect mapid to be involved if you're doing manual
NFSv3 mounts.  I would expect that you would not see mounts occur
out of /etc/vfstab until the mapid service was up and running,
since those mounts are associated with the nfs/client service and
there is an SMF dependency on mapid unless mapid is disabled.
Did you observe what felt like a slowdown in manual mounts before
you disabled mapid, or would the /etc/vfstab thing explain it?

Rob T

Reply via email to