Of course this should have been a reply to Axel. Sorry Jeff.
On Fri, 26 Aug 2005, Ian Kent wrote:
> On Thu, 25 Aug 2005, Jeff Moyer wrote:
>
> > ==> Regarding [autofs] Cascading/recursive autofs mounts?; Axel Thimm
> > <[EMAIL PROTECTED]> adds:
> >
> > Axel.Thimm> Hi, we have a setup, where we would like to have autofs mount
> > Axel.Thimm> over NFS a certain folder and then bind-mount from subfolders
> > Axel.Thimm> to other places.
> >
> > Axel.Thimm> Can this work with autofs? E.g. can autofs cascade mount
> > Axel.Thimm> requests? E.g. if the first autofs request would require a
> > Axel.Thimm> second one to be performed first, would the second one properly
> > Axel.Thimm> mount the NFS share and the outer one continue with bind
> > Axel.Thimm> mounting?. It looks like the first one would lock
> > Axel.Thimm> /var/lock/autofs and the inner autofs request would fail.
> >
> > Correct. The way things are structured in autofs4.1.4, recursive mounts
> > such as this are not supported.
>
> This should be able to be done with Sun style offset mounts (which we are
> calling multi-mounts). We can't do that atm but that's comming.
Saying that this can't be done is not entirely correct.
What can be done now is to specify your mounts as a multi-mount map entry.
The limitation is that, currently, a multi-mount must be treated as a
single unit. Meaning that all the offsets must be mounted on access and
similarly on umount.
For example a map entry like:
example / host:/you/nfs/mount \
/other/one :/dev/sda1 \
/this/two host2:/another/mount
would mount all three of these and not umount them until all three have
been idle for at least timout.
The auto.net example program map script does this for a list of exports
when passed a key corresponding to a server server.
Ian
_______________________________________________
autofs mailing list
[email protected]
http://linux.kernel.org/mailman/listinfo/autofs
