[ovs-dev] [PATCH v7 0/3] Support dynamic rebalancing of offloaded flows

Simon Horman simon.horman at netronome.com
Fri Oct 12 09:44:04 UTC 2018


On Thu, Oct 11, 2018 at 04:01:40PM -0700, Ben Pfaff wrote:
> Thanks for the revision.
> 
> This seems basically OK at a glance but I'd like a second set of eyes.
> Simon, are you willing to review this?  It seems roughly in your area
> too.

Thanks Ben, Thanks Sriharsha,

I am very pleased to see work in this area.

I have a few lingering concerns, which I noted in separate emails regarding
* Extra CPU cost of processing OOR flows and;
* Correctly detecting the offload device of a tunnel
but I think they can be treated as possible further work rather
than holding up this patchset.

I am also not entirely comfortable with the use of ovs_assert() (in general)
but this may be just a matter of personal taste and again I don't think
it needs to hold up this patchset.

In all, as the feature will be disabled by default and should have negligible
impact when disabled I think it would be good to merge in its current form
to allow further testing (and ideally evolution) of this feature.

Acked-by: Simon Horman <simon.horman at netronome.com>


Ben, I would be happy to apply this series but I'd rather
do so once master travis-ci clean for master. Something I am looking at
separately. I'm also happy for you to apply this series.


More information about the dev mailing list