> On Oct 31, 7:33pm, Chris Cowan wrote:
>
> > Most likely, the cell administrators have a cron job configured under
> > bos to clone the *.backup volumes
> > on a daily basis. Look at the "vos backup" command for more info.
> > Do a "bos status .... -noa" against some of the file servers. You
> > should be able to see this
> > cron'ed procedure.
> >
> > This would account for the 1 access per day pattern, that you are
> > observing.
>
> I did an experiment creating a backup volume, and it did not increase the
> access count. From a cursory look at the src, the backup volume cration
> process does not go through the same vnode access routine that would bump
> the access count. I don't think it should.
Correct. The access count is bumped only in the fileserver, when a
client actually accesses the volume. The cloning procedure used by
backups and releases of replicated (in vol/clone.c) does not change
the access count. The dump/restore process doesn't change the access
count either, which means it won't change even if you move the volume
from one place to another.