20:00:05 #startmeeting Octavia 20:00:05 Meeting started Wed Aug 31 20:00:05 2016 UTC and is due to finish in 60 minutes. The chair is johnsom. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:06 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:00:08 o/ 20:00:08 Howdy, folks! 20:00:08 o/ 20:00:09 The meeting name has been set to 'octavia' 20:00:10 o/ 20:00:15 Hi everyone 20:00:20 hihi 20:00:25 hello 20:00:26 o/ 20:00:41 hi 20:00:45 hi 20:01:00 #topic Announcements 20:01:20 o/ 20:01:26 I just want to remind folks that tomorrow is the last day to use your ATC codes for Barcelona without a fee. 20:01:29 o/ 20:01:41 So if you haven't already registered, do so now.... 20:01:57 Even if you think there's only a marginal chance you can attend. 20:01:59 Free is free, eh. 20:02:06 I can't attend 20:02:12 We had a great mid-cycle last week, some notes are here: 20:02:18 Well, we didn't want *you* to go anyway. 20:02:19 #link https://etherpad.openstack.org/p/lbaas-octavia-newton-midcycle 20:02:20 ;) 20:02:25 o/ 20:02:39 sbalukoff, don't make me cut you 20:02:44 Yes, the mid-cycle was awesome. Got a ton of stuff done. 20:02:47 It was a good kick off for the merge activities. 20:02:55 Any other announcements? 20:03:11 Active-active spec merged. 20:03:15 next week is newton FF. anything you want to merge MUST be in gerrit by wednesday. 20:03:33 err, wait, that's *TODAY*, isn't it? 20:03:37 dougwig: Does that include docs? 20:03:50 I thought I'd have another week or two on them... 20:03:51 dougwig, i think it's today 20:04:04 * nmagnezi checks 20:04:07 Ok, so feature freeze: If it ain't in today, it ain't in. 20:04:32 if it's in gerrit today, you can ask for an FFE. if it's not even in gerrit, you're SOL. 20:05:08 Yes, that is today. This is for "features". For Octavia I will not mind docs patches as they don't risk stability of the code. 20:05:09 https://releases.openstack.org/newton/schedule.html 20:05:45 Can the stats patches goes in today ? I think they're ready. :) 20:05:51 #link https://releases.openstack.org/newton/schedule.html 20:06:11 chenli link? 20:06:34 https://review.openstack.org/#/c/354593/ 20:06:43 #link https://review.openstack.org/#/c/354593/ 20:06:56 #link https://review.openstack.org/#/c/355764/ 20:07:09 #link https://review.openstack.org/#/c/356268/ 20:07:35 and the neutron-lbaas side one : #link https://review.openstack.org/#/c/357653/ 20:07:46 TrevorV: thanks 20:07:56 np chenli :) 20:08:44 Ok, cores, let's have a look at this. 20:08:49 Ok 20:08:51 chenli, I'll take a look through them as well. 20:09:14 I am a bit worried about it as I know some teams have done internal work against stats, so please make sure it doesn't break your stuff 20:09:50 thanks all. We really hope stats can at least be 'working' in this release. 20:10:02 Indeed! 20:10:06 Any other announcements? 20:10:26 BTW, LBaaSv1 was removed from the code last week as well. That is worthy of an announcement. 20:10:30 Haha! 20:10:34 Yes, was just about to say that. 20:10:54 finally :) 20:11:04 Also, in other news, the normal LBaaS talk did not get accepted this year. 20:11:34 So, yeah, odd.... 20:11:37 Moving on... 20:11:47 #topic Brief progress reports / bugs needing review 20:12:02 Docs continue apace. 20:12:25 currently still working on the first scenario test migrating from neutron-lbaas to octavia; hope to complete this week 20:12:34 Yep, I finished up my review on the quick start yesterday 20:12:46 Also, Leslie (the Blue Box tech writer) got back yesterday from over a month of dealing with a family crisis; She's editing the two docs patches currently up for review now. 20:12:52 Had a sick weekend, did an internal task, will be working on the quota review again shortly. 20:13:03 #link https://review.openstack.org/#/c/358450/ 20:13:27 Internal stuff. Will be transitioning away from Octavia to another internal RAX team soon. 20:13:32 I will also be snapping up other docs bugs over the next couple weeks which should hopefully encourage adoption for the Newton release. 20:13:35 it is about the bug https://bugs.launchpad.net/octavia/+bug/1615471. 20:13:35 Launchpad bug 1615471 in octavia "Failover fails when the amphora instance is stoped" [Undecided,In progress] - Assigned to li,chen (chen-li) 20:14:06 I am finishing up the routes patch and will be looking at the intermittent bug I saw:1614311 20:14:33 Sweet. 20:15:04 nmagnezi also added this bug to the agenda: 20:15:13 #link https://review.openstack.org/#/c/344658/ 20:15:27 The git -C issue on Centos: #link https://review.openstack.org/#/c/353265/ 20:15:30 This one for lbaas 20:15:37 #link https://review.openstack.org/#/c/317862/ 20:16:13 Yeah, chenli I still don't think that is the right approach on the CentOS issue. I still need to do research on that 20:16:24 yup. this was already merged but got revered due to an issue that is now fixed. this is verified by the namespace scenario gate we fixed last week with the help of dougwig 20:16:49 He also raised the refactor patch: 20:16:55 #link https://review.openstack.org/#/c/346728/ 20:17:51 Lots of stuff to review, eh. 20:18:04 sbalukoff: like any FF. :) 20:18:06 Yeah. 20:18:12 Yep! 20:18:30 #link https://review.openstack.org/#/c/299998/ 20:18:31 johnsom The git -C one ? why ? -C is change the patch before --git-dir, using --git-dir directly should not affect anything. 20:18:34 This one I'm not sure about: https://review.openstack.org/#/c/317862/ 20:18:40 this was also in the agenda johnsom :) 20:18:54 a patch to allow creation by net id instead of subnet id got submitted (similar to how nova does net id). if anyone is interested, to have any hope of newton, it needs some thrashing: https://review.openstack.org/#/c/363302/ 20:18:57 Do we want to mess with tempest plugin work for neutron-lbaas? Or just focus on the plugin for Octavia 20:19:43 johnsom: Focus on OCtavia. 20:19:52 Octavia focus fifrst 20:19:53 first 20:20:00 We're hoping to get things merged in the "O" cycle, right? ;) 20:20:06 the plugin for octavia is already merged, isn't it? 20:20:06 Yes 20:20:08 then once we have that good and working, perhaps apply that to neutron-lbaas afterwards 20:20:17 Remove network from amphorae when delete a member #link https://review.openstack.org/#/c/347121/ 20:20:18 fnaval: +1 20:21:12 nmagnezi As we are merging neutron-lbaas into Octavia, we may just want to focus on building out the tests in Octavia and not migrate neutron-lbaas to tempest plugin 20:21:50 indeed, if that's done, and we swap to running octavia jobs in n-l, we can just stop running the n-l tests. 20:22:17 i also forsee that the tempest plugin dir structure that's needed in neutron-lbaas will not adhere, currently. though, I might be wrong as well. more resarch is needed 20:22:43 johnsom, perfectly understandable. but since it is not going to be completed in the upcoming release maybe it's worth considering 20:23:03 In Newton? 20:23:11 nmagnezi The plan is to complete it in Ocata 20:23:15 Oh, the tempest tests won't make it in there in any case. 20:23:32 sbalukoff, yes, Newton 20:24:11 For Newton, yeah, since the tests won't be converted for Newton, this should probably what to merge until Ocata is opened 20:25:00 johnsom, that was my intention, you phrased it better :) 20:25:21 Ah, ok 20:25:55 So, a bunch of stuff to review. The list is long and given the deadline, some will likely not make it. 20:26:44 If it doesn't, I'm sure it will land early in Ocata 20:27:03 Yep. 20:27:14 #topic Patches requiring discussion/consensus 20:27:27 Any other patch related discussion today? 20:27:29 i really hope to at least have https://review.openstack.org/#/c/299998/ and https://review.openstack.org/#/c/344658/ merged. as they give some sort of flexibility (HA wise) for those who still did not migrate to Octavia atm. 20:28:03 johnsom, not on my end. 20:28:06 need some input on this https://review.openstack.org/#/c/355115/ 20:28:09 Gotta remember how to run the namespace driver. It's been a loooooong time since I've done that. 20:28:13 about the test cases 20:28:28 sbalukoff: fire up a new devstack, is the easiest way. 20:28:31 as I mentioned in the last comment of the patch 20:28:36 sbalukoff, it's actually kinda simple, i can share my local.conf if it helps 20:28:54 nmagnezi: Sure. Send it to me. 20:29:23 Guys, drop that "#link" before your links so it can be tracked easily in the notes. 20:29:25 (Not in this meeting, obviously.) 20:29:32 #link https://review.openstack.org/#/c/299998/ 20:29:41 #link https://review.openstack.org/#/c/344658/ 20:29:43 sbalukoff It looks like you have a -1 on bana_k's patch 20:29:56 #link https://review.openstack.org/#/c/355115/ 20:30:23 to make trevorv happy: 20:30:25 #link https://review.openstack.org/#/c/363302/ 20:30:35 Huh. Ok. I'll have to have another look at it, then. 20:30:52 Only me facing the bug: #link https://bugs.launchpad.net/octavia/+bug/1615471 ?? Can not believe that. :( 20:30:52 Launchpad bug 1615471 in octavia "Failover fails when the amphora instance is stoped" [Undecided,In progress] - Assigned to li,chen (chen-li) 20:30:54 thanks johnsom sbalukoff 20:31:15 sbalukoff, how to run haproxy namespace driver #link https://bugs.launchpad.net/neutron/+bug/1613251/comments/1 20:31:15 Launchpad bug 1613251 in neutron "HAproxy scenario tests cleanup fail with a StaleDataError (LBaaSv1 and LBaaSv2)" [High,Fix released] - Assigned to Nir Magnezi (nmagnezi) 20:31:21 #link https://bugs.launchpad.net/octavia/+bug/1615471 20:31:41 Bot commands have to be alone on their lines... sorry chenli 20:32:24 TrevorV: O! 20:32:28 chenli I expect stopping amphora nova instances is a rare occurrence. I just want to make sure to test that heavily as the previous patch was 20:33:03 johnsom, i believe all failover fails due to the same reason.... At least that happens in my tests 20:33:09 Once I get some test runs, that will be a quick merge 20:33:41 chenli, no, with the recent failover patch it went through a bunch of testing, like 30+ test grid 20:34:13 johnsom, the bug is actually very easy to reproduce. Use a non-admin user to create a loadbalancer, and allow neutron-lbaas create the vip port (the default conifg in devstack) 20:34:28 So unless something changed in the last week or so that broke it, it was working really well 20:35:05 johnsom: nova would check the project-id for port and nova instance, thus==> nova would complain the project id is not the same. 20:35:25 johnsom I just re-produced it in a whole new set-up yesterday. 20:35:45 johnsom and the new set-up is not installed by myself. 20:36:44 Hmm, I will have a look. The ports are plugged into the nova instance and all of them should be owned by the lbaas project 20:37:10 None of them should be owned by the tenant 20:37:17 johnsom doesn't we only plug the vrrp port, and set the vip as the address_pair to the vrrp port ? 20:37:45 Yes, and neutron creates a port for the aap address 20:38:03 johnsom: then no need to plug vip port to amphora directly 20:38:44 I agree on the surface, but previously we were only plugging the port your patch proposes to not plug, so I want to do more testing on this patch. 20:38:56 I think when neutron-lbaas create the vip port, it use user's credential. 20:39:27 I didn't change the plug for vrrp port, just remove the vip port plug. 20:39:54 As I said before, it needs more research and testing. 20:40:27 sure! 20:40:32 #topic Open Discussion 20:40:39 Ok, other topics today? 20:40:59 * Frito observes the crickets 20:41:01 *crickets* 20:41:06 Jinx! 20:41:06 lol 20:41:08 lol 20:41:30 Ok, so just a lot of patches to see if we can merge. 20:41:30 so what octavia related talks got accepted? 20:41:36 None. 20:41:43 No LBaaS talks either. 20:41:45 nmagnezi: none. no load balancing talks *at all*. 20:41:50 nmagnezi Sadly, no LBaaS related talks got accepted 20:41:51 that's very sad 20:41:52 And very few relevant Neutron talks. 20:42:02 the talks this year are heavily NFV, and don't represent the actual networking community much at all. 20:42:15 IMO 20:42:16 Yeah, it's pretty crap. 20:42:28 Yeah, given it was the #1 feature users were interested in from the last user survey I think it is very sad 20:42:38 So hey! Barcelona will probably feature me sitting in a hallway doing mid-cycle type stuff. Y'all are welcome to join me! 20:42:50 sbalukoff: i'm in 20:43:08 i actually learned a lot from your sessions in past summits. shame we won't have another one like this in the near summit. 20:43:13 Yeah, I want to have a LBaaS community discussion meeting and then find a table and work on merge activities personally 20:43:19 unfortunately I'm not going so I will not join 20:43:24 nmagnezi: Come join us in the hall. :D 20:43:44 sbalukoff, sadly i can't travel this time. I hope next time I will join you :) 20:43:47 diltram you can join us in the virtual hall known as IRC.... grin 20:43:52 Yep! 20:43:56 * TrevorV isn't invited 20:43:57 So can you nmagnezi! 20:44:02 johnsom: true :) 20:44:04 No, we don't want to see you TrevorV. 20:44:06 this is my plan :) 20:44:06 ;) 20:44:26 Ok folks. Let's get busy with reviews. 20:44:32 Seriously though, I appreciate your work, TrevorV. 20:44:32 sbalukoff, where can i find details about this? (time, date, how to connect) 20:44:45 Even when it does break L7 in ways we didn't notice for months. XD 20:44:54 hahahaha ;P 20:45:09 nmagnezi: Well, if you're not there in person, the place to be is the #openstack-lbaas channel on IRC. 20:45:27 nmagnezi Why don't you ping us when it gets a little closer. We should know where the neutron-ish rooms will be 20:45:52 Oh, yeah, remote, it will be IRC. Likely on Barcelona time 20:45:57 Yep. 20:46:23 We might be able to do some video stuff if we can find a place where we won't be disturbing others, and if the bandwidth doesn't suck. 20:46:27 But the surest bet is IRC. 20:46:35 sbalukoff, johnsom, ack. i hope I can make it (timezone diffs, it's almost midnight here) 20:46:36 sbalukoff, that's you're fault too, I think you were a +2 on it 20:46:37 o_0 20:46:43 Yeah, ummm, when has bandwidth not sucked at a summit???? 20:47:08 Ok, thanks folks! 20:47:12 TrevorV: Oh yes, I never said I wan't also culpable. ;) 20:47:16 #endmeeting