18:00:29 #startmeeting networking_policy 18:00:30 Meeting started Thu Jan 15 18:00:29 2015 UTC and is due to finish in 60 minutes. The chair is SumitNaiksatam. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:31 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 18:00:34 The meeting name has been set to 'networking_policy' 18:00:46 #info agenda https://wiki.openstack.org/wiki/Meetings/Neutron_Group_Policy#Jan_15th.2C_2014 18:01:11 hi 18:01:25 rkukura: hi 18:01:31 nbouthors: good to see you here 18:01:36 #info http://lists.openstack.org/pipermail/openstack-dev/2015-January/054249.html 18:01:48 release formally announced! ^^^ 18:01:48 SumitNaiksatam: hello 18:02:29 hi 18:02:33 #topic Bugs 18:02:50 KrishnaK_: hi 18:03:02 #link https://bugs.launchpad.net/group-based-policy/+bug/1411237 18:03:02 SumitNaiksatam: hi 18:03:16 mageshgv: thanks for jumping on this right away 18:03:36 SumitNaiksatam: no problem 18:03:49 good to know that it was not affecting the datapath, but just a display issue 18:04:15 i have tested this several times, and it never manifested until yesterday 18:04:27 SumitNaiksatam: yes, in create and update we were directly using the values passed in api 18:04:36 mageshgv: okay 18:04:43 hence the missed db update did not cause functional issue 18:04:52 mageshgv: we would need to backport the fix once we merge on the master 18:04:54 s3wong: hi 18:05:00 hello 18:05:00 SumitNaiksatam: okay 18:05:01 mageshgv: correct 18:05:04 sorry, a little late 18:05:13 s3wong: np, just getting started 18:06:05 mageshgv: we need to backport the doc patch that you fixed too 18:06:46 SumitNaiksatam: yes, I will cherrypick the patch to the stable juno branch 18:07:33 mageshgv: yes, you need to use the -x option 18:07:37 #link https://wiki.openstack.org/wiki/StableBranch#Proposing_Fixes 18:07:48 Youcef: hi, good to see you 18:08:21 SumitNaiksatam: ok. will do that 18:08:33 #link https://bugs.launchpad.net/group-based-policy/+bug/1404412 18:08:37 is ivar here? 18:09:20 his patch is passing the gate: #link https://review.openstack.org/#/c/144816/ 18:09:34 i guess we need recheck a few times to see if its consistent 18:10:18 rkukura: are there any plans to follow up on: #link https://bugs.launchpad.net/group-based-policy/+bug/1383947 18:11:05 I’ll track progress on the neutron subnet pool BP 18:11:16 Plan is to use that as soon as its available 18:11:38 rkukura: can we link the BP? 18:12:00 I’ll look 18:12:39 rkukura: thanks 18:13:24 anyone seen anything critical? 18:13:42 in the release besides this? 18:14:02 okay moving on 18:14:05 BP: https://review.openstack.org/#/c/135771/ 18:14:42 oh good, seems like its merged 18:15:20 rkukura: do we need a follow up patch in gbp? 18:15:32 SumitNaiksatam: The BP merged - are you saying the code merged too? 18:15:42 oops, my bad 18:15:51 thats the spec you were pointing to 18:16:31 right 18:16:36 ok moving on 18:16:41 #topic Packaging 18:17:20 KrishnaK_: thanks for trying out the CentOS-based install 18:17:40 KrishnaK_: How has that gone? 18:18:01 SumitNaiksatam, rkukura: In the middle of it ... Will update tommorrow 18:18:12 KrishnaK_: thanks! 18:18:55 KrishnaK_: thanks 18:19:03 rkukura: any updates over last week on the RDO? 18:19:05 KrishnaK_: Are you using the koji-built or copr-built el7 RPMs? 18:19:33 Next step on RDO is getting our packages included in the RDO repos. 18:19:49 I think that requires they setup some internal CI for the packages. 18:20:14 Do we have any tempest tests they could use as a basis for that? 18:20:15 rkukura: by they you mean RDO? 18:20:25 SumitNaiksatam: yes 18:20:48 They are working to open this up so it can be done externally, but are not quite there yet 18:20:58 rkukura: not really, we had requested a separate repo to house our tempest tests 18:21:32 rkukura: however we were told to integrate the tests with our project 18:21:41 rkukura: I have CentOs installed ... do you recommend copr or koji ? 18:21:43 rkukura: is that a requirement to moving forward? 18:22:23 rkukura: we have some integration test scripts 18:22:27 KrishnaK_: I think the copr-built RPMs (with centos in the names) are the ones we’ll go with going forward 18:22:47 SumitNaiksatam: I’ll find out what is really needed for their CI. 18:23:06 rkukura: thx. will install cope-built RPMs 18:23:22 rkukura: okay 18:23:39 KrishnaK_: OK, please email or find me on IRC if you hit any issues or questions 18:24:04 rkukura: Thanks a lot. will ping you. 18:24:10 the ubuntu packages are also available: #link https://launchpad.net/~group-based-policy-drivers/+archive/ubuntu/ppa 18:24:54 rkukura: any other packaging updates at your end? 18:25:28 SumitNaiksatam: Are their Ubuntu instructions somewhere similar to https://openstack.redhat.com/Neutron_GBP? 18:25:52 rkukura: we dont have those yet :-) 18:26:12 for now the instructions are: apt-get install 18:26:22 SumitNaiksatam: Only other thing is that we will also need to figure out how packages for RHEL OSP will be produced and hosted 18:27:18 rkukura: yeah, what would need to do to start that process? can we start that in parallel while the work on RDO is progressing? 18:27:46 If there are other vendors that plan to offer any kind of supported GBP solutions on top of RHEL OSP, we may want to work to see how the common parts will be provided and supported. 18:29:38 rkukura: okay, lets take this offline with hemanthravi and songole 18:29:48 SumitNaiksatam: +1 18:29:57 rkukura: we will need to sync up with ronak as well 18:30:07 rkukura: what about ODL in that context? 18:30:43 SumitNaiksatam: That’s a good question, but I guess that would need to be supported by Red Hat themselves, unless another vendor was involved. 18:31:55 rkukura: hmmm, so we dont need to do anything for adding the ODL support? 18:32:02 yapeng: good timing to join! ;-) 18:32:21 just discussng packaging the ODL integration for RHEL OSP 18:32:23 SumitNaiksatam: Hi 18:32:51 SumitNaiksatam: My guess is that Red Hat would need a compelling business case for them to offer supported neutron+gbp+odl all on their own 18:33:04 rkukura: can we not provide support? 18:33:23 i mean this team, for ODL? 18:34:01 I don’t think this team can enter any kind of official partneship agreement that would let Red Hat sell support subscriptions 18:34:19 Maybe Red Hat could ship the bits, but not provide support 18:34:42 rkukura: okay got it 18:35:05 This would be as a preview feature or something like that 18:35:13 okay 18:35:34 moving on 18:35:51 #topic Kilo Blueprints 18:36:19 most of the team is still swamped in post-release activities, packaging, support, etc 18:36:29 so we are behind on planning for the next release 18:36:36 LouisF: hi 18:36:54 nbouthors: any updates with the exploration at your end? 18:36:59 hi SumitNaiksatam 18:37:10 in the context of leveraging ODL integration for the L4-7 classification? 18:37:24 SumitNaiksatam: could not start this week. 18:37:33 nbouthors: ah ok, np 18:38:20 LouisF: anything that you wanted to discuss here with regards to your blueprints? 18:38:49 SumitNaiksatam: please review and comment 18:38:53 needles to say, we owe you a review (and apologies for the delay)! 18:39:05 LouisF: indeed! ;-) 18:39:24 LouisF, SumitNaiksatam: can you provide a link to the bp? 18:39:37 s3wong: sure, the specs have been in review for a long time 18:39:52 link to the gerrit review then :-) 18:39:56 #link https://review.openstack.org/#/q/status:open+project:stackforge/group-based-policy-specs+branch:master,n,z 18:40:09 the issue was that i havent created a kilo branch for specs 18:40:21 i mean i posted a patch to do that, but it kept failing the gate 18:40:37 i put that aside and havent gotten back to it yet 18:40:47 but that should not stop from the reviews 18:41:15 sorry i meant “kilo directory” not “kilo branch" 18:41:38 once the kilo directory change is merged, LouisF, you can just move the files around in the same review 18:41:54 SumitNaiksatam: sure 18:42:05 meanwhile the reviews can proceed 18:42:08 nothing stopping that 18:42:11 SumitNaiksatam: Will we update our tempate and tempate-validation to match what neutron is doing in kilo? 18:42:18 template 18:42:21 rkukura: yeah, that is part of the problem 18:42:33 the neutron templates were updated, we did not 18:42:37 and hence that breaks the gate 18:42:43 or at least thats how i understand it 18:44:27 #topic Open Discussion 18:44:37 anything else anyone wants to bring up? 18:45:19 i have been doing some homework on identifying the next set of priorities 18:45:22 kudos to all on recent work 18:45:25 for kilo release 18:45:28 LouisF: thanks 18:45:39 currently i just have a long laundry list 18:45:49 hope to structure it a little better by next week 18:46:27 some of us could meet in person to discuss this as well 18:46:49 or we have #openstack-gbp 18:46:59 SumitNaiksatam: how many new features are on the table now? 18:47:09 s3wong: dont ask! :-) 18:47:22 like i said its a long laundry list 18:47:29 so we have to obviously prioritize 18:47:43 and one of the major input to that is the user feedback 18:47:55 which is one of the reasons i am not rushing to decide on this 18:47:58 we also need to get feedback from users now that we have a release, and be prepared to act on it 18:48:05 SumitNaiksatam: while helping out yapeng and yiyang on ODL policy driver, we found major gap between ODL GBP and OpenStack GBP 18:48:10 rkukura: exacntly, just saying that 18:48:29 s3wong: we need to let people use the latest release 18:48:41 SumitNaiksatam: so I am planning on looking to see if we can perhaps fill some of those gaps by Lithium timeframe 18:48:46 s3wong: at least for a few days, and then gather feedback 18:49:01 SumitNaiksatam: though I am not saying the OpenStack GBP side should stall :-) 18:49:05 s3wong: yes, i believe keith is also working on the ODL on this 18:50:03 so if you have any users looking at the latest GBP release, and have feedback, please send it to me or the whole team 18:50:08 we would really love to hear 18:50:17 SumitNaiksatam: what is this gap between ODL GBP and OS GBP? 18:50:40 LouisF: external policy, service policy, hierarchical contracts... 18:51:03 s3wong: those are additional features 18:51:21 SumitNaiksatam: but it is a gap :-) 18:51:31 they exist in OS GBP but not ODL GBP 18:51:49 s3wong: yes, but i think you alarmed LouisF with the way you put it 18:51:52 :-) 18:51:57 LouisF: hence a gap :-) 18:52:12 the gaps are not in basic workflows 18:52:17 LouisF: but I still encourage you or others to try out ODL GBP with OpenStack GBP :-) 18:52:27 LouisF: don't let me deter you :-) 18:52:29 SumitNaiksatam: so the driver needs to deal with the gaps 18:52:31 although there might be some rough edges 18:52:36 LouisF: yes 18:52:47 just to be very clear - 18:52:58 the OpenStack GBP model is bigger super set 18:53:12 LouisF: it does, it throws a bunch of NOT IMPLEMENTED exceptions :-) 18:53:12 we dont expect all policy drivers to comply with all parts of the model 18:53:27 s3wong: that is exactly as designed 18:53:32 SumitNaiksatam: that is expected 18:53:40 LouisF: exactly 18:54:05 we expect the policy drivers to implement the PTGs, PTs, PRS, PR, PC, PA 18:54:10 I don’t disagree, but from a user perspective, the semantics/features should be as similar as possible independent of the back-end 18:54:45 rkukura: the reference implentation is 18:55:04 rkukura: beyong that its the vendor’s capacity to support a particular model 18:55:27 SumitNaiksatam: What I mean is that the various implementations should ideally be substitutable without the user noticing 18:55:47 rkukura: ideally, yes 18:55:57 rkukura: while that would definitely be a desirable goal 18:55:59 There will be cases where things need to differ, but that should be a last resort 18:56:07 we have noticed that appraoch stymies innovation 18:56:08 odl GBP has a whole lot of stuff to support contract and subject selection: clauses, relators, etc 18:56:27 LouisF: yes, you could identify gaps either way :-) 18:56:47 LouisF, we have some prelimary implementation juno mapping driver 18:57:05 if required we can tag parts of the model as “core” 18:57:09 Ideeally the differences would be in extensions, so its explicit to the user which extensions they are using. Not all vendor backends might support an extension, but those that do should be equivalent 18:57:09 yapeng: yes iv'e looked at that 18:57:22 i talk with s3wong before, we would like to fill the gap in kilo on OS side. also work on ODL side for Lithium release 18:57:48 I need to drop off a couple minutes early - see you all later! 18:58:09 in summary, gaps exist everywhere :-) 18:58:10 SumitNaiksatam: in theory, when we got started on the ODL GBP project, the goal is to have the ODL GBP to be a superset --- though obviously it didn't happen that way :-) 18:59:04 s3wong: i cant comment becaue i was not part of that discussion, however i did not get that memo ;-) 18:59:10 SumitNaiksatam: you mentioned a face-to-face - any plans for that? 18:59:21 yapeng: yes, if we are able to fill a portion of the gap in ODL side, then we will correspondingly update the GBP OBL policy driver on OpenStack 18:59:22 LouisF: yes will update if that comes about 18:59:37 SumitNaiksatam: thanks 18:59:49 we have gaps in the openstack GBP model itself 18:59:57 to satisfy some important use cases 19:00:06 SumitNaiksatam: obviously OpenStack GBP progressed so much better than we all expected :-) 19:00:07 SumitNaiksatam: such as? 19:00:11 and we need address in Kilo :-) 19:00:49 LouisF: policy tags, enhancments to the classification 19:00:58 SumitNaiksatam: ok 19:01:31 ok lets wrap up for today 19:01:34 we are over time 19:01:38 SumitNaiksatam: some of that would be covered by my BPs 19:01:45 thanks for all the feedback folks, keep that coming! 19:01:51 bye 19:01:52 LouisF: yes absolutely 19:01:56 bye 19:02:00 bye all! 19:02:03 bye 19:02:03 bye 19:02:04 by 19:02:05 #endmeeting