[GEDI] gfapi: possible approaches to get lk_owner support

Niels de Vos ndevos at redhat.com
Wed Oct 4 12:10:52 UTC 2017

On Wed, Oct 04, 2017 at 05:03:35PM +0530, Soumya Koduri wrote:
> Thank you all for your comments. Bases on your inputs, I did some POC on
> approach(3) i.e, to store lkowner in glfd itself. API syntax and sample test
> case are updated in [1]. Will refine the patch a bit more before calling for
> formal review. Please let me know if there are any objections.
> > Jiffin's patch does a glfs_dup, if that creates a separate glfd that we
> > can then set lock owner differently on, then it would work.
> > 
> This seems to be a bit tricky. Each gfapi's glfd correspond to a unique fd
> on gluster server side. However if used glfs_dup, same server fd will be
> used for both those glfd objects but with extra reference taken.
> As far as the locks are concerned, looks like when we close a fd, server
> expects us to pass on lkowner as well and it cleans up locks with only that
> lkowner. That way we are right now safe to use glfs_dup and set another
> lkowner on the duplicate glfd returned.
> But if in case in future there comes any requirement to close all the locks
> tied to an fd, then the locks taken on both the glfd's shall be lost.

Sounds good to me. I've left some initial review comments on your patch


> Thanks,
> Soumya
> [1] https://review.gluster.org/18429

More information about the integration mailing list