[ovs-discuss] Question about order of OVSDB transaction commit response v.s. update notification

Han Zhou zhouhan at gmail.com
Thu Mar 9 01:42:54 UTC 2017


On Wed, Mar 8, 2017 at 3:36 PM, Ben Pfaff <blp at ovn.org> wrote:
>
> On Wed, Mar 08, 2017 at 03:25:47PM -0800, Han Zhou wrote:
> > I checked the current python OVSDB IDL implementation, and found out
that
> > the way local cache is handled during transaction commit is like this,
as
> > commented in idl.py:
> >
> >         Committing a transaction rolls back all of the changes that it
made
> > to
> >         the IDL's copy of the database.  If the transaction commits
> >         successfully, then the database server will send an update and,
> > thus,
> >         the IDL will be updated with the committed changes."""
> >
> > If a transaction commit is successfully returned, the client will always
> > see the new data from local cache. It seems this behavior is guaranteed
by
> > the order of transaction commit response and update notification for the
> > same commit: the update notification always come BEFORE the response of
the
> > same commit. I didn't check ovsdb server code to verify. Please correct
me
> > if I am wrong.
>
> Yes, that's how it works.
>
> > If I am right, then I didn't see this order being specified in RFC 7047.
> > Would this order be always guaranteed in ovsdb-server implementation?
Is it
> > better to be specified in RFC?
>
> I don't think that RFC 7047 states or implies the ordering.  This is a
> bug in the RFC, and I wish that it did say that.  I don't think that
> OVSDB is as useful without an ordering guarantee here.

Ben, thanks for your quick answer!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openvswitch.org/pipermail/ovs-discuss/attachments/20170308/d01e2f71/attachment.html>


More information about the discuss mailing list