On Fri, 21 Oct 2005, Jeff Moyer wrote:

> ==> Regarding Re: [autofs] Unable to mount with autofs-4.1.4: aquire_lock: 
> can't lock lock file timed out: /var/lock/autofs; Ian Kent <[EMAIL 
> PROTECTED]> adds:
> 
> raven> On Fri, 21 Oct 2005, Rainer Krienke wrote:
> >> On Mittwoch 19 Oktober 2005 17:55, Ian Kent wrote: > As Jeff has said
> >> this sort of nesting of mounts is, in general, not > supported by autofs
> >> and it worked previously because the locking was less > broken in 4.1.3
> >> than in 4.1.4 (yes its still broken in 4.1.4).
> >> >
> >> > Can this case be done safely?
> >> >
> >> > I'm not entirely sure of the answer to this but I've never liked using
> >> > locking in autofs. I introduced it only to prevent mtab corruption
> >> during > mount (correct me if you have a different recollection
> >> Jeff). It happens > during rapid mount activity such as when there as a
> >> largish number of > entries in the master map (of the order of 50 or
> >> more).
> >> >
> >> > The mtab locking should be done in mount which narrows the scope of
> >> the > lock enough to allow this case to function as requested.
> >> >
> >> 
> >> For me it would be important to learn if this feature will be possible
> >> in the future or if it is decided that for whatever reason it will not
> >> be. The former case would be the best thing for me because I would not
> >> have to find a new solution for mounting home directories and I guess
> >> other people with a large user base would also welcome it. In the latter
> >> case I have to think about a alternative way to go for the future. At
> >> the moment my solution is to use an older version of autofs for suse10.0
> >> installations since this older autofs version simply works for me but
> >> this is of course not an ideal solution.
> >> 
> >> Any comments?
> 
> raven> Yes.
> 
> raven> I was wondering what can be done about this also.
> 
> raven> All that I can offer is to make a patch to remove the locking from
> raven> autofs and another for mount (util-linux) to hopefully fix the mtab
> raven> locking. If you are prepared to carry these non-standard patches we
> raven> can see how it goes. Somewhat more testing would be required before
> raven> removing the locking from the distribution.
> 
> What, exactly, do you think is broken about the mtab locking?  I remember
> spending a few hours with that code, and it looked sane to me.

I'm not sure yet but I had a bug report on it.

It appears that there is a situation in which autofs thinks the lock 
is still held when it is not. A race (no doubt) or maybe a missing test in 
the wait. So my feeling of wanting to eliminate the locking returns again. 
Sorry.

I received a patch that rewrote the lock code as well but I haven't looked 
closely at it yet.

OTOH Adrain Bunk is the util-linux maintainer (for a while now it seems) 
and I recently sent a patch to him for mount that might fix the locking.

I'll forward the mail later? It just explains what I think the problem 
might be and how the cause recently jumped out at me when looking at 
the mount code.

I really think it would be better if the locking could be done in mount. 
It should be much simpler to get it right there instead of in autofs.

Ian



_______________________________________________
autofs mailing list
[email protected]
http://linux.kernel.org/mailman/listinfo/autofs

Reply via email to