[OpenAFS] sqlite on AFS will not work, even with whole-file locking

Matt W. Benjamin matt@linuxbox.com
Wed, 21 Apr 2010 11:34:52 -0400 (EDT)

If the CM omits to flush in this case, that's a plausible result.  Viced will BCB when the lock is released, but if the changed data hasn't been written back, the cooperating clients lose.


----- "Simon Wilkinson" <sxw@inf.ed.ac.uk> wrote:
> I strongly suspect that the copy on the client is up to date, at least
> according to the server. However, it may well be out of data according
> to the machine which made the modification. That machine only writes 
> its changes back to the server on receipt of a fflush or close (or  
> when it's cache fills, and it needs to shed some dirty chunks).
> So, if there is a problem here, it's going to be related to the fact 
> that we aren't flushing when a lock is released, rather than our  
> behaviour on obtaining a lock. Perhaps you could verify that case?
> Cheers,
> Simon.
> _______________________________________________
> OpenAFS-info mailing list
> OpenAFS-info@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-info


Matt Benjamin

The Linux Box
206 South Fifth Ave. Suite 150
Ann Arbor, MI  48104


tel. 734-761-4689
fax. 734-769-8938
cel. 734-216-5309