18:01:30 #startmeeting networking_policy 18:01:30 Meeting started Thu May 26 18:01:30 2016 UTC and is due to finish in 60 minutes. The chair is SumitNaiksatam. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:01:32 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 18:01:34 The meeting name has been set to 'networking_policy' 18:01:36 jagadish should be joinging in a mi 18:01:38 n 18:02:12 hemanthravi: okay 18:02:50 #info agenda https://wiki.openstack.org/wiki/Meetings/GroupBasedPolicy#May_26th.2C_2016 18:03:02 i had the date wrong on the wiki page, just corrected it 18:03:11 #topic Bugs 18:03:31 i did not notice anything critical/high being reported 18:03:58 everything works! 18:04:11 there was a bug reported by songole about the same IP being allocated to a member when the member is associated with multiple groups 18:04:18 ivar-lazzaro: lol! 18:04:22 jagadish: hi 18:04:42 we kind of analyzed what is going on, but will let songole enter the bug first 18:04:53 SumitNaiksatam: can a single member be associated to multiple groups? 18:04:57 * igordcard will be back in 3 minutes 18:05:00 ivar-lazzaro: yes 18:05:01 you mean PT? 18:05:13 ivar-lazzaro: not PT, member 18:05:23 ivar-lazzaro: one PT per groups 18:05:27 *group 18:05:39 but the member has multiple PTs 18:05:49 and each PT is a different group 18:06:12 SumitNaiksatam: oh ok, that's the UI member 18:06:19 ivar-lazzaro: right 18:06:39 the problem in this case is that the PTs belong to PTGs which belong to different L3Ps in different tenants, but they have the same ip_pool 18:06:49 which they are getting by default 18:06:57 and hence it leads to this issue 18:06:59 anyway 18:07:18 if there are no other high priority bugs to be discussed... 18:07:49 #topic Packaging 18:07:56 rkukura: last week you mentioned “GBP integration into DeLorean” 18:08:18 right, but I’ve got no response to my queries on the status of this 18:08:20 any update on that or anything else? 18:08:33 so no update 18:08:53 rkukura: okay, do we need to consider escalating at this point, or you think we have more time on this? 18:08:55 tbachman: hi 18:09:01 SumitNaiksatam: hi! :) 18:09:29 SumitNaiksatam: I’m not sure what’s driving the schedule for the RDO packaging, but the longer we wait, the less likely it is to ever happen 18:10:17 rkukura: okay, let me know if you need any help from me to push this forward, and/or we take this offline to discuss what our options are 18:10:43 #topic Tests 18:10:49 just a quick update from me 18:11:05 i have been trying to get the rally integration job to work for the past couple of days 18:11:34 our rally branch was woefully out of sync, and its taking some time to catch up with all hte changes that have happened 18:11:46 igordcard: there? 18:12:12 okay may be we will discuss NFP first 18:12:19 #topic NFP impl patches 18:12:26 #link https://review.openstack.org/#/q/topic:bp/gbp-network-services-framework 18:12:37 yep 18:12:55 hemanthravi: jagadish and team thanks for updating the patches with the relevant co-authors 18:13:11 i think gate test should also be running now 18:13:14 igordcard: ooops, lets get done with NFP, and we can circle back to your patch 18:13:27 there are few more impl patches, will be submitted in a day or 2 18:13:44 SumitNaiksatam: sure, wasn't looking when you asked 18:14:02 hemanthravi: the integration gate job is still broken starting from the first patch #link https://review.openstack.org/#/c/282292 18:14:16 igordcard: np, thanks for your patience 18:14:35 Most of the review comments on base configurator and reference configurator will be submitted in a day 18:14:52 SumitNaiksatam: how do i get the logs to this failure? 18:15:14 dsvm-functional ? 18:15:32 hemanthravi: yeah, just click on the job 18:16:12 will take a look and resolve it 18:16:18 hemanthravi: i had already provided comments on what might be broken and what can probably fix it 18:16:48 see my comment on May 12th 18:17:25 hemanthravi: jagadish: i believe there were plans to update this wiki page with more information about the patches: #link https://wiki.openstack.org/wiki/GroupBasedPolicy/GerritQueries/NFP ? 18:17:55 SumitNaiksatam: i'm behind on updating the wiki page, will do this 18:18:02 hemanthravi: okay thanks 18:18:20 hemanthravi: when you said the gate job is working, what were you referring to? 18:19:36 hemanthravi: i am referring to your earlier comment “i think gate test should also be running now” (since we did not get a chance to discuss that) 18:19:53 don't have the details, but i was told one of the nfp tests ran as part of the gate. I'll confirm this tonight 18:20:08 SumitNaiksatam: The nfp node driver has been pointed in the configuration for the gate tests to run nfp functionality tests. 18:20:21 to run haproxy in namesapce 18:21:05 jagadish: but the integration job is broken for all the patches and the issue is in the first patch in the chain itself 18:21:23 jagadish: also, which patch are you referring to? 18:22:38 SumitNaiksatam: will check on this tonight 18:23:12 SumitNaiksatam: will check on this tonight 18:23:13 hemanthravi: okay, my understanding is that we are not replacing any of the current integration tests, we are adding new ones 18:23:46 SumitNaiksatam: Yes. The current ones are not replaced. New ones are added. 18:24:00 jagadish: okay good, just wanted to confirm 18:24:14 hemanthravi: jagadish: at this point are there any blockers for you? 18:25:21 perhaps not :-) 18:25:41 not a blocker, but the gate tests had to be done without using the plugin but gbp-patch 18:26:00 hemanthravi: sorry, didnt get that? 18:26:12 gbp-patch? 18:26:32 the gate tests don't using the devstack plugin but the earlier gbp-patch.sh 18:26:37 is that correct? 18:26:40 SumitNaiksatam: Gate tests are done using GBP patch 18:26:58 They don't have devstack plugin support 18:27:04 i think this can be addressed later, once we have the nfp gate tests passing 18:27:05 hemanthravi: jagadish: yes they dont use the devstack plugin, that is expected 18:27:12 hemanthravi: thats not a problem at all 18:27:28 hemanthravi: yeah we can address that later 18:28:12 hemanthravi: jagadish: all the devstack artifacts for the gate tests are contained here: #link https://github.com/openstack/group-based-policy/tree/master/gbpservice/tests/contrib 18:28:59 ok 18:29:03 hemanthravi: jagadish: i believe you are referring to the patching be done here: #link https://github.com/openstack/group-based-policy/blob/master/gbpservice/tests/contrib/functions-gbp ? 18:30:00 SumitNaiksatam: Yes 18:30:09 anyone else in the team have questions for hemanthravi and jagadish on NFP patches? 18:30:13 jagadish: okay good 18:30:46 hemanthravi: jagadish : thanks for the update 18:30:59 #topic GBP QoS Support 18:31:11 #link https://review.openstack.org/#/c/301701/ 18:31:58 just to refresh people’s memories we had merged the spec: #link https://github.com/openstack/group-based-policy-specs/blob/master/specs/mitaka/initial-qos-support.rst 18:32:14 and igordcard had proceeded to do the implementation in the feature/qos branch 18:32:27 igordcard: thanks for your work 18:32:42 anyone got a chance to review the implementation? 18:33:02 I started 18:33:09 rkukura: great, thanks 18:33:14 i put some preliminary comments 18:33:21 nothing major though 18:33:46 as long as we are fine the approach (which is stated in the merged spec), i think the implementation is pretty straightforward 18:33:49 thanks, I'll address them 18:34:15 so IMHO, its really the approach that we are validating there 18:34:31 hemanthravi: ivar-lazzaro: it will be helpful if you guys can take a quick look as well 18:34:41 SumitNaiksatam: +1 18:34:51 ivar-lazzaro: okay cool 18:35:03 SumitNaiksatam: ok 18:35:09 hemanthravi: thanks 18:35:36 rkukura: ivar-lazzaro hemanthravi thanks 18:35:58 igordcard: at this point, my understanding is that is a functional implementation, so you are kind of “done” with your implementation for this first iteration? 18:36:20 *that this is a functional implementation 18:37:15 SumitNaiksatam: Are we not planning to merge this implementation, assuming we are all happy with it? 18:37:28 SumitNaiksatam: do you mean if PoC 01 is the end of it? 18:37:39 rkukura: i agree 18:38:03 SumitNaiksatam: Are you agreeing that we should merge it? 18:38:08 igordcard: i meant to ask, if PoC 01 is complete, and/or there are any blockers? 18:38:26 rkukura: if we are fine with the approack we should merge it 18:38:34 I agree 18:38:50 rkukura: but i think you are asking if we should merge it to master from the feature branch? 18:38:52 SumitNaiksatam: oh ok, so the patch only adds support for maxrate as a first iteration 18:39:20 SumitNaiksatam: maybe, if it is to be merged to master instead of feature/branch, more work should be done to at least support burstrate as well 18:39:29 igordcard: i agree 18:39:30 SumitNaiksatam: but still the same style, as in inner-PTG QoS only 18:39:48 igordcard: were you planning to add burst rate as PoC 2 patch? 18:40:48 SumitNaiksatam: planning to add still to PoC 1 after an initial wave of review since the architecture wouldn't change but I still had to invest a bit of time to make sure both *rates would coexist peacefully 18:40:59 igordcard: ah okay 18:41:35 SumitNaiksatam: I will add a note about that 18:41:53 so i think this is a question for the team - but in my opinion, if we are all fine with this approach, then we can merge the current patch to master, and then have igordcard post the burst rate changes directly to master 18:42:18 this is of course after igordcard himself is comfortable that “both *rates would coexist peacefully” 18:42:47 i think we are using the feature branch to validate the approach, after that i dont think the overhead of the branch is required 18:43:10 SumitNaiksatam: yeah, but for master I'd prefer to add max and burst together in the same patch 18:43:51 igordcard: ah okay, so in that case we can merge this patch into the feature/qos branch (after reviews are complete) and wait for you to post the updates for burst rate 18:44:02 SumitNaiksatam: okay 18:44:16 igordcard: thanks for the update and the work 18:44:28 ;) 18:44:32 any other questions for igordcard ? 18:45:04 okay 18:45:10 #topic Open Discussion 18:45:21 anything we missed for today’s meeting? 18:45:59 if not, we can get 15 mins back :-) 18:46:24 alrighty, thanks all for joining (igordcard for staying up late ;-)) 18:46:31 bye! 18:46:40 #endmeeting