Thursday, 2021-08-05

*** pojadhav- is now known as pojadhav04:07
*** akekane_ is now known as abhishekk06:41
*** pojadhav- is now known as pojadhav07:09
*** rpittau|afk is now known as rpittau07:47
*** diablo_rojo is now known as Guest337108:27
*** Guest3371 is now known as diablo_rojo11:56
*** pojadhav- is now known as pojadhav13:05
*** pojadhav- is now known as pojadhav13:39
*** pojadhav is now known as pojadhav|out14:23
gmannTC meeting in ~8 min from now14:52
* diablo_rojo is ready14:53
spotz_here14:54
dansmithI'm double booked, so will be a little distracted, FYI14:57
jungleboyjI will be there.14:58
gmann#startmeeting tc15:00
opendevmeetMeeting started Thu Aug  5 15:00:08 2021 UTC and is due to finish in 60 minutes.  The chair is gmann. Information about MeetBot at http://wiki.debian.org/MeetBot.15:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.15:00
opendevmeetThe meeting name has been set to 'tc'15:00
gmann#topic Roll call15:00
gmanno/15:00
dansmitho/15:00
diablo_rojoo/15:00
spotz_o/15:01
jungleboyjo/15:01
belmoreirao/15:01
gmann#topic Follow up on past action items15:01
gmanngmann to write to k8s steering team and cc diablo_rojo_phone15:02
yoctozeptoo/15:02
gmanndone, sent  email15:02
gmann#topic Project Skyline (diablo_rojo)15:02
gmanndiablo_rojo: go ahead15:02
diablo_rojoSo! At the last PTG Horizon talked to a group about a project they had been working on to replace Horizon15:02
diablo_rojoThe Horizon team was on board with this given their lack of resources to get horizon the updates it really needs. 15:03
jungleboyjIsn't Skyline the same name that VMware uses for one of their ddashboards?15:03
diablo_rojottx and I have been working with the team on their project- skyline- getting it ready to be proposed. 15:04
yoctozeptoit better not be15:04
yoctozeptothough vmware already has its horizon15:04
diablo_rojoThis one is from 99cloud? 15:04
yoctozeptojungleboyj: it seems it's the name of their support panel15:04
jungleboyjSkyline is the name of one of the VMware products.15:04
gmannand as separate dashbaord project or within Horizon project ?15:04
diablo_rojoThey are almost ready to go- just working on getting their tests setup with zuul. 15:04
belmoreiradiablo_rojo this would be a replacement for Horizon?15:04
diablo_rojogmann, separate to replace horizon15:05
diablo_rojois the main goal from what I understand15:05
gmannhumm15:05
yoctozeptodiablo_rojo: but separate also governance-wise?15:05
dansmithman.15:05
yoctozeptodansmith: hmm?15:05
diablo_rojoI think the groups would merge? I don't know all the details yet. 15:05
dansmithyoctozepto: as in.. "wow"15:06
yoctozeptodansmith: +215:06
diablo_rojoThe horizon folks would be on board though based on the discussions back at the PTG15:06
diablo_rojoThey already have some repos setup on opendev: https://opendev.org/skyline/15:06
yoctozeptohmm, a different namespace for now15:06
diablo_rojoIf anyone wants to take a look15:06
gmannis there ML thread started on this or just PTG discussion ?15:07
yoctozeptoso is the question whether they can move into the official namespace?15:07
yoctozeptogmann +115:07
belmoreirawould "skyline" have the developer/community to drive the new project that Horizon is lacking?15:07
diablo_rojogmann just ptg so far15:07
gmannor collaborate in horizon team instead of separate alternate project?15:07
diablo_rojoOnce they have zuul setup I think the plan is to start a ML thread to start the proposal process15:07
diablo_rojobelmoreira, yes it seems so + ideally the remaining horizon team15:08
fungii expect we'll want to chat with foundation legal/trademark folks on the name too before it gets officially accepted as part of openstack15:08
diablo_rojofungi, +115:08
yoctozeptoI guess we would need devstack integration too15:08
yoctozeptoare they including this in their "zuul setup"?15:08
jungleboyjfungi: ++15:08
dansmithyoctozepto: yeah good point, we would want to see a very stable devstack integration before anything serious I would think15:08
gmannbut before that, we need to discuss with the governance structure they want to propose and how Horizon team thinking on this?15:08
jungleboyjConcerned with the overlap with VMware here.  Despite the fact that I like the name.15:08
diablo_rojoyoctozepto, I am not sure, but if we want to require it before they apply to become an offical project, they are a super receptive team 15:08
diablo_rojojungleboyj, thats an easy fix? changing the name isn't really a concern I don't think? 15:09
yoctozeptodiablo_rojo: ok, then we want :-)15:09
belmoreirawondering about the process to deprecate Horizon and introduce the new project15:09
jungleboyjdiablo_rojo:  Gets harder the longer the project is around.  :-)15:09
diablo_rojobelmoreira, yeah I am not sure what that would look like15:09
gmannyeah, what we will do/plan with horizon?15:09
dansmithnot saying anything negative about this team I don't know, but.. I would really want to avoid making any rash decisions15:09
diablo_rojojungleboyj, then lucky this is rather new ;) 15:09
yoctozeptore horizon: ask horizon folks15:09
dansmithlike, keep them both around as alternatives for many cycles15:09
diablo_rojodansmith, I totally get it. 15:10
diablo_rojoYeah I think we shouldn't get rid of horizon right away or anything15:10
dansmithwe would want to see some uptick in adoption, reports from the field that it's a suitable replacement, etc before we even mark horizon as deprecated, IMHO15:10
yoctozeptoif horizon folks want to refocus15:10
yoctozeptothen let them refocus15:10
yoctozeptodansmith: +215:10
gmanndiablo_rojo: well, Horizon facing resource issue so I prefer the collaboration with horixon team and improve there what is lacking15:10
gmanndansmith: ^^15:10
yoctozeptowe really need devstack + some prod deployment projects to pick it up and users to deploy before declaring anything on horizon15:10
gmannalternate projects things is what we did mistake in past right?15:11
diablo_rojogmann, the horizon team is on board with project Skyline though? 15:11
yoctozeptodiablo_rojo: I would assume this would be q to you15:11
spotz_Yeah having an option to choose would be good for the shorter term as each component would need a dashboard created15:11
fungithe horizon maintainers already concluded on the ml thread earlier this year that the way forward was to rewrite horizon's frontend portion completely15:11
fungi(and that they have their hands full maintaining the version they already have)15:12
jungleboyjWow.15:12
diablo_rojoYou can find some of the PTG discussion here: https://etherpad.opendev.org/p/xena-ptg-horizon-planning on line 14115:12
diablo_rojoyoctozepto, which question? Sorry I am starting to lose track lol15:13
fungiwell, the conclusion was that they picked a dead-end framework, so there's no incremental way to migrate it to newer paradigms. this is the world of web application frameworks apparently15:13
diablo_rojofungi, thank you, yes. Was trying to find that ML thread15:13
fungii can look for it too, though i'm in three meetings at the moment15:13
dansmithfungi: afaict, the "world of web application frameworks" is a total disaster :)15:14
gmannom PTG, it is not clear if they are proposing alternate horizon proejct or within horizon 15:14
yoctozeptodiablo_rojo: I meant yours at gmann, does not matter :-)15:14
diablo_rojoyoctozepto, ahh got it15:14
diablo_rojodansmith, too true, thankfully 99cloud wants to help with that. 15:15
dansmithI mean the entire world of frameworks, a problem no one entity could hope to resolve :)15:15
diablo_rojodansmith, touche. 15:16
gmannmay be once we have the clear proposal then we can discuss further and with Horizon team15:16
ttxohai15:16
diablo_rojottx, hello :) 15:16
diablo_rojoTalking about Skyline15:16
ttxRe: skyline drive vs. horizon, my understanding from teh PTg discussion was that they are using different JS technologies15:17
ttxso it might be easier to get people to contribute to the newer ones15:17
diablo_rojoi.e. the horizon team would merge with the skyline team15:17
dansmithyeah, that's kinda the problem is that they're always a moving target15:18
dansmithso definitely understand that hope, as long as existing people can jump to the new thing and keep contributing15:18
ttxThat was the big question we asked back then... from the horizon team perspective would it be better as a Horizon next version, a separate thing, or just if it did not exist15:18
ttxThey said "separate thing will be easier, even if some would likely be contributing to both"15:18
dansmithpeople that maintain horizon installs now are also more likely familiar with "whatever horizon uses" and moving around a lot generates ops churn that won't be great or welcomed15:19
ttxlike it's probably the only way to ever update our dashboard to more modern webtech15:19
belmoreiradansmith +115:19
ttxBut I would expect both to coeexist15:19
gmannttx: but that will distribute the maintainers too so both might face maintainer issue what horizon is facing ?15:19
ttxgmann: if you assume JS developers know all the frameworks, yes. In reality, you find those using one OR the other15:20
ttxOne of the reasons Horizon is struggling is that it's not a super-new shiny JS thing15:20
diablo_rojogmann, you're not splitting up the skyline folks or the horizon folks really. 15:20
gmanndansmith: +1 usage is also good point15:21
ttxBut yeah, I don;t expect miracles either :)15:21
diablo_rojoYou would have horizon folks maybe still doing horizon things and also doing skyline things15:21
gmanndiablo_rojo: ? is it? horizon team is lacking in resources to do Horizon things15:21
gmannI do not think they are ready to do both?15:21
ttxIt's a bit of a leap of faith, but that's precisely the case we envisioned when we allowed "multiple projects doing the same thing"15:22
gmannlike, system scope support in horizon is one example15:22
diablo_rojoIt seems like they wouldn't be doing both though. Parts or all of horizon would be getting deprecated? 15:22
dansmithI really just think we shouldn't consider any "leaps of faith" and let another project come up alongside,15:22
dansmithuntil it is clearly the better choice15:22
gmannyeah so question is what is our plan for Horizon. if deprecation is what we want then that is other thing15:22
dansmiththat means make it deployable in devstack as an option, include them in messaging, encourage operators to try it, provide feedback, survey who wants to use that instead of horizon15:23
ttxdansmith: that would be a departure from the previous policy, but I can see why15:23
diablo_rojoAt this point, skyline still has a little work to do, like I said. I mostly wanted this conversation to see if there were any other things they need to do before we figure out the proposal + replacement/deprecation etc plan. 15:23
dansmithttx: departure from the previous policy of what? choosing one project per realm? I feel like we've waffled back and forth on that before anyway15:23
ttxprevious policy = let projects incubate within the community rather than outside of it)15:24
belmoreirahonestly I think OpenStack needs a "modern looking" UI. But a completely new project from nothing sounds a risky strategy... Almost every cloud uses Horizon15:24
dansmithoh15:24
dansmithbelmoreira: agree15:24
gmannbelmoreira: true, usage is more imp and whether ops want to migrate it or how easy that will be15:24
dansmithttx: tbh I don't really feel like the current policy is not to incubate projects inside, but I'm sure the letter of the policy law is different15:25
ttxbelmoreira: the path to a "modern looking UI" is through a new project, according to the Horizon folks15:25
yoctozeptothe path is right15:25
dansmithttx: things like cyborg are still trying to achieve deployable-and-usable status15:25
yoctozeptothe worry is about the maturity15:25
ttxanyway that probably warrants a thread... I need to jump to another meeting15:25
gmannyeah also horizon team feedback what they things and plan for horizon15:25
ttxdansmith: change in policy = accept projects once tey are mature, rather than when they are under development15:25
diablo_rojoYes, a ML thread is definitely in the future. 15:26
gmannlet's ask them to start the ML thread and we continue discussion there 15:26
diablo_rojoRepeating myself again-  skyline still has a little work to do, like I said. I mostly wanted this conversation to see if there were any other things they need to do before we figure out the proposal + replacement/deprecation etc plan.15:26
dansmithttx: okay I'm confused, but sounds like we understand something needs to happen here15:26
spotz_And I think a PTG session, either during our leadership one or another added. I think the projects need some input if they make their own UI15:26
ttxfor now they need to work on Zuul integration anyway :)15:26
jungleboyjAgreed.  Something has to be done.15:26
diablo_rojoI would think that it would make sense for them to attend Horizon things again yes. 15:27
gmanndansmith: ttx policy discussion is interesting. would you like to add it in PTG sessions?15:27
diablo_rojoBut I would also think we would like to make progress before then and not just put it on hold till then.15:27
gmannif so please add in PTG etehrpad15:27
gmannmoving on?15:28
dansmithgmann: I dunno, I'm much more of a "let's just figure out what needs to happen and do it" person than a "figure out the current/new policy" but whatever ya'll want15:28
gmanndiablo_rojo: yea ML thread can be good as next step15:28
diablo_rojodansmith, yes, trying to figure out what needs to happen for these Skyline people lol. 15:29
diablo_rojogmann, okay. 15:29
gmanndansmith: sure. please add topic. this is important 15:29
gmann#topic Gate health check (dansmith/yoctozepto)15:29
jungleboyjdansmith:  ++15:29
dansmiththere have been a number of infra-related (I assume) timeout type things I've hit lately15:29
dansmithhaven't had a lot of time to debug them for sure, but things like no logs uploaded, etc which I assume are just infra15:30
yoctozeptowe also had a breakage with grenade due to cinder hardening their validation15:30
clarkbthat can happen if networking crashes on the instance or if the instance crashes15:30
dansmiththere was also the grenade thing, and the nova lvm job thing, both of which are resolved15:30
yoctozeptobut this has been already fixed15:30
gmannyeah cinder API change failed grenade and blocked gate for sometime but it is solved now15:30
clarkbI'm not aware of general upload issues to clouds right now15:30
gmannclarkb: may be dansmith will ping you on infra about it, he said double booked so might be busy in another meeting ?15:33
gmannany other issue on gate we want to discuss?15:33
dansmithI don't have anything to ping about, just noticed the timeouts, but I expect they're normal things15:34
dansmithsometimes there's a rash of them it seems, maybe while providers are doing stuff, I dunno15:34
dansmithmoving on.. :)15:34
gmanndansmith: ok i thought you become away :)15:34
gmann#topic Xena Tracker15:34
gmann#link https://etherpad.opendev.org/p/tc-xena-tracker15:34
gmannany update on items form Xena tracker ?15:34
gmannfor PTL guide goal, it is almost done only murano project left or which I pinged PTL many times but no response. 15:35
gmann#link https://review.opendev.org/q/topic:%2522project-ptl-and-contrib-docs%2522+status:open15:35
gmannrequesting everyone please check your assigned item in etherpad 15:36
gmann#topic Wallaby testing runtime for centos8 vs centos8-stream15:36
gmannWallaby testing runtime is centos8 #link https://governance.openstack.org/tc/reference/runtimes/wallaby.html15:36
gmannbut from current issue it seems liek centos8 cannot be tested on wallaby as no wallab y repo present for centos815:37
gmanndiscussion in #link https://review.opendev.org/c/openstack/devstack/+/80303915:37
gmannnow question is whether 1. we should update the wallaby testing runtime from centos -> centos8-stream and test centos8-stream or 2. remove centos job from testing as there is no way to test it15:38
gmann?15:40
gmannIMO, option2. as I think updating the wallaby testing runtime (and migrating testing to cento8-stream) is little bit extra work now. 15:41
clarkbI would at least suggest not promising anything about centos 815:41
clarkbbecause it will stop updating in 5 months15:41
gmannclarkb: yeah we changed that in Xena so no more testing or cetnos8 in Xena onwards15:41
gmannbut here question is wallaby testing which was working fine but now stop working15:42
yoctozeptoI say just backport stream fixes down to victoria15:42
yoctozeptoand test only stream15:42
gmannit can happen with ubuntu also like Bionic stop working and cannot be fixed  we cannot change all our testing to focal for old stable 15:42
yoctozeptothis is more sensible reality-wise15:42
yoctozeptogmann: we don't support openstack stuff beyond ubuntu lifetime15:43
gmannyoctozepto: like Pike? not sure what version we support there but it is working15:43
gmannI think supporting centos8-stream from Xena onwards make sense and not in old stable15:44
clarkbgmann: I think thei mportant piece of info here is centos-8 is a dead end. The other platforms we run and test on aren't, but when they hit their end of 5 years of support we shut them off too15:44
fungion the bionic example, ussuri was the last release to officialy support ubuntu bionic according to the pti, and it's scheduled to transition to extended maintenance in a few months15:45
gmann "shut them off" you mean for current or for all old stable testing too?15:45
fungi(november looks like)15:45
yoctozeptoextended maintenance is not eol15:45
fungiextended maintenance also does not guarantee we keep running all the same tests/platforms15:46
fungiem is a "best effort" testing only15:46
yoctozeptoyup15:46
gmannand we stop testing it anything stop working in EM15:46
fungiwe no longer tag point releases in em, we no longer guarantee we'll even fix security vulnerabilities in em branches15:46
clarkbgmann: we shut off hte test platform. Anything use that test platform will stop testing there15:46
gmannclarkb: +1, yeah that is my understanding 15:47
yoctozeptocome on, I don't think it's that painful to just make sure devstack works on stream for stable branches and just communicate this at tc level that we lost centos-8 earlier than planned15:47
clarkbgmann: so you can either turn off centos-8 now or in 5 months15:47
clarkbyoctozepto: that depends entirely on whether or not stream updates function :) I know libvirt has updated in ways that caused problems already15:47
yoctozeptoclarkb: true that15:48
gmannyoctozepto: we can do that i am not saying stop testing centos8-stream there but in TC defined testing runtime which is min thing to do we should not update it to centos8-stream15:48
jungleboyj:-(15:48
gmannany project can test anything extra there is no limit on that15:48
yoctozeptojungleboyj: why the sad face?15:48
gmannyoctozepto: need to know every expression/feeling :)15:49
jungleboyjyoctozepto:  The fact that a libvirt update already broke things.15:49
yoctozeptogmann: so you suggest to not change anything and just tell people the platform disappeared and it's platform's fault?15:49
yoctozeptoworks for me :-)15:49
yoctozeptojungleboyj: well, there was a huge bump in version15:49
yoctozeptobut it was not that bad15:49
gmannyoctozepto: yeah and if you wan to add centos8-stream job in wallaby we are all ok15:49
yoctozeptogmann: I don't really "want", I just don't oppose it :-)15:50
gmannyoctozepto: ok :) ditto15:50
yoctozeptoperhaps some other tc-members opinions?15:50
spotz_The original plan for Xena was for RDO to be oon CS9, we're looking at optionns annd will discuss in our meeting next week15:50
yoctozeptobecause now only qa and infra spoke :-)15:50
yoctozeptospotz_: well, we are not yet thinking about CS9 at all tbh15:51
yoctozeptobut yeah, good to refresh this info15:51
mnaserisnt the only reason oepnstack is essentially tested on centos is pretty much tripleo?15:51
yoctozeptowe will likely need to adapt the version tested late in the cycle15:51
mnaserbecause reality is we only run jobs on focal15:51
yoctozeptomost jobs run on focal15:51
yoctozeptolike, most = 99.5%15:52
yoctozepto;-)15:52
gmannmnaser: yeah we have job in devstack and devstack based too but mostly tripleo15:52
gmannwe are running out of time..15:52
fungithe devstack job has been failing again for a while, last i looked15:52
fungiand is non-voting15:52
gmannfungi: yeah15:52
yoctozeptothe interesing question is do redhat people actually develop on centos or ubuntu? ;D15:52
mnaserand essentially centos 8 / cs8 is not that huge of a delta to be honest15:52
gmannso to summarize, 1. no update in wallaby testing runtime 2. devstack or any project can test centos8-stream in wallaby as extra thing. 3. no guarantee of testing in centos8 or centos8-stream in wallaby as per current situation in platofrm 15:53
TheJuliayoctozepto: depends on the project and focus15:53
yoctozeptoTheJulia: hello :-)15:53
gmanneveryone ok with the above plan?15:53
TheJuliaSo, what about the fact that centos8 is EOL15:53
gmannif not then we can continue discussion in next meeting..15:53
TheJuliaare we doing the right thing by saying the tested against runtime is an EOL thing?15:54
yoctozeptogmann: I can accept it15:54
fungialso rhel has "free" (gratis) licenses for developers, but that doesn't seem to extend to running in public clouds so not applicable for our ci/cd15:54
mnaseri mean15:54
mnaseri know our tested runtime list is supposed to be set hard-set-in-stone15:54
mnaserbut we can append a change to it and explain why we did it15:54
fungiabsolutely15:55
yoctozeptoeither approach works for me15:55
gmannmnaser: but appending it need  move testing to new version15:55
TheJuliaBut is that still the right thing for those who wish to consume our software?15:55
yoctozeptonot blocking, not endorsing15:55
mnaserTheJulia: well, when we promised c8, we didnt expect it to be eol'd15:55
gmannanyways I will keep it for next meeting, moving to next topi5 min left15:55
yoctozeptoTheJulia: people consume openstack on rhel and debian and neither is listed ;-) 15:55
TheJuliamnaser: true15:55
mnaserand so i guess since things changed... we'll just have to adopt..15:55
gmann#topic PTG Planning15:56
mnaserif anything, i'd say we'd do disservice if we didn't adapt and left an eol os there :) -- but onto the next!15:56
gmann#link https://etherpad.opendev.org/p/tc-yoga-ptg15:56
TheJuliamnaser: I concur, fwiw15:56
gmannok, let's continue it in next meeting or after meeting15:57
gmann#link https://etherpad.opendev.org/p/tc-yoga-ptg15:57
diablo_rojo3 min left. 15:57
gmannplease add topic there^^15:57
gmann#topic Board informal Brainstorming sessions about "community health and resource management"(gmann)15:57
gmannno update on this15:57
gmann#topic OpenStack newsletter items15:57
gmann#link https://etherpad.opendev.org/p/newsletter-openstack-news15:58
diablo_rojoJust a reminder that the ML etherpad is ongoing really. 15:58
gmannplease add topic there15:58
diablo_rojoDrop things as you find them. 15:58
diablo_rojoDon't need to wait for me to put the call out. 15:58
gmanndiablo_rojo: when is deadline to finalize ?15:58
fungiin about a month15:58
gmannis it for Aug or next month?15:59
fungi(they're published ~monthly, and today was the most recent one)15:59
gmannohk, we missed that may be15:59
gmannanyways let's add item there as it comes15:59
fungiusually the deadline is a day or two before it's distributed15:59
gmannfungi: yeah that is not clear to me :)15:59
gmannanyways I will check and make it more advance next time16:00
diablo_rojogmann, it varies month to month16:00
gmann#topic Open Reviews16:00
diablo_rojoI added a few of the things that were mentioned yesterday16:00
diablo_rojoAnd will keep the rest for next month. 16:00
gmannok16:00
gmann#link https://review.opendev.org/q/projects:openstack/governance+is:open16:00
gmannonly one review which is waiting on projects side16:00
gmann\o/16:00
gmannthanks all for joining today16:01
gmann#endmeeting16:01
opendevmeetMeeting ended Thu Aug  5 16:01:11 2021 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)16:01
opendevmeetMinutes:        https://meetings.opendev.org/meetings/tc/2021/tc.2021-08-05-15.00.html16:01
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/tc/2021/tc.2021-08-05-15.00.txt16:01
opendevmeetLog:            https://meetings.opendev.org/meetings/tc/2021/tc.2021-08-05-15.00.log.html16:01
spotz_Thanks gmann, everyone!16:01
diablo_rojoThank you gmann!16:01
jungleboyj\o/16:01
jungleboyjThank you!16:01
yoctozeptothanks gmann16:01
*** akekane_ is now known as abhishekk16:01
fungito clarify the newsletter situation, the editorial team usually lets us know at least a few days, sometimes as much as a week, before they plan to publish a new edition. usually we would try to give bodies like the tc as much advance warning as possible to submit topics, but if things can be added to that etherpad on a continuous basis then at least we have something we can publish in the16:03
fungiopenstack subsection16:03
fungithe cadence is roughly monthly, but varies based on a number of external factors16:03
fungiso there's no rigid schedule to plan for, unfortunately16:03
gmannfungi: yeah, knowing the final deadline is good as we can finalize the added things.16:04
diablo_rojoYes. It gets pushed forward and pulled back depending on other Open Infra news + projects releases etc16:04
diablo_rojogmann, yes and normally I am more on top of that. 16:04
gmannyeah, thanks16:07
*** rpittau is now known as rpittau|afk16:41
-opendevstatus- NOTICE: The Gerrit service on review.opendev.org is going down for a quick restart to adjust its database connection configuration, and should return to service momentarily20:04

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!