Thursday, 2018-07-26

*** ricolin has joined #openstack-tc00:45
*** ricolin has quit IRC00:46
*** cdent has joined #openstack-tc00:58
*** cdent has quit IRC01:08
*** edmondsw has joined #openstack-tc01:17
*** dtruong_ has quit IRC01:20
*** edmondsw has quit IRC01:21
*** ricolin has joined #openstack-tc02:31
*** openstack has joined #openstack-tc02:49
*** ChanServ sets mode: +o openstack02:49
*** edmondsw has joined #openstack-tc03:05
*** edmondsw has quit IRC03:09
*** mriedem_afk has quit IRC03:20
*** dtruong_ has joined #openstack-tc04:06
*** rosmaita has quit IRC04:12
*** spsurya_ has quit IRC04:48
*** spsurya_ has joined #openstack-tc04:49
*** edmondsw has joined #openstack-tc04:53
*** edmondsw has quit IRC04:58
*** e0ne has joined #openstack-tc05:06
*** spsurya_ is now known as spsurya05:07
*** dklyle has quit IRC05:52
*** dklyle has joined #openstack-tc05:53
*** e0ne has quit IRC06:00
*** edmondsw has joined #openstack-tc06:41
*** edmondsw has quit IRC06:47
*** jpich has joined #openstack-tc06:58
*** guvnah has quit IRC07:04
*** tosky has joined #openstack-tc07:38
*** dtantsur|afk is now known as dtantsur07:46
*** e0ne has joined #openstack-tc08:17
*** edmondsw has joined #openstack-tc08:30
*** edmondsw has quit IRC08:34
*** cdent has joined #openstack-tc08:54
*** dtruong_ has quit IRC09:02
*** cdent has quit IRC09:58
*** cdent has joined #openstack-tc09:58
*** dtruong_ has joined #openstack-tc10:10
*** edmondsw has joined #openstack-tc10:18
*** edmondsw has quit IRC10:23
*** edmondsw has joined #openstack-tc10:49
*** rosmaita has joined #openstack-tc11:26
*** openstackgerrit has quit IRC12:36
*** mriedem has joined #openstack-tc12:36
dimso/12:41
smcginnisHeya dims12:42
TheJuliagood morning13:05
* cdent waves13:23
dimssmcginnis : TheJulia : cdent : yo! so i got the ball rolling on a constellation - https://review.openstack.org/#/c/586212/13:24
*** mriedem is now known as mriedem_afk13:25
cdentyay dims !13:26
smcginnisAwesome, thanks dims!13:26
dimsyou all are welcome to edit it too!13:27
TheJuliadims: awesome13:42
*** hongbin has joined #openstack-tc13:52
*** mriedem_afk is now known as mriedem13:53
*** pabelanger has quit IRC14:07
*** pabelanger has joined #openstack-tc14:07
*** annabelleB has joined #openstack-tc14:10
mnaseri would like to have eyes on this interesting issue14:19
mnaserhttp://lists.openstack.org/pipermail/openstack-dev/2018-July/132475.html14:19
scasvery interesting.14:20
smcginnisOh right, I was going to respond there.14:20
*** annabelleB has quit IRC14:20
cdentthe crux is that they are not yet ATCs?14:20
smcginnisI was going to suggest he stays PTL primarily as a contact point and "mentor", but delegate as much as possible to this new team so it's not much of a time commitment for him.14:20
TheJulialooks like 3 people should qualify as atcs14:21
smcginnisThen over stein one of the new folks can learn the ropes and get actual patches merged and be ready to be the official PTL in T.14:21
smcginnisOr even part way through Stein if that's what they work out.14:21
TheJulia+1, anything to soften the transition14:21
TheJuliaBut I suspect Samsung folks would need almost a crash course as well to get started14:22
scasas someone who has assumed maintainership of a project developed by others, it's a trial by fire14:23
scas3 years in, i still forget to check launchpad on occasion14:23
TheJuliascas: we are all only human14:24
TheJuliaand it is a LOT of work :(14:24
TheJuliaI regularly forget to go look at storyboard and beat myself up for it14:25
TheJuliasorry, looks like 3 people have done a review from samsung on trove in this cycle, and one code commit :\14:25
scasin my head, i'd get them to commit a cosmetic change, which would tick the boxes. they still have to learn, but it gets them part of the way there14:26
scasi'm thinking back to my drive-by chinese friends14:27
smcginnisIf they have some additional test coverage they can contribute, that would be great.14:28
scas+114:29
scasone of my more aggressive changes in chef was spawned by a one-line test change14:30
TheJuliaI personally would like to see an existing contributor step up and be willing to take on the reins, but Samsung folks likely have some direction and roadmap in mind which is a huge bonus, where as the existing roadmap priorities and goals for the project need to be re-evaluated.14:33
TheJuliaLikely, we need to put out a call for contributions or perhaps call for a specific use case need to be fulfilled to help bring contributors to it somewhat more organically.14:34
scasif i put myself in that situation: another team wants to adopt chef openstack for their own plans -- i'd probably go with a phased transition where i'd make myself available for a time to get them up to speed14:34
scasit would be clear that they would have the stick, but i'd be around for consultative means14:35
smcginnisscas: That seems like a good approach to me.14:36
scasi'd also make no hard limit on how long that would take, but make it apparent that there would be an eventuality where they'd be going on their own velocity14:38
*** annabelleB has joined #openstack-tc14:38
scasbut, that's me14:38
*** hongbin has quit IRC14:46
smcginnisOK, finally had a chance to respond on there.14:49
*** tbarron has quit IRC15:01
smcginnisAbout that time I think.15:01
*** aclark has joined #openstack-tc15:03
cdenttc-members, let's listen to smcginnis15:03
smcginnis:)15:03
mugsiethis is troves 3rd leadership change in a year or so - which is worth bearing in mind15:03
smcginnisGood to see we have a few things on the PTG etherpad - https://etherpad.openstack.org/p/tc-stein-ptg15:03
smcginnismugsie: Can't remember if you were there last time in Denver, but it was about to go in maintenance mode when the team from IBM stepped in.15:04
mugsieyup - I was in the room for a lot of that15:04
dimsyep, i remember that15:05
smcginnisI honestly wouldn't be too surprised if it does eventually end up there, but it seems like companies keep coming out with new commercial offers using it.15:05
mugsieand then that IBM team handed off to the curent team, and now this team is handing over to samsung15:06
mugsieit is hard to open source stuff that you have done internally for a product - a lot of people using trove seem to have forked trove internally and now are trying to unfork.15:07
*** tbarron has joined #openstack-tc15:08
zanebsee also: StarlingX15:09
mugsieaye15:09
smcginnisThose internal forks always seem to sound good to the business forks until they realize a few releases later how much work and expense it adds.15:09
zanebyup15:10
smcginnis*business folks ;D15:10
zanebbut I also wonder if there were some factors specific to Trove driving that as well15:10
pabelangersmcginnis: +115:10
zanebthey had a security model that in my view was clearly incompatible with a public cloud deployment, and were very resistant to changing it15:11
mugsiea lot of the issue is the way trove was structured - it wasn't easy to do plugins, (see https://github.com/openstack/trove/blob/master/trove/common/cfg.py ) and there was not real easy to replicate ways of doing custom images for the VMs15:13
mugsiezaneb: yeah - the fact that all the public cloud providers had special extra special sauce bits of code they used for deployment was problematic15:14
mugsiethere was a proposal a while ago for a re-arch, but I am not sure it ever went into the public domain, and I am not sure there is the appetite anymore for it15:16
zanebif it's the one I am thinking of, it was more of a do-over, and I don't think there were ever resources available to sink into it15:17
mugsieno, there was a HPE internal one before the one that was floated in Denver15:17
mugsiebut question that is in my mind is, if we were looking at this as a project joining the project, would we accept them?15:19
mugsies/joining the project/joining the OpenStack project/15:20
smcginnisProbably not in its current state, but not sure if there's anything to do about that at this point.15:20
zanebyeah it would be a hard sell right now15:21
zanebit's a shame, because Trove was the first service outside of Nova/Keystone/Cinder/Neutron/Glance that every public cloud wanted to offer to their users (thinking of e.g. Deutsche Telekom)15:22
persiaPossibly worth reaching out to the Public Cloud team to get some input/guidance on the issue.  They are fairly active and representative.15:23
zanebworth noting that being unable to get something like https://review.openstack.org/438134 in Nova also contributed greatly to their inability to find a sane security model upstream15:23
zanebpersia: it's still in their top 5 priority list last time I checked15:23
zanebdemand is there15:24
zanebbelief that an upstream collaboration can deliver is, sadly, not15:25
mugsieor the ability, or willingness to put resources towards it15:27
persiaOr close communication between people willing to invest resources in it and the public cloud teams that are investing resources maintaining the existing limping solution15:28
mnasertrove is awesome, we'd love to have it, don't think we'd have the resources to re-architect it unfortunately15:34
mnaserhowever i think that they can learn a lot from octavia15:34
mnaserin terms of the service vm model15:34
*** annabelleB has quit IRC15:36
mugsiemnaser: yeah, octavia had the advantage of being devloped after trove, and see the issues.15:37
johnsommugsie Ha15:37
johnsommugsie No respect at all....15:37
mugsiejohnsom: I could have mentioned other projects that the teams worked on that also used VMs :P15:38
* mugsie mutters something about gearman15:38
mugsieThe spec that zaneb linked would have really helped them IMO15:39
johnsommugsie THAT is true. There were other examples of how not to grow up....15:39
mugsieI know kfox1111 was fighting that battle for a very long time, but we never managed to move the needle15:40
zanebAIUI the Octavia-style service VM model was in fact how people deployed Trove in practice, but was not the default upstream15:41
zanebthat wasn't the only issue - security of the message bus was also a big problem (now somewhat improved) - but it was a significant one, and both were related15:42
mugsieyeah, the octavia model really helped the message bus issue15:43
cdentwas that referenced spec not well received, or did it simply not have sufficient energy behind it?15:43
*** annabelleB has joined #openstack-tc15:44
clarkbpersia: I think that was actually a major part of the root of the struggle for a long time you had a team at rax and a team at HP and they sort of tried to smash code toegether every 6 months before the summit15:44
zanebcdent: I think by the time the spec was written, there was not much energy behind it15:44
mugsieyeah, there had been a lot of pushback before that, and I think it just ran out of steam15:45
clarkbyes there were external issues, but there was no cohesive driving force to address them because everyone had their own internal standups and just worked around them in their own local way15:45
persiaclarkb: Yes.  More coordinated comms, ideally integrating deployers/operators (some of whom can also contribute some changes) tends to be a more reliable model in this space.15:45
zanebcdent: but AIUI that was preceded by ~5 years of not even pursuing it because everyone had the impression that Nova had closed the door on it15:45
mugsieclarkb: that did improve over time, but by then there was not the drive to fix the big issues15:45
persiaIt is hard to do that from a product viewpoint, as the identities/input of the deployers/operators are considered the valuable thing ("customer").15:46
clarkbmugsie: there was effectively never a trove team to fight for their needs15:46
clarkbuntil amrith took over15:46
* clarkb sat next to the HP team for ~2 years and had to help with the just jam it in locally fun15:47
smcginniswin 2415:47
smcginnisDoh15:48
zanebsmcginnis: fail15:48
mugsieclarkb: yeap - I was part of the dublin HP PaaS team - I remeber it well :)15:48
smcginniszaneb: Yep! :)15:48
*** hongbin has joined #openstack-tc15:49
clarkbmaybe I'm just optomistic but I like to think that if there had been a more unified team to fight for the use cases that we would have seen the run a service in cloud problems solved much earlier. But instead that effort was put to forking the downstream cloud to handle it15:51
scaschef openstack is an interesting one. i somehow have the capacity to operate a private cloud and develop on a development project with a case of urban sprawl, but attracting people to maintain it has been difficult at best15:51
scasnot to derail15:51
zanebclarkb: imo it basically came down to a lot of the requirements not being available at the right time15:55
scason the hype cycle scale, chef probably goes in the first generation of tools given its vintage. yet there are still users that self-identify15:55
zanebthey needed e.g. service-locked VMs in Nova and Zaqar messaging instead of Rabbit in 201315:55
clarkbzaneb: yup, and they had them, just none of it was upstream15:56
zanebbut those things weren't available, there was demand right then, so everybody went off and did their own downstream hack15:56
clarkbzaneb: because it was the HP trove team and they had an HP cloud15:56
clarkbnot the OpenStack Trove team with an OpenStack cloud15:56
zanebafter which nobody had the incentive to solve the core problem upstream (in fact, they had the incentive to change as little as possible upstream)15:56
*** edmondsw has quit IRC15:57
*** edmondsw has joined #openstack-tc15:57
*** edmondsw has quit IRC15:59
scassome have the 'open' mindset, some have the 'meet the goal at any cost' mindset. unfortunately, they often come at odds in a space as where openstack lives16:01
*** e0ne has quit IRC16:02
scasin the democratic model, both can be allowed to flourish even under scrutiny16:02
scasopen source is no exception16:03
scasmy armchair observation is that things in this space operate akin to a large scale enterprise that just so happens to release code on the internet16:05
*** jpich has quit IRC16:09
scasthat's the bsd acting up again.16:10
*** ricolin has quit IRC16:27
*** dtantsur is now known as dtantsur|afk16:38
*** annabelleB has quit IRC17:15
*** e0ne has joined #openstack-tc17:20
*** annabelleB has joined #openstack-tc17:25
*** e0ne has quit IRC17:37
TheJuliaI kind of feel like the bridge is solving the common problem, and if that common problem or goal was there to motivate then that could help. But yeah, the meet the goal at any cost mindset does hamper the ability to even make sure that there is a common problem first.17:38
*** edmondsw has joined #openstack-tc17:48
*** DuncanT_ has left #openstack-tc17:58
*** dklyle has quit IRC18:01
*** e0ne has joined #openstack-tc18:22
*** aclark has quit IRC18:35
*** tosky has quit IRC18:57
*** ianychoi_ has joined #openstack-tc19:15
*** ianychoi has quit IRC19:18
*** annabelleB has quit IRC19:27
*** dklyle has joined #openstack-tc19:51
*** cdent has quit IRC19:59
*** e0ne has quit IRC20:13
*** diablo_rojo has joined #openstack-tc20:34
*** annabelleB has joined #openstack-tc20:43
*** e0ne has joined #openstack-tc20:59
*** e0ne has quit IRC21:00
*** edmondsw has quit IRC21:43
*** edmondsw has joined #openstack-tc21:45
*** edmondsw has quit IRC21:49
*** mriedem is now known as mriedem_afk22:14
*** pabelanger has quit IRC22:46
*** pabelanger has joined #openstack-tc22:46
*** mriedem_afk is now known as mriedem22:47
*** annabelleB has quit IRC22:55
*** hongbin has quit IRC23:00
*** lbragstad_ has joined #openstack-tc23:01
*** lbragstad has quit IRC23:01
*** knikolla has quit IRC23:01
*** david-lyle has joined #openstack-tc23:24
*** dklyle has quit IRC23:26
*** knikolla has joined #openstack-tc23:37
*** david-lyle has quit IRC23:51
*** dklyle has joined #openstack-tc23:52

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!