[ovs-dev] [PATCH v7 3/3] Translation of fast failover groups

[Simon Horman]

Hi,

I guess it depends a little on what you want to look at.

If you want to actually be able to add and groups and use them to forward
packets then I suggest this patch series, which you can find in git.

https://github.com/horms/openvswitch.git devel/groups-v7

I plan to post v8 some time this week and I will both
post the patches to this ML and push a branch to my git tree
when that happens.

On Mon, Oct 28, 2013 at 04:29:47AM +0000, P Balaji-B37839 wrote:
> Hi Simon,
> 
> Can I take HEAD and start exploring on groups. I want to try this for VXLAN and 'flow based tunneling'.
> 
> Regards,
> Balaji.P
> 
> > -----Original Message-----
> > From: Simon Horman [mailto:horms at verge.net.au]
> > Sent: Monday, October 28, 2013 8:19 AM
> > To: Ben Pfaff
> > Cc: dev at openvswitch.org; Neil Zhu; Jarno Rajahalme; Casey Barker; P
> > Balaji-B37839; Addepalli Srini-B22160
> > Subject: Re: [PATCH v7 3/3] Translation of fast failover groups
> > 
> > On Mon, Oct 21, 2013 at 03:04:25PM -0700, Ben Pfaff wrote:
> > > On Tue, Oct 15, 2013 at 05:17:50PM +0900, Simon Horman wrote:
> > > > Fast failover groups use the actions in the first bucket that is
> > > > alive.
> > > >
> > > > Signed-off-by: Simon Horman <horms at verge.net.au>
> > >
> > > Please make bucket_is_alive() log a rate-limited warning if it reaches
> > > the liveness depth limit.
> > 
> > Sure, will do.
> > 
> > > xlate_ff_group() abandons translation entirely, with ctx->exit, if
> > > there is no live bucket.  I think that this reads too much into what
> > > the standard says, which is just "If no buckets are live, packets are
> > > dropped."  I think the intention is just that, if there's no live
> > > bucket, don't send the packet to any bucket, but (as usual) the
> > > wording is ambiguous.
> > 
> > Yes, I agree. Actually I had come to the same conclusion while reading
> > over the patch at around the same time that you wrote this email.
> > I'll remove setting ctx->exit.
> 
>