15:00:21 #startmeeting large_scale_sig 15:00:21 Meeting started Wed Jun 23 15:00:21 2021 UTC and is due to finish in 60 minutes. The chair is ttx. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:21 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:21 The meeting name has been set to 'large_scale_sig' 15:00:26 #topic Rollcall 15:00:32 Who is here for the Large Scale SIG meeting, in our new location? 15:00:35 o/ 15:00:36 o/ 15:00:47 o/ 15:00:59 I'll ping people on #openstack-meeting-3 just in case 15:01:22 Our agenda for today is at: 15:01:24 #link https://etherpad.openstack.org/p/large-scale-sig-meeting 15:01:56 #topic Next OpenInfraLive session 15:02:04 Two weeks ago we produced a great episode, continuing our upgrade discussion 15:02:29 The "Large Scale openStack" episodes are some of the most popular on the platform for now 15:02:39 Our next "large scale openstack" show will be... July 15 15:02:51 We need to decide on the topic ASAP so that we can gather guest speakers and promote the show 15:03:01 We have three top candidates, taken from suggestions at https://etherpad.opendev.org/p/large-scale-sig-lightning-talks: 15:03:06 - Inception: Run your OpenStack control plane on top of the infrastructure that it manages 15:03:10 - Monitoring a large scale deployment 15:03:14 - How OpenStack large clouds manage their spare capacity 15:03:17 Any preference? 15:04:02 I dont have personal preference 15:04:15 personally, I'm interested in "How OpenStack large clouds manage their spare capacity" :) 15:04:35 thats a big topic 15:04:35 "Inception" it will be cool, but maybe later? 15:04:41 I also don't have a strong preference 15:04:57 all of them are good candidates IMHO 15:05:16 My only concern with this one is who would do the base presentation? 15:05:18 Inception is maybe not *only* related to largescale 15:05:23 In terms of format, we need to avoid the "too many presentations, no time for discussion" issue we had with upgrades 15:05:35 So I think we should prefer a narrow topic, with one single 5-10min presentation, followed by a panel discussion between 4-5 people 15:06:02 with "spare capacity" do we have anyone ready to do the base presentation? 15:06:02 Monitoring IMO is a very large topic 15:06:18 genekuo: yeah we should probably be more specific 15:06:47 about monitoring, we can talk, but my speaker is not available easily on 15 15:07:04 about spare, belmiro, you were willing to talk, right? 15:07:18 belmoreira 15:07:40 I think is an interesting topic to discuss... I have a lot of questions :) 15:08:16 belmoreira: do you think we can do the show without any leading presentation? 15:08:18 amorin, would you be willing to talk? 15:08:37 I think we have few things to say, but I know we are not the best on this topic... 15:08:47 I think public clouds is even more challenging 15:08:53 I can still ask the team and come back to you asap 15:09:15 We can discuss who the guest speakers could be but i want to nail down the format first 15:09:31 if we do presentation + panel, we need a presenter 15:09:31 I can also ask the team that manages capacity to see if they can talk 15:09:48 alternatively we can do pure panel 15:09:58 if we have many questions, that could work 15:10:04 ttx having only one presentation may be very narrow scope. Probably people are doing different things... public/private clouds are completly different 15:10:45 it's narrow but it's a just a trick to get people engaged 15:11:00 or maybe it's more a presentation to introduce the topic 15:11:00 ttx true! 15:11:01 I will prefer presentations + panels 15:11:29 short presentations about how currently people manages capacities first 15:11:30 Like " the problem with excess capacity: rather than "what we do at CERN with extra capacity" 15:12:05 I agree with belmoreira that private/public cloud may have a completely different view to this topic 15:12:12 genekuo: short presentations did not work well for us last time, just took too much time and not enough engagement between participants 15:13:08 but maybe we can start the discussion by introducing the challenge we want to discuss, using a few slides 15:13:17 that would set the stage 15:13:25 a presentation stating thee problem and then the discussion about what people usually do 15:13:27 and then we can ask the panel questions 15:13:50 sounds like a good plan 15:13:53 I think I just said the same as ttx :) 15:13:55 Sounds good to me 15:14:10 belmoreira: since you find the topic interesting, I nominate you for stating what the challenge is 15:14:53 nomination accepted :) 15:15:25 Also... if people have a point to make that could really use a slide to explain, we can prepare it in advance and plug it during the show 15:15:38 (or keep it in reserve) 15:15:51 That makes the show a bit less like a Zoom call 15:16:07 OK so... 15:16:28 #agreed Topic should be "How OpenStack large clouds manage their spare capacity" 15:16:46 #agreed Belmiro will do the initial "challenge statement" short presentation 15:17:07 question: is any of you using something similar to spot instances? 15:17:08 Beyond Belmiro, who here has a great story on that topic to share and wants to be on the show? 15:17:26 and... who else should we invite to join? 15:17:27 I think I will find someone, or at least something to say 15:17:39 I just need to talk back about this with the team 15:17:39 not a "great story" :) I just think it's a real problem 15:17:45 so you can count on us 15:17:54 I'd like us to rotate speakers if possible, to avoid making it look like a club too much 15:18:00 agree 15:18:08 I can ask the team managing this part. I think they can at least provide some input to the problems 15:18:16 Like we could have someone from CityNetwork or Binero for the public cloud side 15:18:23 rotate the speakers, or rotate the companies? 15:18:33 ok yes 15:18:43 "CityNetwork or Binero" that would be great 15:18:49 That could be in addition to you arnaud if you have a good story to share 15:19:04 I can also be "spare speaker" 15:19:09 if we have enought 15:19:10 since the challenge is clear on public clouds, I would not mind having two 15:19:16 \ok good 15:19:26 Do you have any contact with fuga.cloud? 15:19:31 no 15:19:49 I know they are using openstack, maybe I can find a contact 15:20:00 InMotion could be interesting, they have a "openstack-as-a-service" offering that should generate a lot of spare capacity 15:20:27 Otherwise, i'd love to have someone from Verizon Media (like James Penick since they have a huge capacity 15:21:31 Cloud&Heat could be fun too 15:22:00 I'll reach out to those and ask if they want to join. Any other participant suggestion? 15:22:12 genekuo: do you have a story to tell on that topic? 15:23:24 Capacity is actually manage by a separate team. I can ask if they are willing to talk 15:23:30 #info Belmiro will participate, Arnaud is available in case we need him, Gene will ask the capacity team 15:24:10 #info thierry to reach out to CityNetwork / Binero / InMotion/ Cloud&heat / Verizon Media 15:24:56 belmoreira: if you do the problem statement, my added value is limited on the show, so I was wondering if you could just do the intro/outro and follow the episode script 15:25:00 Then I can focus on triaging questions and selecting good ones... it's been hard for me to do both 15:25:15 So I'll be around, just not "on" the show 15:25:37 belmoreira: are you ok with leading the discussion on that one? 15:25:50 That's a big challenge. i'm happy to try 15:26:01 (We'll work on the script of questions together) 15:26:35 #info Belmiro will emcee the show, thierry will be backstage selecting questions 15:27:05 Alright.. anything else on that? I'll keep y'all posted on the progress in recruiting guest speakers 15:27:49 Sounds like a good plan to me 15:27:56 alright then 15:27:58 #topic Progress/Blockers 15:28:00 * Stage 1 - Configure (amorin) 15:28:04 #link https://wiki.openstack.org/wiki/Large_Scale_SIG/Configure 15:28:22 I havnt been working on this topic for a while, dont have much time at the moment 15:28:58 ++ 15:29:10 * Stage 2 - Monitor (genekuo) 15:29:15 #link https://wiki.openstack.org/wiki/Large_Scale_SIG/Monitor 15:29:31 Not much update on the wiki part 15:30:01 Mostly working on oslo.metrics stuff 15:30:10 no blocker? 15:30:35 I would like to know the gap to a 1.0.0 release 15:30:50 I actually put this topic in open discussion 15:31:13 ok 15:31:26 * Stage 3 - Scale Up (ttx) 15:31:29 #link https://wiki.openstack.org/wiki/Large_Scale_SIG/ScaleUp 15:31:42 no progress recently, openinfra.live has been eating my bandwidth 15:31:46 # Stage 4/5 - Scale Out, upgrade & maintain (belmoreira) 15:31:48 #link https://wiki.openstack.org/wiki/Large_Scale_SIG/ScaleOut 15:31:50 #link https://wiki.openstack.org/wiki/Large_Scale_SIG/UpgradeAndMaintain 15:32:11 I will use the info from the previous shows to update it. Will try to do it for the next meeting. 15:32:26 yeah... Was wondering if there was any insight to extract from our upgrades show that we could use 15:32:39 alright, next topic 15:32:50 #topic Next meetings 15:32:57 We'll have one more IRC meeting between now and our next OpenInfra Live show 15:33:05 So the schedule for next meetings looks like: 15:33:07 - July 7: IRC meeting 15:33:11 - July 15: OILive 15:33:21 After that my proposal would be to skip the meeting on July 21 15:33:37 then have a meeting on August 4 with whoever will be around, one on August 18 and book August 26 for the next live episode. 15:33:53 (I booked August 26 for our next episode) 15:34:05 How does that sound? 15:34:24 No problem for me 15:34:54 #info next meetings: July 7 IRC meeting, July 15 OpenInfra Live, July 21 skip, August 4 IRC meeting 15:35:03 august will be hard for me 15:35:10 I can't join for the July 7 meeting 15:35:10 july ok 15:35:18 hmm 15:35:56 belmoreira: July 7 we would have discussed lastminute prep for the OILive episode, so if you;re not around it's probably not worth it 15:35:59 but I can catch up later with ttx 15:36:12 ok, still useful to have async point 15:36:20 a sync point I mean 15:37:16 amorin: re: August I'm on the same boat... if we think we can't make a show on Aug 26, better push back the date 15:37:25 amorin: when would you be back? 15:37:36 not before september 15:37:48 september 8 I think 15:37:50 maybe we could do the Inception episode on Aug 26 15:37:53 1 will be hard also 15:37:56 since you will sit out that one I guess 15:37:59 yup 15:38:18 then we can plan one around monitoring early October 15:38:30 ok 15:38:35 please keep the meeting for the 7. I have already my Wednesday afternoon blocked, but let's see... I may be able to join a little bit later 15:38:58 #info Next OILive episode after the Jul 15 one will be Aug 26, probably around Inception 15:39:14 #topic Open discussion 15:39:19 Floor is yours genekuo 15:39:56 Yeah, I would like to know if there is any general guideline or a checklist to go to 1.0.0 release for projects 15:40:16 I saw in wiki page that we currently still have 2 gaps (bandit and test coverage) 15:41:15 Currently I've tested the basic features and confirm it working in devstack 15:41:33 re: bandit we should enable it, and make it skip the predictable file path using "# nosec" 15:42:00 test coverage, I have no idea how mucgh we test but it's probably minimal 15:42:12 would be good to add more if we can 15:42:25 but we should aim at 1.0.0 in Xena release 15:42:58 if you think it's not going to significantly change in the near future 15:43:41 OK, I'll start working on bandit and test coverage 15:43:42 in terms of general guidelines, 1.0.0 means you apply deprecation rules, backward compatibility etc 15:44:04 so if you remove an option, you need to continue to support it as deprecated over two openstack releases, etc 15:44:28 so it's good to be suer about the API and config options and general architecture before declaring 1.0.0 15:44:33 sure* 15:45:22 genekuo: you mentioned talking with another user recently... how did that go? 15:45:34 It's generally a good test that the lib works for several use cases 15:45:53 They actually figured out how to test it with the half done code I proposed 15:46:15 Haven't got back to them to see if they have any feedback 15:46:16 ok 15:46:30 maybe that would be good to do before planting a 1.0.0 flag 15:46:30 Let me ask them if there's any point for improvement 15:46:48 ok, does that answer your questions? 15:47:02 Yep, thanks a lot! 15:47:07 Alright then... Anything else, anyone? 15:47:34 Another small question come up in my mind 15:48:01 Do we gather statistics for wiki view count? 15:49:19 Will be interesting to know how much people actually reach the wiki page 15:49:28 guys, I need to run right now, see you next week! 15:49:42 Thanks amorin! 15:49:52 unfortunately no 15:49:56 no usage stats on wiki 15:50:02 ok let's close 15:50:04 Thanks everyone 15:50:09 Thanks! 15:50:10 #endmeeting