Re: Clarifications: CVS Import Bug - Please Respond

[Mark D. Baushke]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Derek Price <derek@ximbiot.com> writes:

> Mark D. Baushke wrote:
> 
> > some odd race conditions with more than
> > one user trying to create write locks in the same directory at the same
> > time both believing they have obtained their exclusive lock.
> 
> 
> IIRC, this one had to do with clients caching the state of a directory. 
> Something like both clients would query the lock directory's existence
> and if they had cached that it didn't exist, both would return a
> successful status code to a program before they get the failure response
> from the NFS server.

Yes, this is the problem.

> I've fixed that at least once (for running sanity.sh between two hosts)
> by setting the dir-cache-time for my NFS client to 0.  On my Linux
> client the options are acdirmin=0,acdirmax=0 and can be set separately
> for each mount-point.

True, but most folks do not set them in any special way which leads to
the problem.

        -- Mark
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (FreeBSD)

iD8DBQFC+5MbCg7APGsDnFERAmhMAJ9qxtd4MIlxH2Pjmt9QrR3zniQXgACeNWVT
vWH16NvFWHrtoWE8N1RUzms=
=6HOG
-----END PGP SIGNATURE-----