Messages in this thread | | | Date | Wed, 02 Feb 2000 17:05:55 -0800 | From | "H. Peter Anvin" <> | Subject | Re: autofs 4.0.0 (pre1) |
| |
Frank van Maarseveen wrote: > > On Thu, Jan 27, 2000 at 12:20:49AM -0800, Jeremy Fitzhardinge wrote: > > >> What version are you comparing it to? > > > 2.2.10-ac11 > > > > > > In case you're wondering why: almost no 2.2.x kernel is usable (at my > > > work) because they all suffer from an NFS client bug which is fixed by > > > Trond's patch I'm using in 2.2.13. Without this patch, redirecting setuid > > > program output to an NFS mounted FS fails: data is discarded because I/O > > > is done with the credentials of the calling process (e.g. setuid root) > > > instead of the opener of the file. > > > > Well, my patches do nothing to the behaviour of NFS. Are you saying that > > autofs 4 is more stable than a previous version of autofs with 2.2.x? > Yes. > > 2.2.10-ac11 autofs didn't panic/oops the system but gave some troubles > when a (non-Linux) server disk was resized or was temporarily > offline. I don't recall the details but at times it was difficult > to get rid of old mounts and/or to restart autofs. It had all to do > with error handling (or maybe also a race) during mount/umount actions. >
That version had a known bug (not in autofs, in the VFS.)
> > First impression is that the new autofs together with a new mount > behaves better. I still have to test it more systematically but I'll > use this new autofs anyway. We'll see what happens. > > One concern is that all automounts here are hard NFS mounts: except for > a stall server/network problems should never disrupt operations on NFS > clients. On the other hand, when a server disk has been resized every > NFS operation on old mounts will yield ESTALE. Mount expiration and even > restarting autofs should still work in this case and not block on the > error. > > Another concern is the mount program which must try forever and not > timeout when the server is not responding at mount time. I think mount > expiration should block as well in this case except when forced by an > autofs stop/restart. >
These are NFS operations, and don't have anything to do with autofs.
-hpa
-- <hpa@transmeta.com> at work, <hpa@zytor.com> in private! "Unix gives you enough rope to shoot yourself in the foot."
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |