18:01:30 #startmeeting networking_policy 18:01:31 Meeting started Thu Jan 18 18:01:30 2018 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' 18:02:01 Do we want to discuss the potential talk in vancouver summit? 18:02:07 annakk: sure 18:03:04 do we want it to be demo-like? for both aim and nsx? 18:03:07 #topic Vancouver Summit 18:03:24 #topic-fail :( 18:03:24 #topic Vancouver Summit 18:03:26 maybe with focus on NFV since its in trend 18:03:29 SumitNaiksatam: thx! 18:03:50 #chair tbachman rkukura annakk 18:03:51 Current chairs: SumitNaiksatam annakk rkukura tbachman 18:04:09 tbachman: thanks :-) 18:04:17 rkukura: SumitNaiksatam: we should talk next week about the summit 18:04:29 tbachman: sure 18:04:33 * tbachman and rkukura will be in SJC next week 18:04:40 tbachman: sure 18:05:11 BTW I will be in PA Feb 5-8 18:05:20 annakk: when are the proprosals due? 18:05:26 annakk: are you going to propose a talk? 18:05:31 PA as in Pennsylvania, or PA as in Palo Alto? 18:05:36 I don't know, probably soon 18:05:39 Palo alto :) 18:05:43 :) 18:05:54 due Feb 8 18:06:17 annakk: good to know 18:06:30 I have to confess I'm very bad with talks in front of big audience :) 18:06:39 feb 8 gives a reasonable amount of time to plan 18:06:42 annakk: I’m most certainly worse 18:06:43 I tend to change colors and forget english words 18:06:52 lol 18:06:58 I tend to forget english words as well, even though it’s my native tongue 18:07:04 :) 18:07:11 I even do that when I’m not nervous :P 18:07:13 annakk: dont worrry about it you will be fine 18:07:15 tbachman: lol 18:07:45 annakk: so are you doing anything with GBP+NSX+NFV on the product side? 18:08:27 Not yet, I was dragged out of openstack lately, but hopefully will be back to it in couple of months 18:08:42 So hopefully we'll have something to show in may 18:08:58 annakk: okay 18:09:09 Seems like services, NFV, etc. are certainly hot 18:09:50 annakk: were you able to test GBP/Pike with NSX? 18:10:05 not yet unfortunately.. 18:10:08 okay 18:10:22 i am just trying to get a baseline of where we stand 18:10:55 but i think we can safely assume that at least GBP/Pike+NSX will work at the time of the summit 18:11:14 yes 18:11:16 SumitNaiksatam, tbachman: Where do we stand on GBP/Pike+APIC testing? 18:11:31 rkukura: I believe we have pike artifacts, but no installer yet 18:11:41 SumitNaiksatam: is that correct? 18:11:47 tbachman: right 18:11:55 we are waiting for the installer 18:12:14 ok 18:12:15 the coverage at this point is the devstack AIM gate job 18:12:50 annakk: I haven’t checked, but is there a VMware gate for pike? 18:13:08 tbachman: yes there is, but not for master/pike 18:13:25 no, I'll move it from ocata to pike once its tested 18:13:36 annakk: thx! 18:14:24 okay so i think we should at least be ready with pike to show for the summit 18:14:40 however, the community would already have moved to Queens by then 18:15:09 but i think we should still be okay with Pike, agree/disagree? 18:15:56 SumitNaiksatam: unless you have a secret Queens branch in your back pocket ;) 18:16:15 was GBP in sync with the community in previous talks? 18:16:35 annakk: mostly yes, but it depends on what we want to show 18:16:50 if we want to show integration with product older release in perfectly fine 18:17:49 ok 18:18:33 my hope is that queen’s sync might be easier, but i havent tried 18:19:06 will it help someone to get travel approvals if he proposes the talk? I don't need travel obviously 18:19:21 i think we are good 18:19:42 if talk gets accepted we should be able to travel, at least one of us 18:19:57 (most likely more than one though) 18:20:22 ok 18:20:32 rkukura: tbachman: right? 18:20:45 at least we dont have visa issues 18:20:52 SumitNaiksatam: I’d think so — we’ll obviously need to talk w/folks out there next week 18:21:03 tbachman: right, but i think it shoud be fine 18:21:04 I think topics is going to be the biggest question. 18:21:12 we havent done this for a bit, so it should be okay 18:21:16 It’s 5/21-24? 18:21:27 Vancouver is just a puddle-jumper for you all in SJC ;) 18:21:28 we might have other things lined up as well, so this probably be in addition to that 18:21:54 nice time to visit there, too 18:22:01 tbachman: you bet 18:22:05 yes :) 18:22:24 tbachman: i forget had you done the OpenStack @ Vancouver last time? 18:22:33 SumitNaiksatam: alas, no 18:22:45 * tbachman has only been to the (2nd) Austin summit 18:22:53 then you should definitely go 18:23:02 really loved Vancouver, rkukura what say? 18:23:12 * tbachman used to visit Vancouver when he grew up in the PNW 18:23:12 definitely would like to go back! 18:23:31 tbachman: oh of course, i forgot about that! 18:23:35 anyway 18:24:01 annakk: so apart from the “NFV” integration, what would be other things that you would like to highlight? 18:24:51 just the fact that we have a new driver 18:24:53 btw, tactically speaking #link https://www.openstack.org/summit/vancouver-2018/summit-categories/ 18:24:59 annakk: okay :-) 18:25:21 annakk: did garyk have something specific in mind? 18:25:42 I'll talk with him 18:25:54 annakk: okay 18:26:04 annakk: do you work on the container’s side at all? 18:26:32 no, its the other part of the team 18:27:28 I'm currently involved with terraform.. 18:27:53 btw maybe we should write a terraform provide for GBP :) 18:27:55 annakk: nice 18:28:00 *provider* 18:29:09 annakk: good pointer, will look 18:29:44 I hear field is shifting to terraform from heat 18:29:50 apart from the NFV track, i was wondering if this “Container Infrastructure” was fit 18:29:56 annakk: ah okay 18:30:38 its clear that we are fishing a bit here 18:30:52 but i think its fine, we brainstorm a bit, and go back and think 18:31:25 do you have use cases with containers? 18:32:15 annakk: yes with containers, but have to think about GBP 18:33:42 annakk: does your team work on cloud foundry? 18:33:56 I never thought of combining gbp with nsx container solution, will have a think about that 18:33:59 yes 18:34:43 annakk: okay 18:34:57 rkukura: your thoughts? 18:35:53 SumitNaiksatam: Definitely would like to participate, but we need to put some thought into possible topics 18:36:00 rkukura: right 18:36:30 i think where we are mostly at is, we have focussed less on new features in the past few cycles and more on the production side 18:36:43 so most of the shiny new features have already been presented 18:36:49 Does it make any sense to kick-off trying to make GBP an official project? 18:37:20 rkukura: sure 18:37:21 rkukura: do we need to be in sync with the release cycle for that? 18:37:34 annakk: i would imagine 18:37:51 perhaps the talk/presentation can serve as a springboard for that as well 18:37:59 Agreed we don’t have much in the way of new features. But we might want to talk about GBP’s NFP vs. Neutron’s SFC 18:38:28 Can you say a few words about sfc integration? I've seen the commit 18:38:32 annakk: to your point, there might some procedural effort involved to get it to the point of applying 18:38:55 annakk: this is mostly on the aim driver side for now 18:39:08 we would need to discuss the GBP mapping 18:39:27 that discussion can hopefully happen in this meeting/community 18:39:59 ok 18:40:08 annakk: but we can definitely discuyss that patch if you are interested 18:40:19 assuming some of us go, would we want to procure a meeting room for a GBP breakout dev session? 18:40:23 we can check if Ivar can join the meeting, better to hear it from the author 18:40:50 I’ve gotten confused about the different summit tracks again 18:40:56 is there a dev session associated with this? 18:41:00 is it all dev? 18:41:36 tbachman: if they do provide space for unofficial projects, we should (assuming we are still not official until then) 18:41:37 tbachman: I think the talk and the dev session are not related 18:41:57 thx! 18:42:23 I haven’t been to a summit in a while, but I don’t think there are official development meetings for projects anymore 18:42:47 rkukura: I think you’re the most recent summit-attendee from our group 18:42:48 annakk: even showing the mere integration, say GBP+NSX is a good demo 18:42:56 SumitNaiksatam: +1 18:42:58 tbachman: lol 18:43:10 wait annakk did sydney, right? 18:43:18 right — meant from our SJC group 18:43:18 thought so 18:43:24 sorry 18:43:25 tbachman: in australia there were quite a few for neutron, I think they are called workshops or something like that 18:43:32 * tbachman meant no slight to annakk 18:43:44 ok 18:43:53 tbachman: no my bad, didnt mean it that way :-) 18:44:13 np :) 18:44:43 one way or the other, we are one happy OpenStack family 18:44:56 :) 18:45:09 * tbachman queues Disney music 18:45:15 :) 18:45:21 okay shall we table the discussion for today on this? 18:45:29 sure 18:45:34 SumitNaiksatam: ack 18:45:37 may be chew on this a bit more and regroup over emails 18:45:43 #topic Pike Sync 18:45:51 so thanks for all the reviews 18:46:08 SumitNaiksatam: thanks for all your hard work! 18:46:10 in my estimation all the work is mostly done, including some clean up 18:46:13 tbachman: np 18:46:19 but please do try to test it 18:46:25 oh one big thing 18:46:31 will do 18:46:39 i actually watned to put it on the agenda, completely forgot 18:46:52 so there is a quirk with rollbacks 18:47:13 SumitNaiksatam: in pike only? 18:47:32 tbachman: yes 18:47:50 so earlier: with session.begin(subtransactions=true) 18:47:59 will create new nested transactions 18:48:42 so in the innnermost transaction, if you fail (exception is raised), nothing is committed 18:49:13 however with the new pattern: with context_manager.writer(context) 18:49:40 when you nest multiple of those, there is no new nested transaction 18:49:53 the same transaction is preserved 18:50:35 and some kind of a soft commit happens after the scope of each “with” ends 18:50:56 I think there was a thread in emails about making sub-transaction behavior default with new pattern 18:51:01 I can try to find it 18:51:29 SumitNaiksatam: Can you explain what you mean by “soft commit”? 18:51:33 so if you happened to raise an exception and caught that exception somewhere in the nesting, and did not raise it, the soft commits might actually get committed 18:52:02 So the inner transaction might commit even when the outer transaction rolls back? 18:52:05 rkukura: i was waiting for that question from you :-) 18:53:02 rkukura: so the outermost transaction actually did not rollback, because the exception was caught inside the outermost transaction 18:53:45 the way this is different from using the older pattern, is that the soft commit would not happen in the first place with the exception being raised from the inner transaction 18:54:19 regarding soft commit - i meant update of sqlalchemy data structures (which it uses for bookkeeping) 18:54:34 we saw this happen in Ivar’s patch with the UTs 18:54:58 So if the exception is raised in the inner TX and not caught in either the inner or outer TX, do both inner and outer TX roll back? 18:55:12 i was hoping to explain with that specific example, but sorry i didnt do my prep work, so i dont have it ready to copy-paste here 18:55:24 rkukura: yes, that case works fine 18:55:25 SumitNaiksatam: Are you saying the datastructures reflect updates that are not committed in the DB? 18:56:00 rkukura: right 18:56:13 and those are eventually committed if the exception is not re-raised 18:56:30 so at the face of it might seem thats how it should work 18:56:59 but its different from how it used to work 18:57:21 it might not actually have a deployment implication, but at least manifested in tests 18:57:32 and good to understand the different in the usage patterns 18:57:46 we got around this by nesting the older pattern with the newer pattern 18:57:50 more details next time 18:57:54 we are out of time 18:58:12 anything else urgent? 18:58:13 ok, thanks SumitNaiksatam! 18:58:23 nothing urgent from me 18:58:26 i meant difference in usage patterns 18:58:50 rkukura: annakk tbachman: thanks for joining 18:58:52 SumitNaiksatam: bye! 18:58:59 bye 18:59:01 keep the discussion going on the summit talk 18:59:03 bye 18:59:07 #endmeeting