18:02:55 #startmeeting networking_policy 18:02:56 Meeting started Thu Nov 13 18:02:55 2014 UTC and is due to finish in 60 minutes. The chair is SumitNaiksatam. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:02:57 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 18:02:59 The meeting name has been set to 'networking_policy' 18:03:28 #info meeting agenda https://wiki.openstack.org/wiki/Meetings/Neutron_Group_Policy#Nov_13th.2C_2014 18:03:41 #topic Paris summit follow up 18:04:07 I think the summit was pretty exciting from a GBP perspective 18:04:21 thanks to everyone for contributing and participating 18:04:53 #info GBP presentation at conference: #link https://www.openstack.org/summit/openstack-paris-summit-2014/session-videos/presentation/group-based-policy-extension-for-networking 18:05:10 #info GBP design session etherpad: #link https://etherpad.openstack.org/p/kilo-gbp-design-summit-topics 18:05:20 Youcef: welcome! ;-) 18:05:41 rkukura: thanks for joining 18:05:44 hi - sorry I’m late 18:05:59 calendar didn’t stick with UTC 18:06:07 any thoughts, feedback from the summit that we need to discuss up front? 18:06:26 rkukura: yeah, i think most people have the issue, i should have sent a note out earlier 18:06:39 so any feedback from the summit, that we want to discuss? 18:07:09 (we will get into specific technical items next, just checking for any general comments) 18:08:00 okay, seems like people are still catching up ;-) 18:08:14 #topic GBP Juno milestones 18:08:43 during the breakout session on thursday it was proposed that we have the following immediate milestones 18:08:47 well, even Kanzhe decides to join :-) 18:08:54 hi all. 18:09:14 oh yeah, we volunteered Kanzhe for a work item 18:09:23 i bet he has an update! ;-P 18:09:31 Kanzhe: good to see you back 18:09:37 so on the milestones 18:09:42 thanks. 18:10:05 prior to the summit, we wrapped on juno-gbp-2 milestone 18:10:33 the proposal is to do juno-gbp-3 on nov 28th, and release on dev 19th 18:11:31 any issues with the above time lines (we will discuss next what we are targeting for those milestones) 18:12:07 SumitNaiksatam: depends on the items :) 18:12:18 ivar-lazzaro: yes, okay so we will defer the questions 18:12:25 #topic Pending work items 18:13:06 so based on the discussions and feedback during the summit, we will have to push some more essential features for juno 18:13:39 let me quickly try and run through those 18:13:47 feel free to interject at any time 18:14:07 first thing we want to target is the resources’ renaming 18:14:42 so i posted a patch for that: #link https://review.openstack.org/#/c/133949 18:14:57 ivar-lazzaro: will need your help on a few UTs that are failing 18:15:00 SumitNaiksatam: thanks for taking care of this! 18:15:10 ivar-lazzaro: np 18:15:20 we also need to refactor the client code for this 18:15:39 i will post the patch later today (unless someone else wants to pick this up) 18:15:40 SumitNaiksatam: at a first glance, It seems a problem related to the fact that APIC driver uses an external library for getting the work done 18:15:50 ivar-lazzaro: ok 18:15:53 SumitNaiksatam: and renaming also those calls is breaking things :) 18:16:27 ivar-lazzaro: ah ok, i tried to do it in a way that the calls to the external libraries were not broken, but i might messed it up! :-( 18:17:03 SumitNaiksatam: np, I'll try to fix it later today 18:17:27 ivar-lazzaro has a spec for shared resources #link https://review.openstack.org/#/c/133603/ 18:17:47 this helps to target the external network use case 18:18:00 kindly review that spec 18:18:32 SumitNaiksatam: is that targeting Juno also? 18:18:40 yes please, especially let's try to understand together which resources we want to make "sharable" on the reference implementation 18:18:42 s3wong: yes all this for Juno 18:18:52 based on what neutron allows us to do ofc :) 18:19:12 and by Juno we mean preferably Nov 28th, because that is feature 18:19:20 *feature freeze 18:19:55 rkukura is also working on the extensions loading mechanism for vendor drivers 18:20:00 the spec is approved 18:20:14 rkukura: is a patch posted for this? 18:20:21 not yet - next week 18:20:40 rkukura: ok cool 18:21:09 Do we need extensions for all resources initially? Or just the ones with mappings? 18:21:11 anyone planning to write the vendor drivers please take note of the extension loading mechanism (similar to that proposed in ML2) 18:21:41 rkukura: ah, so you are planning to move the mapping to the driver as well? 18:21:58 SumitNaiksatam: That would be a 2nd phase, but maybe also next week 18:22:04 rkukura: ok 18:22:21 i think ivar-lazzaro and banix have requirements for the extension loading from drivers 18:22:23 Plan is for initial patch to support extensions for L3P, L2P, PTG, and PT only 18:22:30 so they might be able to speak to this 18:22:33 Cathy_: hi 18:22:46 SumitNaiksatam: hi, sorry to join late 18:23:01 Cathy_: no worries, the time shift was confusing for everyone 18:23:31 ivar-lazzaro banix can you sync with rkukura offline on which resources you need support for? 18:23:36 rkukura: what does that imply? Will contracts still be bound to SGs for all the drivers? 18:24:12 SumitNaiksatam: ok will do 18:24:16 banix: thanks 18:24:23 ivar-lazzaro: We don’t expose the SG mapping, so shouldn’t be an issue for that 18:24:41 A followup patch would add support for extending whatever other resources we need 18:24:43 rkukura: yeah just realized that, nevermind :) 18:24:56 rkukura: ah, that would be ideal! 18:25:05 rkukura: that's great! 18:25:05 rkukura: thanks for that update 18:25:15 next item ... 18:25:38 s3wong is working on getting the updates to work for the policy rules and rule sets 18:26:00 s3wong this would involve DB schema changes, right? 18:26:04 SumitNaiksatam: indeed 18:26:24 just asking if this warrants a spec 18:26:36 but dont want to introduce unnecessary process overhead 18:26:50 we can have a separate design review session if required 18:27:08 SumitNaiksatam: yes, I need to have several more tables to get which SG a policy-rule/classifier/action maps to 18:27:16 s3wong: makes sense 18:27:19 SumitNaiksatam: OK 18:27:32 next item is incorporating the redirect action in the hierarchical contract composition 18:27:57 magesh is not here, but he might be interested in doing that, i will follow up offline with him 18:28:03 (currently I am operating under the assumption that I don't need a separate spec for that... because TBH, it should have been done along with the last SG patch :-) ) 18:28:04 s3wong: have you looked into a joined query? 18:28:22 rkukura: +1 18:28:24 s3wong: fair enough :-) 18:28:25 s3wong: I agree a spec should not be needed 18:28:46 rkukura: no, but that does sound like a better thing to do... 18:29:23 SumitNaiksatam, i 18:29:35 hemanthravi: hi, you made it :-) 18:29:39 SumitNaiksatam, will follow up magesh on the redirect... 18:29:55 SumitNaiksatam, forgot about the time change...call at 11 18:30:01 hemanthravi: great thanks! 18:30:14 s3wong: thanks for the update, anything more on that work item? 18:30:43 SumitNaiksatam: no, but will definitely get it done by Nov. 28th :-) 18:30:55 s3wong: sweet, thanks! :-) 18:31:06 next item - vendor drivers 18:31:16 i believe ronak is not here 18:31:28 i believe he is working on this 18:31:53 banix: any chance that you will posting one for this release? 18:32:40 SumitNaiksatam: There is a colleague who is working on it and I will have to check and see how that is going. Have not been in close contact with that work. 18:32:51 banix: no worries 18:33:02 banix: please let us know if you/he needs any help on that 18:33:16 SumitNaiksatam: sure. thanks. 18:33:22 banix: we have a couple of vendor drivers in, so we have some basic experience on that 18:33:37 just noticed s3wong posted the ODL driver, #link https://review.openstack.org/134285 18:33:40 sounds good. 18:33:53 SumitNaiksatam: the ODL policy driver probably won't make the 11/28 cutoff (though I posted the spec under juno) 18:33:53 s3wong: great that you got this in 18:34:07 s3wong: lets work on that 18:34:07 the spec for it 18:34:52 s3wong: is yapeng planning to help out with the implementation? 18:35:14 SumitNaiksatam: Yapeng is more focus on ODL side 18:35:20 s3wong: ok got it 18:35:41 s3wong: lets have a separate short session to review that spec, so that we can move ahead 18:35:44 he is looking into having another renderer and implement any gap between ODL and OpenStack GBP in ODL GBP 18:35:58 s3wong: okay sounds good 18:36:24 next major work item is testing 18:36:47 we need to increase UT and FT coverage 18:37:00 i will work on weeding out the problem areas 18:37:12 s3wong: where can we find the specs for the ODL GBP driver requirements 18:37:15 Kanzhe is looking at the scenarios that we would need to test 18:37:31 nbouthors: #link https://review.openstack.org/134285 18:37:42 nbouthors: SumitNaiksatam posted a link above.. but here it is #link https://review.openstack.org/#/c/134285/ 18:37:51 ok thanks 18:37:58 Kanzhe: probably you did not get a chance to look at this yet 18:38:09 SumitNaiksatam: Not yet. 18:38:27 Kanzhe: np, some kind of a summary test plan would be good 18:38:31 Plan to start working on it later the week. 18:38:42 Kanzhe: that way multiple people can work on it in parallel 18:38:44 Kanzhe: thanks 18:39:01 just the high level scenarios 18:39:02 SumitNaiksatam: sure. 18:39:11 thanks 18:39:40 that work items on client, ui and heat side 18:39:59 some things are broken in the client (apart from the renaming) 18:40:14 bugs are posted, so please feel free to pick them up 18:40:37 uday is not here, but he is working on several UI pieces 18:40:55 perhaps we can have a separate session to discuss what the team would like to see in the UI 18:41:06 is susaant here? 18:41:37 hemanthravi: i believe the heat patches are mostly merged now? 18:42:05 SumitNaiksatam, yes, will check if susaant pushed the devstack changes 18:42:15 hemanthravi: ok thanks 18:42:52 so the above is mostly what we want to focus for the juno release 18:43:02 #topic Kilo work items 18:43:32 Cathy_: Louis and nbouthors have been working on a number of specs 18:43:52 they are all listed here: #link https://review.openstack.org/#/q/status:open+project:stackforge/group-based-policy-specs+branch:master,n,z 18:44:09 so in parallel we will try to make progress on these spec reviews as well 18:44:53 we have other features like policy tags which will be immediate priorities in Kilo 18:45:20 but i dont want to get too far ahead of ourselves, we can revisit priorities and features when we are actually done with Juno 18:45:34 SumitNaiksatam: +1 18:45:37 SumitNaiksatam: sounds good. 18:45:45 we will discuss and have the Kilo topics as a standing agenda item in the meetings though, until then 18:46:03 we should all be making progress in parallel, ideally! ;-) 18:46:31 SumitNaiksatam, +1 18:46:50 igordcard: that would also include the traffic steering and classifiers :-) 18:46:57 probably it is already too late, but I would like to re-propose the traffic steering blueprint for Kilo 18:46:57 We will submit another BP for integrating with Service Funcitons instantiated outside of GBP or SC, which we think is higher priority than the other BPs we proposed earlier on 18:47:02 SumitNaiksatam, exactly :) 18:47:10 Cathy_: sounds good 18:47:39 igordcard: we will have to focus on getting the current model to work correctly (with the current set of classifiers) 18:48:03 SumitNaiksatam:ok 18:48:05 igordcard: but yes, i agree, lets discuss steering part in parallel, and figure out the scope and applicability 18:48:25 #action SumitNaiksatam to follow up with igordcard on steering spec 18:48:44 SumitNaiksatam: for integrating with Service Funcitons instantiated outside of GBP or SC, we would like to target it at Kilo 18:48:51 SumitNaiksatam, alright :) 18:49:36 Cathy_: yes, mandeep and tgraf (thomas) had some thoughts around this, so lets get everyone on the same page 18:49:55 #action SumitNaiksatam to follow up with Cathy_ on SFC 18:49:57 SumitNaiksatam: since in reality (based on my talk with other service function appliance companies) quite some SF will be instantiated outside GBP and SC 18:50:16 Cathy_: agreed 18:50:39 #topic Open Discussion 18:50:41 SumitNaiksatam: I am looking forwward to talking with you. We will help develeping this. 18:51:04 Cathy_: yes, hence added the AI for me :-) 18:51:08 okay we have 9 mins 18:51:25 we circle back to discussion on the timelines etc 18:51:42 ivar-lazzaro: so does it sound reasonable? 18:52:10 SumitNaiksatam: yes seems fair 18:52:22 ivar-lazzaro: ok, doable? ;-) 18:52:27 Cathy_, we should also discuss if we can use SC to represent the SF instantiated outside 18:52:42 SumitNaiksatam: the work is well distributed, apart from major blockers that may arise it seems a reasonable timeline 18:52:51 hemanthravi: yes 18:52:53 ivar-lazzaro: yeah 18:53:08 one dependency might be rkukura’s extension loading for the vendor drivers 18:53:18 oh i forgot to mention 18:53:24 The TS bp at its latest state is at https://review.openstack.org/#/c/92477, should its topic just be changed, or completely abandoned and recreated afterwards? Also, I'm not sure if gerrit allows changing the ownership... 18:53:30 hemanthravi: I have some thought on this. Let's have a meeting to discuss it 18:53:35 rkukura is working furiously on the packaging 18:53:38 Cathy_, ok 18:53:48 and my expense report ;) 18:53:56 rkukura: :-) 18:54:14 so rkukura would be working on the extension loading after that 18:54:24 hemanthravi: SumitNaiksatam : will Sumit call for a seperate face-to-face meeting for the discussion? 18:54:28 so we need to coordinate a bit so that we work in lock step here 18:54:35 SumitNaiksatam: no problem for rkukura, he still has THREE whole weeks :-) 18:54:36 Cathy_: hemanthravi sounds good 18:54:43 s3wong: :-) 18:54:54 igordcard: to your question 18:54:56 SumitNaiksatam: actually TWO 18:55:06 igordcard: note that your earlier spec was submitted in the neutron specs 18:55:14 rkukura, do the vendor drivers need to make any changes for the new mech 18:55:22 igordcard: however, the GBP specs are now separate from the neutron specs 18:55:42 igordcard: hence i was saying we need to understand which part is relevant where 18:56:30 SumitNaiksatam, igordcard: traffic steering is interesting for GBP; on one hand, we already have classifier in GBP, on the other, we operate on an abstraction level such than TS may be a bit low level 18:56:43 s3wong: agree 18:56:48 but we still need it 18:57:09 igordcard: so --- looking forward to seeing your spec in context of GBP :-) 18:57:11 since the service chain should ideally leverage this 18:57:44 igordcard: i can bring you up to speed if you are cofused between the projects and the processes ;-) 18:58:01 igordcard: lets take it offline 18:58:12 okay anything more to discuss to today? 18:58:19 SumitNaiksatam, Okay then 18:58:20 need to log off, bye 18:58:30 okay thanks everyone for joining 18:58:33 bye 18:58:38 Cya all 18:58:39 till next week, bye! 18:58:43 #endmeeting