00:02:26 <thinrichs> #startmeeting CongressTeamMeeting 00:02:27 <openstack> Meeting started Thu Jul 7 00:02:26 2016 UTC and is due to finish in 60 minutes. The chair is thinrichs. Information about MeetBot at http://wiki.debian.org/MeetBot. 00:02:29 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 00:02:31 <openstack> The meeting name has been set to 'congressteammeeting' 00:02:35 <ekcs> hi all 00:02:35 <aimeeu> hello! 00:02:37 <ramineni_> hi 00:03:01 <thinrichs> masahito might still be out of town this week 00:03:18 <thinrichs> Agenda for the week... 00:03:23 <thinrichs> 1. mascots 00:03:39 <thinrichs> 2. status updates 00:03:42 <thinrichs> Anything else? 00:04:30 <thinrichs> #topic Mascots 00:05:20 <thinrichs> I just heard that OpenStack marketing is going to be making all the big tent projects their own logo 00:05:44 <thinrichs> Then they'll be using those logos to promote OpenStack 00:06:23 <thinrichs> What they're saying is that we need to pick out a mascot by July 27, and then they'll have an illustrator make the logo 00:06:32 <thinrichs> So that's pretty cool 00:06:33 <aimeeu> I had heard the OpenStack logo is going to change but hadn't heard about individual project logos 00:06:48 <thinrichs> They'll announce next week on the -dev mailing list 00:07:32 <thinrichs> The one restriction they're placing on the logo we choose is that it needs to be from the "natural world"... 00:07:34 <ekcs> mascot…. like an animal? I guess I’ll see on the ML. 00:07:48 <thinrichs> an animal or natural feature like mountain or waterfall 00:08:06 <aimeeu> that's not going to be easy 00:08:09 <thinrichs> And we can't use something some other company, etc. already has 00:08:54 <thinrichs> We'll need to come up with a few (say 3) so that they have some to choose from in case of conflicts 00:08:58 <ekcs> hmmm. we could choose the stonehendge to make them determine whether it’s natural =p 00:09:13 <aimeeu> lol 00:09:15 <thinrichs> :) 00:09:41 <thinrichs> Something related to laws governing the universe seems good for congress 00:10:01 <aimeeu> wracking my brain and not coming up with much 00:10:29 <thinrichs> Isaac Newton came to mind. They didn't mention people though. 00:10:38 <thinrichs> (Not that I like Newton, BTW) 00:10:49 <aimeeu> can we steal a Da Vinci drawing? 00:11:37 <thinrichs> Isn't the beaver MIT's mascot b/c it's nature's engineer? 00:11:53 <thinrichs> DaVinci drawings would be cool. 00:12:23 <thinrichs> Anyway, next time you're surfing the web, try to find something that would fit. 00:12:35 <thinrichs> 3 weeks from today is the deadline 00:13:16 <thinrichs> Anything else on this topic? 00:14:10 <thinrichs> #topic 00:14:13 <thinrichs> #topic Newton2 00:14:23 <thinrichs> 2nd milestone for Newton is next week 00:14:45 <thinrichs> which means we need to figure out what we think we can get finished by newton3 00:15:18 <thinrichs> I've seen a couple of specs up in addition to HA 00:15:45 <thinrichs> Those are both masahitos, so I guess we'll wait til next week to figure out if he thinks those can get done by mitaka3 00:16:08 <thinrichs> Besides HA, what items are people hoping to finish by mitaka3? 00:16:42 <ekcs> nothing from me. I couple things I’d like to do that for sure won’t make this cycle. 00:16:47 <thinrichs> Aug30 is newton3 00:17:12 <ramineni_> i thought of completing the migration to horizon plugin 00:17:48 <ramineni_> i think that should be done,eventhough im facing some issues now 00:18:59 <thinrichs> ramineni_: got it. Keep us up to date on how that's going over the next 2 months. 00:19:19 <thinrichs> What about HA? Can we get that done by newton3? 00:19:21 <masahito> hi, sorry late 00:19:28 <thinrichs> masahito: hi! 00:19:41 <thinrichs> masahito: we're discussing what we think we'll finish by newton3 00:19:52 <masahito> got iy 00:19:56 <thinrichs> masahito: we're starting with the non-HA work 00:20:14 <thinrichs> masahito: are you hoping to implement either of your specs by newton3? 00:20:51 <masahito> I forgot when is newton3 00:20:57 <thinrichs> end of August 00:21:02 <thinrichs> So about 2 months 00:21:26 <masahito> oh, I think one of my specs I'm hoping implement 00:22:53 <thinrichs> The selectability of translators one? 00:23:04 <masahito> yes 00:23:39 <thinrichs> masahito: sounds fine. Just trying to get a sense of how much work we're planning and how many changes to the API we're planning. 00:24:01 <thinrichs> masahito: that one extends the current API, correct—so that you can update the config for a datasource? 00:24:40 <masahito> thinrichs: yes, it adds a new config option for datasource driver. 00:25:05 <thinrichs> ok 00:25:10 <thinrichs> Now for HA… 00:25:30 <thinrichs> ekcs: do you think we can get the changes from your spec done in 2 months? 00:26:17 <ekcs> I think it’s within reach. 00:26:33 <ekcs> worth shootng for. 00:26:48 <thinrichs> I also wonder whether we can eliminate the old DSE code so we don't have those 2 branches anymore 00:27:30 <ekcs> I can focus most of my time on it, but I’m still going to need some help I think. 00:27:41 <thinrichs> ekcs: do we have bugs that we can divvy out to people? 00:27:44 <thinrichs> I'm happy to help 00:28:14 <ekcs> haven’t set up the bugs yet. I can do it. 00:28:36 <thinrichs> #action ekcs will set up bugs necessary for HA deployment 00:28:41 <ekcs> thinrichs: for removing old DSE code, we can do that after newton-3 right? 00:29:07 <ekcs> officially stop supporting old DSE in newton-3. do the actual cleanup later. 00:29:27 <thinrichs> I guess we could, but I think usually you save post milestone-3 for testing, bug fixing only 00:29:49 <ekcs> ok. 00:29:51 <thinrichs> The intermediate releases aren't real releases, so we don't need to worry about supporting a feature 00:30:34 <thinrichs> I wonder if removing the old dse would break bryan_att's installers 00:30:57 <thinrichs> We have a new dependency on rabbitmq, right? 00:31:12 <thinrichs> Even if we're using the in-memory version? 00:31:38 <ekcs> umm we shouldn’t need rabbit for inmem. 00:31:51 <thinrichs> That's what I don't remember... 00:31:56 <ramineni_> inmem shouldnt require i guess 00:32:03 <masahito> I think in-memory doesn't need rabbit 00:32:06 <thinrichs> so the in-mem version is implemented in the oslo library 00:32:25 <thinrichs> ok. So install shouldn't be that different. 00:32:51 <ekcs> it’s implemented by kombu. 00:33:26 <thinrichs> The only difficulty I see with removing the old DSE is that the change could touch a bunch of files, and it'd be easy to end up with rebase problems. 00:33:41 <bryan_att> thinrichs: hi joining late. If you removed the dse in Ocata and beyond we can adapt the installers as needed, but I don t know of any explicit dependency 00:34:20 <thinrichs> bryan_att: thanks. 00:35:07 <thinrichs> So we should find a lull in development where pulling out the old dse isn't going to drive us all crazy 00:35:30 <thinrichs> Maybe I'll take a look at how hard that is and let everyone know next week. 00:35:43 <bryan_att> Afaik the approach to defining datasources has changed so there would probably be some effect anyway, but none that would be too hard to handle probably 00:35:44 <thinrichs> #action thinrichs will investigate removing the old dse 00:36:13 <thinrichs> bryan_att: that's the hope—to minimize changes in the installers 00:36:45 <thinrichs> Anything else on this topic? 00:37:03 <masahito> about datasource_manager 00:37:19 <masahito> will we support datasource manager in dse2? 00:39:04 <thinrichs> masahito: that functionality is built into DseNode I believe 00:39:08 <thinrichs> so no 00:39:17 <masahito> ok 00:39:42 <thinrichs> #topic status updates 00:39:46 <masahito> I'll check there is no dependency or not 00:39:58 <thinrichs> aimeeu: want to go first with a status update? 00:40:06 <thinrichs> masahito: sounds good. I'll try to look to 00:40:13 <aimeeu> Sure. 00:41:13 <aimeeu> Thanks for all the feedback on the plexxi bug. I'll be working on the other bugs this week and continuing to learn the code. I've started reverse engineering code into UML diagrams. 00:42:29 <thinrichs> aimeeu: sorry about the trouble with the plexxi bug… 00:42:42 <thinrichs> hope you didn't mind that I pushed a patch on top of yours.. 00:43:15 <aimeeu> Not an issue. That's how I can learn the code. 00:43:46 <thinrichs> I learned something too—guess I didn't remember (or maybe ever understand) how that code worked 00:43:55 <aimeeu> ;) 00:44:27 <thinrichs> Do you know what bug you're working on next? 00:45:39 <aimeeu> I've got the two mentioned in last week's meeting , so one of those.... unless you all have something else you'd like me to tackle 00:46:48 <thinrichs> I haven't found anything new. Maybe I'll find some things as I dig into my action item later this week. 00:47:34 <aimeeu> I'm willing to work on anything, so just assign to me and I'll try to get it fixed 00:48:23 <thinrichs> aimeeu: Terrific! I'll let you know if something comes up. 00:48:29 <thinrichs> ramineni: want to give us a status update? 00:48:39 <ramineni_> yes 00:49:46 <ramineni_> spinned up a new patch to add tests to keystonev3 datasource and added patch to remove the pe dependecy on api 00:50:16 <ramineni_> thats it from my side 00:50:59 <thinrichs> Speaking of your dependent patch… 00:51:56 <thinrichs> ekcs: does it make sense to merge masahito/mine and then push another patch with additional tweaks that's dependent on ramineni's? 00:52:17 <ekcs> yea that makes sense. 00:52:36 <thinrichs> ok... 00:52:39 <thinrichs> running short on time. 00:52:44 <thinrichs> masahito: status update? 00:52:49 <masahito> ok 00:53:23 <masahito> I started to revise my spec and can push it in today 00:54:19 <masahito> and I'll have talk "Congress deep dive" in OpenStack Days Tokyo today 00:54:32 <ekcs> cool! 00:54:34 <thinrichs> Cool! Let us know how it goes 00:54:35 <masahito> oh, today means Thu JST 00:55:04 <masahito> that's from my side. 00:55:11 <thinrichs> masahito: thanks! 00:55:14 <thinrichs> ekcs: status? 00:55:20 <ekcs> 1. I did a manual thread-safety review in the core Congress code. Good news is I didn't identify anything really serious. #link https://review.openstack.org/#/c/335721/ 00:55:20 <ekcs> Essentially two issues: 00:55:22 <ekcs> a) datasource delete needs fixing to avoid problem similar to datasource create 00:55:23 <ekcs> b) When API call comes in specifying a datasource by UUID, the database conversion to datasource name may be interrupted, by the time the greenthread resumes, the returned name may no longer be the name of the datasource specified by the input UUID. 00:55:24 <ekcs> Neither issue is specific to the new_arch. Of course, there are most likely more issues. I'm sure they'll get identified over time. That's one reason the "blocking call" notes are in there, to assist others in identifying more issues in the future. 00:55:25 <ekcs> 2. Minor compatibility changes to pass py35 #link https://review.openstack.org/#/c/338555/ 00:55:26 <ekcs> Oh and back to HA, I think it'd be good to get DSE2 things settled (like the separate nodes patch and the synchronizer patch) before merging major changes. But the work can start. 00:56:52 <ramineni_> ekcs: is there locking mechanism inplace in our DB code, like one thread is updating ..other thread doesn read..something like that 00:57:45 <ekcs> I’m sure there is, but that doesn’t stop other greenthreads from running. 00:57:45 <thinrichs> ekcs: on HA.. are the separate-nodes patch and the synchronizer patch already in review? Or are they changes that we need to make before beginning the HA work? 00:58:06 <ekcs> the separate nodes patch I mean the one you and masahito worked on. 00:58:25 <ekcs> the synchronizer, ramineni_ started something but I’m not sure how close it is. 00:58:34 <thinrichs> And the one ramineni is working on 00:58:49 <ramineni_> i can push up the udpate on synchronizers patch today 00:59:22 <thinrichs> It'd be good to have the synchronizer work for the DSDs and PE, but we only plan on needing it for the PE, right? 00:59:40 <ekcs> thinrichs: yes. 01:00:03 <ramineni_> im thinking of sepearte threads to start synchronizer for PE and DSD 01:00:36 <thinrichs> Okay, so let's try to get those couple of patches in before (i) HA work and (ii) removing old DSE. 01:00:42 <ekcs> that makes sense. Every PE instance should probably run a synchronizer? 01:00:54 <thinrichs> Out of time. Continue for a couple minutes in #congress? 01:00:55 <ramineni_> ekcs, ya 01:01:03 <ekcs> cool. 01:01:13 <thinrichs> #endmeeting