08:00:18 <oanson> #startmeeting Dragonflow
08:00:19 <openstack> Meeting started Mon Jul 24 08:00:18 2017 UTC and is due to finish in 60 minutes.  The chair is oanson. Information about MeetBot at http://wiki.debian.org/MeetBot.
08:00:20 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
08:00:21 <oanson> Hello.
08:00:22 <openstack> The meeting name has been set to 'dragonflow'
08:00:24 <lihi> Hi
08:00:25 <leyal> Hi
08:00:28 <dimak> Hey
08:00:31 <oanson> Who's here for the Dragonflow weekly?
08:00:36 <irenab> hi
08:01:00 <oanson> Cool. Let's start
08:01:03 <oanson> #topic Roadmap
08:01:22 <oanson> SFC
08:01:29 <dimak> Almost in :)
08:01:30 <oanson> dimak, Are we done?
08:02:11 <dimak> I think it is pretty done
08:02:16 <oanson> What do you mean? How long does it take to get a tiny patch of 2222 lines reviewed?
08:02:19 <dimak> Unless someone has more comments
08:02:20 <dimak> https://review.openstack.org/#/c/424146/
08:02:34 <dimak> 😉
08:02:50 <lihi> 😇
08:03:00 <lihi> On it
08:03:15 <dimak> Thanks!
08:03:43 <oanson> I'll report that I didn't do squat last week. In my defense I was sick. But LBaaS and RPM packaging didn't see any cpu time
08:04:23 <oanson> L3 flavour - dimak I saw you updated the patch. I asked there a question
08:04:33 <irenab> dimak, is there some networking-sfc test cases that you checked ?
08:04:40 <oanson> lihi, leyal, do you have any questions about it, so we can move forwards?
08:04:55 <dimak> oanson, I'll check and get back to your comment
08:05:05 <dimak> irenab, I'm looking into that as well
08:05:10 <lihi> Nope
08:05:23 <oanson> dimak, I'd rather discuss it here so the spec can be accepted today, if everyone agrees
08:05:37 <dimak> I'll upload L3 service provider for Dragonflow today or tomorrow
08:05:49 <irenab> based on my recent experience with Trunk ports, we should pay more attention on the test coverage
08:06:01 <dimak> My plan is to retire DF's L3 plug in in favor of using the vanilla L3 router plugin with DF service provider (driver)
08:06:19 <oanson> irenab, there is only so much a test can cover - it's not an integration.
08:06:44 <oanson> I asked if the solution in the spec *requires* modification in the other l2 implementations. We should support the other implementation out-of-the-box if possible
08:07:07 <irenab> oanson, maybe we can discuss it after the roadmap, but coverage should be improved
08:07:11 <oanson> dimak, that's L3s mechanism similar to L2s ML2, right?
08:07:20 <dimak> Yes, as I said above, I'll check the ref implementation first and post a reply
08:07:27 <dimak> oanson, yes
08:07:27 <oanson> irenab, agreed. But I recommend low expectations :)
08:07:46 <irenab> oanson, its a long journey
08:08:01 <oanson> irenab, sorry?
08:08:57 <irenab> oanson, adding more testys and improve stability takes time. So, agreed about expectations
08:09:09 <oanson> Sure
08:09:29 <oanson> dimak, regarding the l3 flavor - did you answer my question? (I may have missed it)
08:09:40 <dimak> Yes
08:09:48 <irenab> regarding L3 flavor, it should not impact L2 plugin, only if there is something broken there
08:09:57 <dimak> Service provider is a kin to mech driver in ML2
08:10:05 <dimak> Well, not exactly
08:10:08 <irenab> there is similar work done by yamamoto for Midonet
08:10:12 <oanson> dimak, I meant regarding my comment on the spec.
08:10:28 <oanson> Like irenab said, we shouldn't have to modified other people's l2 implementation
08:10:32 <dimak> oanson, I answered that as well, I need time to check ref impl
08:10:38 <oanson> I see.
08:10:50 <oanson> Sorry - I didn't realise that was the answer.
08:10:57 <oanson> All right. Let's continue then.
08:11:04 <oanson> ETCD publisher
08:11:07 <oanson> lihi, ?
08:11:26 <lihi> I thought I made the new library to work, but I was fooling myself. But I made some progress, so hopefully now this time it will work (I'm still testing it )
08:11:59 <dimak> If you need some extra eyes, upload a patch :)
08:12:22 <dimak> Tag it with 🚧
08:13:09 <lihi> Does gerrit supports emojis?
08:13:33 <dimak> I'm not sure :(
08:13:41 <oanson> dimak, Under construction?
08:13:46 <dimak> Yes
08:13:56 <oanson> No emoji for work in progress?
08:14:28 <lihi> I guess that letting the gate fail will be faster than opening the systemd logs in my VM
08:14:42 <oanson> Or anywhere
08:15:34 <oanson> All right, let's move on
08:15:48 <oanson> leyal, what about the PXE/DHCP ?
08:16:21 <oanson> I see the spec is merged
08:16:36 <leyal> 2 patches in 2 waiting for review and 1 i still  need to do some changes
08:16:52 <irenab> links?
08:16:55 <leyal> The spec and the change in the DB ..
08:17:15 <leyal> https://review.openstack.org/#/c/475167/27
08:17:24 <leyal> https://review.openstack.org/#/c/475718/23
08:18:54 <oanson> Sure. We'll look at them
08:19:10 <leyal> I also write a draft (in dragon-flow) of about requirement for Ironic support..
08:19:27 <oanson> That would be cool to see
08:19:54 <irenab> leyal, not sure if related, but there are wierd failures at the gate on second patch
08:20:23 <irenab> non voting
08:20:26 <oanson> irenab, I think it's a gate thing
08:20:28 <leyal> I will look at that ..
08:20:55 <irenab> oanson, any known issue on gate?
08:21:04 <oanson> Not that I know
08:21:05 <dimak> Gate should be good for now
08:21:28 <oanson> leyal, I think you should leave the gate for now
08:21:39 <leyal> Ok
08:21:51 <oanson> If this shows up more often we'll assign someone to look into it (then it may be you)
08:22:14 <oanson> Anything else for roadmap?
08:22:57 <oanson> #topic Bugs
08:23:20 <oanson> dimak was nice enough to close bug 1690775 for us :)
08:23:20 <openstack> bug 1690775 in DragonFlow "Remove special handling for lport/ofport in local controller" [High,In progress] https://launchpad.net/bugs/1690775 - Assigned to Omer Anson (omer-anson)
08:23:32 <dimak> Not closed yet though :(
08:23:52 <dimak> The last one kinda fails
08:24:29 <oanson> still the py35 thing?
08:24:42 <dimak> Nope, the one that removes update_lport
08:25:03 <dimak> https://review.openstack.org/#/c/486411/
08:25:21 <dimak> But most of the code is gone
08:26:19 <oanson> Do you need another set of eys?
08:26:23 <oanson> eyes*?
08:26:48 <dimak> Nope, its the issue we discussed earlier today, about port locality
08:27:13 <oanson> Sure
08:27:39 <oanson> Anything else for bugs?
08:27:47 <dimak> I've opened some bugs with irenab on trunk app
08:28:08 <dimak> most of those have been fixed or have a fix at the gate
08:28:20 <dimak> This one is still not though https://bugs.launchpad.net/dragonflow/+bug/1705503
08:28:20 <openstack> Launchpad bug 1705503 in DragonFlow "Trunk subport will not be available after controller restart" [High,Confirmed]
08:28:20 <irenab> oanson, there is one on child port not being set to Active
08:28:24 <oanson> Sure. Where they classified with importance?
08:29:34 <dimak> Yes
08:29:35 <oanson> irenab, this one: bug 1705397 ?
08:29:35 <openstack> bug 1705397 in DragonFlow "Sub port of Trunk port is not updated to Active status" [Undecided,New] https://launchpad.net/bugs/1705397
08:29:46 <irenab> yes
08:29:58 <oanson> irenab, is it blocking integration with kuryr?
08:30:26 <irenab> similar to nova, kuryr expects port to be aCtive before moving on with processing Pod creation request
08:30:37 <irenab> yes, blocking
08:31:07 <irenab> without port being Active, it assumes that Data Plane is not set yet
08:31:19 <oanson> Sure. Then set to high.
08:31:43 <irenab> oanson, I do not have the power ...
08:31:44 <oanson> I'll try and fix it this week.
08:31:53 <oanson> Already done.
08:32:36 <oanson> We have many bugs of Undecided importance
08:33:03 <oanson> Anyone feels like taking a look at those? Just to sort the importance?
08:33:12 <dimak> Sure
08:33:26 <oanson> dimak, thanks.
08:33:40 <oanson> Anything else for bugs?
08:34:32 <oanson> #topic Open Discussion
08:34:43 <oanson> I want the floor
08:35:04 <oanson> I will be on PTO next week, and the one after that.
08:35:10 <oanson> Anyone wants to take over the meetings?
08:35:49 <oanson> Cool
08:35:57 <irenab> I can take the one next week, the week after I am on PTO
08:36:15 <oanson> irenab, thanks.
08:36:34 <irenab> oanson, it will be the US friendly time, right?
08:36:35 <oanson> Worst comes, we'll cancel the one in two weeks
08:36:38 <dimak> Yes
08:36:39 <oanson> Yes
08:37:05 <oanson> Anything else on this?
08:37:30 <oanson> Second item is test coverage.
08:38:36 <oanson> tox has a thing where it shows the coverage by unit tests
08:39:00 <oanson> And the fullstack tests are a bit of a mess, seeing as they test flows rather than functionality.
08:39:18 <dimak> maybe we can do more tempest
08:39:28 <oanson> dimak, we need to get tempest to work first
08:39:30 <dimak> or at the very least make the job we have green
08:39:36 <dimak> Yeah
08:39:49 <oanson> But then, yeah, tempest is the best fullstack thing we can do
08:40:03 <dimak> Maybe we should roadmap it
08:40:18 <oanson> Let's finish the dnat issue first, and take it from there
08:40:37 <dimak> I promise the DNAT patch will be done one day :)
08:40:37 <oanson> That is our most critical point right now
08:40:47 <oanson> Before the PTG?
08:41:03 <dimak> Note that metadata service is not getting any requests either
08:41:18 <irenab> dimak, oanson , context/bug?
08:41:40 <oanson> Tempest is broken because dnat doesn't work out of the box. That is bug 1636829
08:41:40 <openstack> bug 1636829 in DragonFlow "Conflict between flat network and DNAT app" [Critical,In progress] https://launchpad.net/bugs/1636829 - Assigned to Dima Kuznetsov (dimakuz)
08:41:41 <dimak> 1 moment
08:42:17 <dimak> I have a fix, that puts dnat on top of provider app
08:42:17 <oanson> Not *just* because of dnat, but dnat is the first thing we need to fix
08:42:40 <oanson> Linky?
08:42:47 <dimak> With the lport/ofport bug mostly out of the way, I hope I'll have cleaner approach
08:42:58 <dimak> https://review.openstack.org/#/c/475362/
08:43:18 <dimak> It is somewhat half-baked right now
08:43:44 <oanson> dimak, could we maybe have a workaround: Increase the priority for dnat just for dnat's IPs?
08:43:58 <dimak> I can try that
08:44:10 <oanson> It will at least make the bug Hgih, not Critical
08:44:13 <dimak> 🤢
08:44:14 <oanson> High*
08:44:24 <oanson> I don't have that emoji
08:44:48 <dimak> 😷*
08:45:25 <oanson> Sorry, :(
08:45:59 <oanson> I am releasing the floor
08:46:06 <oanson> Anyone wants it?
08:47:09 <oanson> Anything else for open discussion?
08:47:17 <oanson> Anything else in general?
08:47:31 <dimak> Lunch!
08:47:40 <oanson> Hooray. Let's go eat!
08:47:46 <oanson> Thanks everyone for coming.
08:47:55 <oanson> #endmeeting