Paul Keating wrote:

> yes, the backup was traversing the snapshots as well.
> however, it shouldn't have been an issue, since "follow NFS" was 
> turned off.
>  
> I'm pretty sure at this point that it was happening because the mount 
> point was created (/mnt/filer) and a symlink created /filer -> 
> /mnt/filer and then the share was mounted onto /filer, rather than 
> /mnt/filer.
>  
> this link deeked out netbackup, so it couldn't see that the mount was 
> of type "NFS".
>  
> I haven't been able to thouroughly test this theory yet, though.
>  
> Paul
>
>     -----Original Message-----
>     *From:* Jeff Cleverley [mailto:jeff_cleverley AT agilent DOT com]
>     *Sent:* November 2, 2005 10:17 PM
>     *To:* Paul Keating
>     *Cc:* Mark.Donaldson AT cexp DOT com; veritas-bu AT mailman.eng.auburn 
> DOT edu
>     *Subject:* Re: [Veritas-bu] cross mount points? NFS?
>
>     Paul Keating wrote:
>
>>     I didn't do it, and I was as suprised as the next guy when I saw it.
>>     The backup ran real long (250 Gig, rather than the expected 2
>>     Gig) so the admin of the box sent me a df output, so I just
>>     thought it was mounted on /mnt/filer then I browsed the backup
>>     image and said "WTF???"
>>
>     Paul,
>
>     I'm seeing a similar thing associated with a new post I'll do.  I
>     found the df or bdf commands were not listing the "."
>     directories.  The backups were walking all of the .snapshot
>     directories and backing up considerably more data than expected.
>
>     Jeff
>
Cant see how that could fool netbackup unless follow symbolic links is 
checked. There is an option to follow them. I can see if this was set 
and becuase your NFS volume was mounted from a sym link that it may 
ignore the fact follow NFS wasnt checked, but it should have picked 
/filler up as a sym link to start with and just backed the link up and 
not where it goes.

Dave