SUMMARY: nfs mounts, file owners and groups problem

Thank you to all who replied, in the end 16 people.  I am very grateful for 
all your suggestions!

The replies varied regarding the strategy to get the job done:
- taring in 2gb chunks, scp the chunks, untar
- a perl script defining getperm and setperm, run over the whole fs
- check for matching userids in etc/passwd (made no difference)
- use ufsdump (no chance to mount ro, or unmount because of time)
- use scp -rp, (fails to preserve owner/group over network)
- use rsync with public key authentication (this worked! Danke Jan)
- share the fs with the options anon=0, root=<client system>
- mount the share with the option vers=3 if I am on Solaris 10 (I am)

I had already been sharing the fs with the root=<client system> option, 
however on a handfull of userid's, this was seen on the client side as 
nobody.  I added the anon=0 option, however this made no difference.  I also 
added the problem userid's to /etc/passwd on the client system with no change 
in behaviour.  The behaviour is very odd for me, as it seems that what I was 
doing with nfs should have worked.  I will have to spend more time on this 
issue as it will come up in the future again under other circumstances.

Jan D. pointed out that rsync should work, however I was obviouly doing it in 
a far too complicated manner - setting up an rsync server, then rsyncing 
through that.  His suggestion was to use key authentication, and
# rsync -e ssh -aHvp S1:/path/to/copydir S2:/copy/me/here/
This got me out of trouble for now.

Thanks again to everyone who replied.

regards
Markus



On Wednesday 26 September 2007, Markus Mayer wrote:
> Hello all,
>
> I need to transfer a large amount of data (about 730Gb) from one machine to
> another and am not able to take anything off line.  As an rsync
> client/server setup doesn't preserve the file ownership and groups at all,
> I am trying to do it using an nfs share.  This method is also running into
> problems.
>
> I have a share on one machine, set up with the following command:
> S1# share -F nfs -o ro=<servername>,root=<servername> /share/location
> On the second machine, I can mount the share using:
> S2# mount -F nfs-o ro <servername>:/share/location /mnt/remoteserver
>
> The problem I run into is that on the first machine, files have thier owner
> and group, however on the nfs mount on the second machine, often the owner
> is changed to nobody.  An example of such a file is:
> S1# ls -ln /share/location/dir1/afile
> -rw-rw-r--   1 206      201         1348 Sep 11  2006 afile
> On the second machine however, I get:
> S2# ls -ln /mnt/remoteserver/dir1/afile
> -rw-rw-r--   1 60001    201         1348 Sep 11  2006 afile
>
> There does seem to be some consistency in this behaviour, all be it
> strange. All files on S1 with owner root have their ownership seen as
> nobody on S2. For other files, if the file owner on S1 has id 206, the
> owner is seen as nobody on S2, however if the owner is 205 or 207, the
> owner is preserved.  On S1, there is an user entry for id 206, so I put
> that id on S2, however with no change in behaviour for ownership
> preservations.
>
> I'm at an absolute loss as to what's happening here and would be grateful
> for any help anyone can give me.
>
> regards
> Markus
> _______________________________________________
> sunmanagers mailing list
> sunmanagers@sunmanagers.org
> http://www.sunmanagers.org/mailman/listinfo/sunmanagers
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers