19:00:02 #startmeeting Dragonflow 19:00:03 Meeting started Mon Oct 23 19:00:02 2017 UTC and is due to finish in 60 minutes. The chair is oanson. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:04 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:00:06 The meeting name has been set to 'dragonflow' 19:00:09 Hey 19:00:10 Hello 19:00:12 Hi 19:00:16 Who's here for the Dragonflow weekly? 19:00:49 o/ 19:00:56 Hmm... Let's wait another minute then. Maybe pino or mlavalle will join 19:01:20 dimak, leyal, hi! 19:01:43 hi oanson 19:02:25 All right, let's start 19:02:33 hi 19:02:40 irenab, hi. Glad you could make it! 19:02:48 Now we have a critical mass :) 19:02:56 #topic Queens Kick-off 19:03:09 So as you know (cause you've been there), we had the Queens vPTG last week 19:03:19 Lots of interesting discussion. 19:03:36 Can be seen in the following link: (Coming right up) 19:03:50 #link Queens vPTG summary https://etherpad.openstack.org/p/dragonflow-queens 19:04:06 Also summarised nicely in this trello board: (Again, link coming right up) 19:04:16 #link Dragonflow Trello board https://trello.com/b/PM7nah4Z/project-dragonflow 19:04:42 Note that the launchpad is still the main project tracker. The trello board just helps by making things visual. 19:05:05 While here, I'd also like to announce irenab as our new core reviewer! 19:05:34 irenab, your superpowers should already be in effect. 19:05:43 Woohoo 🎉🎉🎉 19:05:46 oanson: thanks , will try it soon :-) 19:05:47 With great power comes great responsibility, so please don't break the project more than anyone else :) 19:05:50 Congratulations 19:06:07 10x 19:06:08 congrats :) 19:06:33 Before we continue, also note I reworked the agenda a bit (link shortly) 19:06:40 #link Agenda https://wiki.openstack.org/wiki/Meetings/Dragonflow 19:07:01 Mostly I wanted to break things down to sub-topics, e.g. features, gates, deployment, etc. 19:07:16 Bugs still gets it's own space :) 19:07:36 All right. Let's get cracking. 19:07:41 #topic Roadmap 19:07:55 Let's start with features. 19:08:37 LBaaS: the spec is up for a while. I'll update it with leyal's comment tomorrow 19:09:00 Then I think it's ready to be torn through and poked holes till it's a fly-net. 19:09:27 oanson: I will take a look asap 19:09:33 irenab, thanks. 19:10:11 DNS is something we just spoke of in the vPTG. lihi is the owner for now (unless someone volunteers). I'm guessing there won't be anything at least till we get back from Oz. 19:10:44 oanson: I think Oz is closer than where you go … 19:11:01 I thought Oz was short for Australia? 19:12:03 Aus, then :) 19:12:15 :-) 19:12:28 No update on L3 flavour. I'll read through the ML to see if there's anything I missed, then talk to the relevant people in Neutron, then update the spec. 19:13:03 ETCD publisher has a review that's up. I ran it and it seems to work nicely. But I only had a chance to check single node 19:13:09 Linky in a second 19:13:29 #link ETCD pub/sub implementation https://review.openstack.org/#/c/495741/ 19:13:48 dimak, irenab, it needs another core, so if one of you could go nuts..... 19:14:04 I'll review it tomorrow morning 19:14:06 Anything anyone wants to add for features before we continue? 19:14:12 dimak, great, thanks! 19:14:29 oanson, Upgrades 19:14:48 dimak, I put it under gates, but we can move it here 19:14:52 I've addressed some of irenab's comments on the spec, and I have a few more 19:14:52 Fire away about upgrades 19:15:07 will be happy for more reviews 19:15:33 shall we discuss the questions on upgrade now or keep it n the review? 19:15:49 I prefer the review 19:15:54 Easier to keep track 19:15:55 fine 19:16:21 I didn't get a chance to read the question yet :) So I refer to your preferences 19:16:43 #link Upgrades (Database Migration) spec https://review.openstack.org/#/c/500647/ 19:16:56 thanks :) 19:16:57 #link Upgrades (Database Migration) impl https://review.openstack.org/#/c/401210/ 19:18:00 Another thing for features: The DHCP overhall spec should be ready to be merged. I'll ask lihi tomorrow to give it her vote so it can move forwards 19:18:15 Hopefully she won't find anything wrong with it :) 19:18:38 Anything else for features? Anything to add or revisit? 19:19:02 Great! Deployment! 19:19:36 OSA - lihi started tackling the gate. OSA upgraded the dragonflow gate as well to work with zuulv3. We're now trying to get the dragonflow gate to do the same 19:19:50 #link Run OSA on DF gate: https://review.openstack.org/#/c/514268/ 19:20:11 Running gates at this hour works less well than UTC morning :) 19:20:39 Then we'll have to figure out why the gate fails *grin* 19:21:07 RPM packaging: No progress. I'll contact the assignee on the python-jsonmodels RFE to see if we can move forwards 19:21:30 I've also added Migration into Dragonflow. I'd be happy if someone with more imagination can think of a better name. 19:21:51 But other than that, there's no progress yet. Will probably also have to wait till we return from Down Under. 19:22:03 Anything to add for deployment? 19:22:19 oanson: I do not understand what you say. what does it mean “Added migrations into Dragonflow”? 19:22:58 irenab, apparently, finishing sentences is the first thing to go when I'm tired :) 19:23:07 I meant adding Migration into Dragonflow to the agenda. 19:23:19 Because it's a target for Queens 19:23:31 ovs ->df? 19:23:41 Yes 19:23:46 ok 19:23:49 But more generally, Neutron DB -> DF 19:24:11 Since the backend shouldn't matter. We can scrub all the info we need from the Neutron DB, hopefully using Neutron API 19:24:53 is there any draft for the solution? 19:24:59 No 19:25:05 Not yet* :) 19:25:27 I can whip something up so we'll have something to work with. Shouldn't take too long 19:25:41 I just asked since was not sure 19:25:45 Is there an owner for this in the queens etherpad? 19:25:59 Trello shows it's orphan 19:26:22 So does the etherpad 19:26:30 :( 19:26:36 (At least they are synchronized) 19:27:09 I can take it. I don't think it's too complex. And I suspect it'll take a while for LBaaS spec to be accepted 19:27:16 maybe we will have some contributor who needs this for his cloud … 19:27:32 try pitching it in sydney :P 19:28:06 irenab, don't hold your breath :) 19:28:09 dimak, I can try 19:28:24 In the project update, I can have a list of orphaned features we want to complete 19:28:42 Maybe throw in L3 flavour as well 19:28:51 sounds like a plan 19:29:09 Great 19:29:26 Anything else for deployment? 19:29:53 All right! Gates! 19:30:10 Grenade gate: patch needs rebase. (Linky soon) 19:30:24 #link Grenade gate patch https://review.openstack.org/#/c/496837/ 19:30:37 Now with zuulv3, that patch can also be updated to include the actual gate 19:30:52 oanson, do we need it in-tree? 19:31:00 Maybe we can use a generic job 19:31:14 In any case, we'll need to reference it in tree. 19:31:21 Hopefully, we can use a generic definition 19:31:21 yes 19:32:01 Tempest gate: snapiri found an interesting bug with dnat. 19:32:04 I saw that my current upgrade patch is zuul happy, so I'll rebase this one on top 19:32:14 dimak, I understand you guys were hacking at it together? 19:32:20 yes 19:32:20 dimak, great! Thanks! 19:32:38 we'll continue tomorrow morning 19:33:13 Great! 19:33:19 there's no clear conclusion yet :( 19:33:38 link to bug? 19:34:04 irenab, no bug yet, we'll open one but I think we need more details 19:34:50 Last gate is the kuryr-integration gate 19:35:11 I didn’t have a chanceto get to it yet 19:35:31 No worries. 19:35:40 I will start with deployment gate probably 19:35:50 I recall the general idea was to take ODL's tests and reuse them, assuming they are black-box? 19:36:06 I think the deployment gate is already up and running and green, no? 19:36:36 oanson: this does not exist yet, I will sync with dmellado once on the progress 19:37:31 All right. Just making sure we're aligned 19:37:44 Anything else for gates? 19:38:18 Last one: Troubleshooting 19:38:37 OSProfiler: snapiri has a very advanced patch up (linky pending) 19:38:58 #link OSProfiler patch https://review.openstack.org/#/c/507851/ 19:39:27 I think it can be already tested and reviewe 19:39:30 reviewed* 19:39:46 I think the WIP tag is pending more testing 19:39:55 any of the gates enables it? 19:40:00 any plans to document it? 19:40:21 dimak, not as of now. Any thoughts how it can be gated? 19:40:30 oanson, maybe rally? 19:40:43 irenab, we should. But seeing the state of our documentation, I don't want to hold back the feature for it. 19:41:09 I jsut would expect some documentation to be part of the patch 19:41:52 Sounds like a valid comment for -1 :). The truth is I haven't considered documentation regarding this. 19:42:00 But it's a good idea. 19:42:27 oanson: reminds me that we have AI for project policies 19:42:38 AI? 19:42:44 Action Item 19:42:55 Yes. 19:43:03 I think I have a draft somewhere. I'll see if I can find it later 19:43:07 butI like the way you’re thinking ;) 19:43:28 Sounds like a research direction 19:43:33 * oanson grins 19:44:16 We also have plans for Skydive and a topology viewer (a tool that creates topological graphs from the DF DB) 19:44:29 topological graphs - not the kind that shows mountains... 19:44:30 per se... 19:45:01 But I think these will have to wait till later in the cycle. 19:45:14 I can also suggest them in Sydney 19:45:36 +1 19:45:39 Anything else for Troubleshooting? 19:45:49 Also, I lied, this isn't the last one 19:46:12 Documentation 19:46:32 The plan is to overhaul our documentation 19:46:47 My name is on there but we haven't made much progress yet 19:47:08 dimak, no worries. We set this action item last week EOW. 19:47:14 yes 19:47:40 Even you need positive time to get things done (as opposed to traveling back in time) 19:48:08 lihi, dimak and snapiri will give a new structure for the docs. Once that's done, we'll split the work between the team to update the docs for both structure and content 19:48:33 In parallel, irenab and I will construct the project policies. 19:48:39 (We haven't had much progress either) 19:49:06 Right. I think that's all I have for roadmap. 19:49:11 Anyone wants to chime in? 19:49:50 #topic Bugs 19:50:04 I don't want to say too much here. 19:50:25 I'll note the good news: We have two new bugs that were reported from outside the core dev team. 19:50:32 That means someone out there is trying us! :) 19:50:45 links? 19:50:47 The bad news is that the critical DF+Octavia bug is still critical. 19:51:10 Bug 1: https://bugs.launchpad.net/dragonflow/+bug/1724541 19:51:11 Launchpad bug 1724541 in DragonFlow "No flow for arp from qg-xxx" [Undecided,New] 19:51:22 dimak: did you have a chance to add details to the bug description? 19:51:28 Bug 2: https://bugs.launchpad.net/dragonflow/+bug/1722839 19:51:29 Launchpad bug 1722839 in DragonFlow "df-controller.service start failed,due to OVS database connection to Open_vSwitch fail" [High,New] 19:52:14 irenab, not yet 19:52:16 The second one is fairly straightforwards. I'll set up an environment tomorrow to try and reproduce, and send a solution for review 19:53:13 Anything else for bugs? A specific one anyone wants to discuss? 19:53:49 #topic Open Discussion 19:53:57 Anyone wants the floor? 19:54:26 Are you sure? It's a dang good floor! 19:54:52 Fine then! 19:54:58 Thanks everyone for coming! 19:55:07 good night! 19:55:23 #endmeeting