[ovs-dev] [PATCH 3/3] cfm: Expose detailed fault status in the database.

Ben Pfaff blp at nicira.com
Wed Feb 8 20:48:26 UTC 2012


On Wed, Feb 08, 2012 at 12:45:07PM -0800, Ethan Jackson wrote:
> > So, my interpretation of the database has been that cfm_fault says
> > whether there is a fault and cfm_fault_status gives the reason for the
> > fault.  How should a controller interpret it if there is a fault
> > (cfm_fault=true) but no reason, or almost equivalently if there is a
> > fault but the reason is just "override"?  Especially if the goal of
> > providing the ability to override is to allow for testing (that is the
> > reason, right?) then I'd expect that one would want to simulate a
> > specific kind of fault.
> >
> > Perhaps I just misunderstand.
> 
> Right sorry perhaps I wasn't clear, my plan is to add the override
> fault status to the database.
> 
> > Right, recv_fault would be a CFM_FAULT_* bitmap.  On an unexpected
> > MAID, recv_fault |= CFM_FAULT_MAID.  On an overflow, recv_fault |=
> > CFM_FAULT_OVERFLOW.
> 
> Oh, I had misunderstood you.  This makes sense to me I'll go ahead and do it.
> 
> > Not sure you saw my other comments since I didn't see any replies.
> 
> Your other comments seemed reasonable and will be folded into the
> incremental I intend to send shortly.

Thanks, all that makes sense to me.



More information about the dev mailing list