Hello!
On Wed, Feb 28, 2007 at 09:45:29AM -0700, Peter Braam wrote:
> However, I would like to hear why there are no race conditions in this
> protocol. How can we be certain that a canceled lock isn't used by the
There are, they do not matter.
We already release the lock halfway through its processing (even for vfs
clients).
Server will do all the validation itself later anyway.
There are no write locks for metadata on clients (yet), and so they are
merely a hint that "we just checked this file and during the check it
was in place (and his attributes were ...)"
Or what protocol do you mean
> client? Also, is there a point to having these locks at all?
Yes. Cache.
Bye,
Oleg
_______________________________________________
Lustre-devel mailing list
[email protected]
https://mail.clusterfs.com/mailman/listinfo/lustre-devel
