08:01:21 #startmeeting publiccloud_sig 08:01:21 Meeting started Wed Mar 15 08:01:21 2023 UTC and is due to finish in 60 minutes. The chair is tobberydberg. Information about MeetBot at http://wiki.debian.org/MeetBot. 08:01:21 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 08:01:21 The meeting name has been set to 'publiccloud_sig' 08:01:28 I've re-used the agenda from last time 08:01:36 Perfect! 08:01:48 since from what I gathered, nothing in depth was discussed due to low attendance 08:01:55 At least we are 3 this morning :-) 08:02:05 morning guys 08:02:56 Exactly 08:02:57 Morning! 08:02:59 Lets kick it off 08:03:00 Agenda here: #link https://etherpad.opendev.org/p/publiccloud-sig-meeting 08:03:12 Please put your name in there as well 08:03:34 #topic 1. Support in 3rd party tools that support specific clouds. (Pick up the discussion from last time, when we ran out of time) 08:03:51 To be honest, not really sure where we left that... 08:04:01 I don't think we reached it. 08:04:56 Then I can stop looking for that discussion then ;-) 08:05:03 me neither, can't really get what exactly it is about. At least partially it may be to "let's run some form of acceptance tests with SDK/OSC and give soft of verified label for them" 08:05:05 I've used a couple of tools that are either OpenStack compatible or S3 compatible, and they'll sometimes have preconfiguration for OpenStack based clouds. I was wondering about how we can make this easier for public clouds. 08:05:33 s/preconfiguration for/preconfiguration for some" 08:05:45 problem with all those tools and public clouds is that they all have different configurations and expectations 08:05:57 aye 08:06:11 i.e. openshift installer is assuming networking resources are having tags - but this is not necessarily the case 08:06:31 and not every tool is doing proper validation or discovery of facts and capabilities of the cloud 08:06:48 microversion support can be intersting as well 08:06:53 I can understand those however - mostly this is a relatively unknown info 08:06:55 A little bit like the work in SDK where we try to deal and abstract the complexity 08:07:32 aye 08:07:42 yes, SDK target is exactly to get rid of this whole complexity and based on discovery do one thing or another thing, or just return exception - sorry, not possible 08:08:11 amount of code right now in sdk is pretty much loudly describes this complexity - 150k+ loc 08:08:31 so it is no surprise others are eventually not doing things right 08:08:56 yea, that is a hard nut to crack indeed 08:09:42 Are private cloud "configurations" considered in the SDK work as well, or only focus on public clouds? 08:10:12 well, focus is on the capability of the cloud to let info be discoverable 08:10:36 so in reality there is absolutely nothing what would prohibit it running well for private clouds, assuming they do things right 08:10:48 I think private clouds should certainly be supported, however I'm sure a number won't want their details leaked. So if configuration files, or discoverability can be used, that'd be best. 08:10:56 ok ok 08:10:58 right 08:11:04 in reality we are running sdk in the backend of our cloud with custom profiles and hacks - so it works. You just need to know how to cook it 08:12:33 Okay, so this is kinda "hmmm, they should use the SDK if possible", although not all the tools are Python, so, . 08:13:37 yes, we spent really hell amount of effort to switch OSC/Ansible/Zuul/... towards using SDK. So this is definitely what everybody in the py world should do once trying to work with any OpenStack based cloud 08:14:56 Basically it comes down to the discoverability that is lacking information? 08:15:09 Or just that OpenStack can be configures in millions of ways so it is impossible to actually make everything discoverable 08:16:03 both statements are correct 08:16:20 Maybe. Some of the tools will have drop downs with a list of pre-known clouds that include URLS or required tweaks. 08:16:47 that is precisely the point why interop tests in current form are currently telling nothing about real interoperability for users 08:17:07 +1 08:17:33 puck: knowing URLs is not a point here - it's more that once you know identity endpoint of the cloud everything else must be discoverable 08:17:40 So, I guess this loops make to the verification tools we're talking about. 08:17:44 gtema: +1 08:18:09 for that I am currently working on reforming SDK tests so that they can be executed against any cloud (public or private) 08:18:17 nice 08:18:49 and that would then tell: if those tests are fine chances are great that SDK/OSC/Ansible/.. will work fine on this cloud for user (not for admin) 08:18:57 Yea, I assume that is the way to go to make that real, probably not super easy but having "real" interop test is a must have as a starting point... 08:19:12 exactly 08:19:45 BTW ... do you have what you need for that work to continue gtema or can we helt or assist in any way? 08:20:49 so far I run those tests for Cloudera and still [ :-( ] wait for other clouds. But in reality we should consider how to treat those results 08:20:55 Credentials to clouds or what not 08:21:07 that was the point for the forum discussion - discuss further items on that 08:21:28 yeps, which is a super good topic 08:21:58 I'm happy to facilitate some access to Catalyst Cloud as a test target. 08:22:55 so, jumping to next topic here... 08:22:56 #topic 2. How to proceed with the starters in https://etherpad.opendev.org/p/publiccloud-sig-specs 08:23:01 gtema: i think, i did point you towards https://github.com/SovereignCloudStack/docs/blob/main/community/cloud-resources/cloud-resources.md - that way we can hook you up with access to the clouds we have at hand there 08:23:22 great 08:23:24 I said I would give it a stab ... haven't been able to find the time unfortunate 08:23:54 fkr - right. But I was sofar not able to proceed on that 08:25:36 tobberydberg: is that something where pairing would help? 08:25:46 aye should be no issue to get you an account on scs1 08:26:03 s/scs1/gx-scs 08:26:08 tobberydberg: I'd be up for that, likely I'm not able to assist you with knowledge, but it would help me ;) and it might help you allocating the time :) 08:26:56 In general for the spec ... good goal to have a first version submitted before the Summit and have a forum session about it? 08:26:58 hahaha, good point fkr, and maybe just what is needed :-) 08:27:42 I'm totally up for that, lets se if we can find a time in the upcoming weeks 08:28:01 +1 08:29:08 would it make sense to have a session during the upcoming virtual ptg on that as well? 08:31:15 (did I break the conversation? :) 08:31:31 no, everyone is thinking ;-) 08:31:36 for my pov - +1 08:32:56 +1 08:32:58 Lets have that as one focused session for a start 08:32:59 With that said, we need to book some time for that as well 08:33:58 But that is the next topic. Lets leave this topic as is and plan for a focused session around this at the PTG 08:34:10 how does that work? 08:34:10 (booking the slot during ptg) 08:34:56 #topic 3. Anything to prep for PTG session? 08:34:57 Yea, saw some email about that... Will see if I can find it 08:35:01 using PTG bot it seams like 08:37:02 Is the session scheduled now? I don't see a time. 08:37:14 https://ptg.opendev.org/ptg.html 08:37:26 If we decide on time I can make sure to book 08:38:22 we can pick the day as well? (sorry, stupid first timers questions...) 08:38:58 MONDAY, TUESDAY 1300 UTC ? 08:39:09 sweet 08:39:27 Monday > tuesday 08:40:00 Can we solve this without a poll? ;-) 08:40:09 Just do it. ;) 08:40:15 +1 08:40:26 I might be able to attend, unsure, I'll have my kids that week. 08:41:06 Monday 1300 and then 3h?, so 13-16 UTC 08:41:33 +1 08:42:56 Can't SDK team in there yet, was trying to find if that clashed with your team gtema 08:43:08 But I'll book that for now and we hope for the best 08:45:09 sorry, was afk - I have not booked anything for SDK yet 08:45:13 Done ... except ptg-bot hanged on my last request 08:45:15 so feel free to pick suitable time 08:45:21 +1 08:45:29 tobberydberg: thanks for taking care of that 08:45:50 Now it is booked and PTG bot woke up again :-) 08:47:48 heh 08:51:15 Okay, anything else today? 08:51:33 nothing from my side 08:52:16 not from my pov 08:52:36 Possibly of interest to folks here: https://www.politico.com/news/2023/03/10/white-house-cloud-overhaul-00086595 08:54:06 lol 08:57:38 guess we can close it then? 08:57:42 Okay, sounds like we're done then. 08:58:05 yup, have a nice day folks 08:58:07 \o 08:58:41 \o 08:58:50 Yes 08:58:54 * puck heads home from the office, 21:58 here 08:59:26 \o 08:59:50 got disconneced here... 08:59:51 Thanks for today and have a great day! 08:59:52 #endmeeting