[ovs-dev] [RFC 1/8] system-traffic: Load balancing.

Jarno Rajahalme jrajahalme at nicira.com
Mon Mar 14 20:57:57 UTC 2016


> On Mar 14, 2016, at 10:42 AM, Ben Pfaff <blp at ovn.org> wrote:
> 
> On Sun, Feb 28, 2016 at 10:33:17PM -0800, Gurucharan Shetty wrote:
>> From: Jarno Rajahalme <jrajahalme at nicira.com>
>> 
>> Signed-off-by: Jarno Rajahalme <jarno at ovn.org>
> 
> The core of the change here seems to be:
> 
>> diff --git a/lib/ofp-actions.c b/lib/ofp-actions.c
>> index becf02d..8f6e02e 100644
>> --- a/lib/ofp-actions.c
>> +++ b/lib/ofp-actions.c
>> @@ -5913,7 +5913,8 @@ ofpacts_execute_action_set(struct ofpbuf *action_list,
>>      * not be sent anywhere. */
>>     if (!ofpacts_copy_last(action_list, action_set, OFPACT_GROUP) &&
>>         !ofpacts_copy_last(action_list, action_set, OFPACT_OUTPUT) &&
>> -        !ofpacts_copy_last(action_list, action_set, OFPACT_RESUBMIT)) {
>> +        !ofpacts_copy_last(action_list, action_set, OFPACT_RESUBMIT) &&
>> +        !ofpacts_copy_last(action_list, action_set, OFPACT_CT)) {
>>         ofpbuf_clear(action_list);
>>     }
>> }
> 
> The change to ofproto-dpif-xlate seems undesirable, because if I read it
> correctly it means that every packet translation that involves a group
> bucket will put that into the log, without even rate-limiting.  I think
> that this change was probably meant for debugging, without meaning to be
> applied permanently.
> 

Right, I sent my unpolished patch to Guru, and did not remember about the debugging I added to ofproto-dpif-xlate. So, to be clear, the ofproto-dpif-xlate changes in this patch should not be applied, otherwise this should be good to go.

  Jarno

> I didn't read the change to the tests, but I appreciate them.




More information about the dev mailing list