Monday, 2018-07-30

*** diablo_rojo has quit IRC02:09
*** ricolin has joined #openstack-tc02:43
*** e0ne has joined #openstack-tc05:28
*** jaosorior has joined #openstack-tc05:51
*** e0ne has quit IRC06:00
*** melwitt has quit IRC06:31
*** melwitt has joined #openstack-tc06:32
*** melwitt is now known as Guest971406:32
*** eumel8 has joined #openstack-tc06:53
*** jpich has joined #openstack-tc07:21
*** PapaOurs is now known as bauzas07:41
*** e0ne has joined #openstack-tc07:50
*** cdent has joined #openstack-tc07:55
cdentAt what point do we become nervous about PTL nominations?08:12
*** dtantsur|afk is now known as dtantsur09:35
*** ricolin has quit IRC10:01
*** jaosorior has quit IRC10:26
*** rosmaita has joined #openstack-tc11:08
*** diablo_rojo has joined #openstack-tc11:09
*** diablo_rojo has quit IRC11:16
*** diablo_rojo has joined #openstack-tc11:28
*** jaosorior has joined #openstack-tc11:32
scasi'm not tc, and i'm there11:39
scasto my uncaffeinated nose, it smells a bit like the shrinking effect that has been mused about11:40
scasit aligns with the hype cycle fairly evenly, though this is the trough of disillusionment11:42
cdentscas: I think that's certainly an aspect of it11:42
scasi'm not saying it's the whole thing, but the major catalyst feels like a bit of corporate burnout. their experiments with openstack aren't as aws-killer-like as initially believed, and there's aws sitting there being itself11:46
cdentI think there's some localized burnout too, individuals who have been PTL or near-to-PTL needing a break11:47
scasquite so11:47
scasdespite my soloness, i don't feel the symptoms yet, but it happens to everyone with enough time11:48
mugsiecdent: I think you may be hitting the nail on the  head for PTL burn out, some of us have been PTLs for a very long time11:54
* cdent nods11:54
cdentI reckon burnout (of all sorts) is a bigger deal in troughs.11:56
mugsieyeah, I think so. "if I could just get $THING done, people might start getting excited again, but E_NO_TIME_OR_PEOPLE to do $THING" is a real downer11:59
scasstein marks my fifth cycle, and ^ marks some of my sentiment11:59
cdentmugsie: exactly. and that's a trap.12:00
mugsiecdent: it is. This time it is a trap I am trying not to fall into12:00
scasi temper the treadmill with just doing what i can to keep things going. if i don't get all the things, okay, that's for tomorrow or sometime next week12:00
mugsiescas: that is all we can do12:01
scashaving a succinct answer to 'what is openstack' -- with or without the question mark -- may just help to stem the bleeding. the blagosphere has been super unkind to OpenStack: The Project since certain big names distanced themselves. looking at a hype graph, the pattern still aligns, but the statement or question still has a subjective or non-answer12:05
* ttx is back from vacation, happy to report I did not check on IRC at all12:11
* cdent gives ttx a cookie12:12
cdentwell done12:12
ttxre: Trove and public cloud -- their model is not as incompatible as you'd think. OVH is seriously considering adopting Trove (with a few caveats to work around security issues) once they are done updating their remaining regions to Newton+12:13
ttx[ttx back: gone 190:55:00] says my weechat12:15
scaspart of the problem is that most users' entry to OpenStack: The Project is by way of deploy projects, and right now openstack.org says maybe 3 paths exist, which mostly involve burning down the incumbent solution12:32
scasthe homelab crowd actively shies away from openstack's complexity, which can be a great source of 'labor of love' type maintainers12:34
scasnot that things should cater to any one individual or group, but provide clear-cut paths. it's an idealistic thing to say because reality says it'll take me, in a deploy project, way longer to make good on those aspirations12:35
scaschef has a good story surrounding the allinone monolith, which is a fine entry to the ecosystem. the larger deploy projects are getting there with allinone/quickstart type scenarios emerging12:36
scasthe same primitives exist for getting to a full production deployment with some assembly required12:37
scasi know these things, and now folks here do, but 'the world' has no idea i even exist12:39
scasthe point about deploy projects being products still rings in my head in this particular conundrum12:40
*** jaypipes has quit IRC12:43
scasthat we all work in unstructured, coordinated stovepipes is likely to be the crux of the problem12:43
cdentthat is a recurring theme12:43
*** jaypipes has joined #openstack-tc12:43
mugsiettx: I think the "few caveats" is what we were talking about. It has gotten better recently, but in the ye olden days it was a real issue12:43
*** mriedem has joined #openstack-tc12:49
*** mugsie has quit IRC12:49
*** mugsie has joined #openstack-tc12:49
*** mugsie has quit IRC12:49
*** mugsie has joined #openstack-tc12:49
scasi'd love to start expanding in terms of services supported again. if one looks at the chef cookbook list, some half a dozen are unmaintained despite deployments existing in the broader ecosystem. as upstream, nobody was providing feedback or seemingly showing interest in the future maintenance12:51
scasthat's a hard one to handle when the number of services is "as much as can be supported". the point about it being a product ringing especially loud12:51
scasbut if deploy projects are purpose-built for openstack, and they're not 'openstack', what are they?12:52
scasmashing the deploy projects into a sig might be too little, too late12:53
*** edmondsw has joined #openstack-tc12:56
scason the other hand, it could be signaling of a refinement of what the 'big tent' notions muddied12:59
scasthere's no clear-cut answer to 'what is openstack' because i can make the same claim as keystone and nova12:59
scasarguably, my deploy project is less significant in terms of success criticality, despite being pretty useful to those who use it13:00
scasit is, however, one of the paths to 'openstack'13:01
dimso/13:02
scasthe possessive terms aren't meant to signify anything but my position13:02
scasit's still 'by openstack, for openstack', regardless of me keeping the rope tied to the wheel13:02
*** edmondsw has quit IRC13:06
*** jroll has quit IRC13:10
*** jroll has joined #openstack-tc13:10
openstackgerritMohammed Naser proposed openstack/governance master: Add vice-chair to check-review-status  https://review.openstack.org/58706213:41
openstackgerritMohammed Naser proposed openstack/governance master: Fix crash with None value when no topic is set  https://review.openstack.org/58706313:41
*** lbragstad has joined #openstack-tc13:46
smcginniscdent: I was getting nervous last PTL election, then there seemed to be a rush near the end to fill all the open roles.13:48
pabelangerI see 6 that still need to be validated / approved but still missing a bit of nominations13:54
smcginnisI wonder if we should think about some targeted recruiting. We still have just over a day left, and it's a Monday, so maybe folks will get on that today.13:57
openstackgerritMohammed Naser proposed openstack/governance master: Add vice-chair to check-review-status  https://review.openstack.org/58706214:07
openstackgerritMohammed Naser proposed openstack/governance master: Fix crash with None value when no topic is set  https://review.openstack.org/58706314:07
* dtroyer "getting on that today"14:13
dtroyerI hope I'm not alone in procrastinating14:13
*** hongbin has joined #openstack-tc14:14
mnaseri wonder if we're better off getting rid of the PTL role itself14:20
cdentmnaser: for at least some project they provide a very useful orchestration role. what would replace that?14:21
cdent(I agree that the PTL as currently defined, may  be too much)14:21
ttxExternally-speaking PTL is just a way of saying "contact point". Ideally PTL would delegate all work, esp. release work14:23
mnaserright, but that can be split amongst cores14:24
ttxThe trick is that people see it as a glorified position, so if they are not "the elected one" they shy away from doing any coordination work14:24
ttxmnaser: I'd say it should be split. Nothing prevents that14:25
zanebttx: yeah, I was about to say that14:25
zanebfew people are excited about signing up for grunt work while somebody else gets all the attention14:25
ttxFrom a governance perspective, we need two things: at least one name to contact for anything related to that team, and a way to make final calls if needed14:25
ttxWe collapsed both needs into a single solution14:26
jrollFWIW, in the few projects I interact with, the coordination work is shared between cores. e.g. dmitry helps a ton with ironic things, matt and dan help a ton with nova things14:26
ttxjroll: yes, but in smaller teams there is definitely a "you signed up for it" mentality14:26
smcginnisI could see having one "contact point" for a set of low activity projects too.14:27
jrollsure, just throwing out a data point :)14:27
ttxThe PTL position was also a convenient way to proxy more "local" governance structures. you can organize internally however you want, externally we just need a name14:28
mnasera liaison vs a ptl?14:28
ttxI see it as an API we can use to interact with the project team. Internally the roles can be filled by different people, but externally we don't need to know that, as long as we know of one default person14:30
ttxalthough in some cases we codified liaisons14:30
ttxto avoid bothering the PTL all the time14:30
zanebrename it 'community contact person'?14:31
ttxWell, they serve two roles. One is ambassador/contact person14:31
ttxThe other is "Bucket stops here"14:31
ttxwhich is needed so that you don't need it14:31
*** mriedem has quit IRC14:31
ttxWe conflated the two needs into a single role14:32
ttxso if you renamed it 'community contact person', that person would still have final say in case it's ever needed14:32
smcginnisBenevolent dictator for stein.14:33
ttxwhich probably only happened a handful of times since the creation of OpenStack14:33
*** dklyle has joined #openstack-tc14:34
ttxbut contests would happen a lot more if we did not have that safety valve14:34
ttxas demonstrated in other communities14:34
mnaseri feel like a lot of people might look at PTL as being this extra management-like position14:35
ttxMaybe we should force the release coordination to be done by another human14:35
cdentplenty of people want them to be a PM14:35
ttxonce you have the PTL elected, just use them as safety valve / contact point and let them designate someone to do coordination14:36
ttxbut then we are back to the trick I mentioned earlier14:36
mugsiecdent: a lot of the role can feel like PM'ing14:37
jrollis there a specific problem we're trying to solve by the changes we're discussing here? lack of candidates or something else?14:37
mugsiebut, we need people to PM, and PM'ing a project by committee is tough14:37
jrollside note: zun moved to a single +2 for approvals to speed up the review process. good on them, interested to see how this goes14:37
*** ChanServ changes topic to "OpenStack Technical Committee office hours: Tuesdays at 09:00 UTC, Wednesdays at 01:00 UTC, and Thursdays at 15:00 UTC | https://governance.openstack.org/tc/ | channel logs http://eavesdrop.openstack.org/irclogs/%23openstack-tc/"14:37
*** dtantsur is now known as dtantsur|brb14:38
mugsieand, for some people having the PTL role is the only reason they can spend time on a project - if we got rid of that role, I think we would see people drop off14:38
dtroyermugsie: would codifying the two roles ttx mentions as separate positions be a benefit in that regard?14:39
dtroyerin the case of empoyers I know first-hand it might be14:39
ttxjroll: yeah, let's wait to see if we actually have a problem, before jumping to solutions14:39
pabelangermugsie: reason, to an employeer?14:39
jrollttx: too late? :)14:39
*** mriedem has joined #openstack-tc14:39
mugsieI think in smaller projects it would end up being the same person anyway - but it might allow some breathing room14:40
mugsiepabelanger: yeap14:40
dtroyerin my case it might, or it might also be a way to let an additional person get blessing to spend (more) time on a project because of the recognied position14:41
smcginnismugsie: "if we14:41
smcginnis                got rid of that role, I think we would see people drop off" - I agree14:41
mugsieI think that possibly the "buck stops here" person would be allowed time, but I worry the community contact person would be too abstract to be able to justify time with some employeers14:42
mugsieand unfortunately the "OpenStack infuencer" job market* has dried up a bit from the good old days, so we don't have as much negotiation power as we once had14:43
smcginnismugsie: Yeah, I agree.14:43
mugsie* That is an  awful phrase, but I can't think of a better way to put it14:44
smcginnisHah, it pretty much sums it up though.14:44
TheJuliaReading the discussion, I can't help but feel like there is a desire to see a problem where people may just be taking their time to reflect and give the opportunity for others to step up. I concur it is too early to know if it is a problem or not. +1 to mugie's statements re PTL basically being a lot of PM'ing.14:46
TheJuliaIf there is a last minute surge, it would be good to poll those to try and understand why.14:47
cdentI did ask "At what point..." not "Fire fire, the sky is falling!" :)14:47
dtroyerI didn' think we are at the point where action is necessary but we should stay on top of this sort of trend (if it is one) so we do detect it early14:50
smcginniscdent: Jeez man, stop panicking all the time.14:51
cdentfire fire14:51
mugsieTheJulia: when people ask what I do as PTL, I generally have something about herding cats, who actually have different managers, and as such listen about 30% of the time :)14:53
TheJuliammmm FIRE14:55
mugsieI would say it is time to worry for some projects - e.g. Oslo / Docs / Octavia / Barbican / Trove are all missing candiates right now. They are all important projects.14:56
*** diablo_rojo has quit IRC15:29
*** dtantsur|brb is now known as dtantsur15:37
*** dtruong_ has quit IRC15:39
*** dtruong has joined #openstack-tc15:47
*** jpich has quit IRC15:52
scasi reluctantly put my name on the hat five times, each time the argument to myself is that "if i don't do it, nobody else will"16:01
scaswhile chef may not be as crucial as, say, barbican, i consider it one of the many paths to openstack16:02
scaswithout paths, it becomes a monoculture. the balance is kept by someone doing what i do, after the band has all packed up and the house lights came on16:03
scasnot to mention, if i let it wither completely, there are some major decisions that have to be made in the corporate ecosystem. the average chef deployment, in terms of managing openstack, is not small16:04
scasone could argue that a cosmic shift needs to happen, but that's more of a product roadmap for something that has paying users16:05
cdentI sometimes, but not always, wonder if letting things wither is the only salient strategy for lack of corporate contribution16:05
scaspossible. more than once i've been tempted to let nature do whatever it wants to16:06
scasi don't because i know the knock-on effect16:06
scasthe sunk cost fallacy of the existing first gen deployments is strong. unless you do a drastic shift, decisions from five years ago are still in play16:07
scasthe adoption is slower than the development, usually by a couple of named releases16:08
scasin the greater blagosphere, i've seen murmurings of the mental cost of deploying openstack to be too great to undertake, even with the second and third gen deployers16:09
scasdocker/k8s evolve so quickly that something documented a week ago may or may not work. on the other hand, i can slap a monolithic openstack instance on my laptop in about the time it takes to cook a pizza and burn your mouth on the first bite16:10
scasbut that flexibility may or may not be super apparent. my gut says it's not.16:11
mugsiedeployment tools are great at install. what people have issues with is long term maintence. (k8s is having the same issues, but have less moving parts that need to be deployed on the metal so it is not a pronounced)16:14
*** Guest9714 is now known as melwitt16:15
scasyeah. the long-term maintenance is a pain unless you've a team behind the effort. usually three.16:15
mugsiee.g. doing a deploy is easy(ish) but I need to convert 5 nodes to CPU pinned, wiht weird kvm emulator thread pinning, but the tool doesn't support it, so it is manually done. and this happens so much that the deployer cannot use the tool to update ever again16:16
scasquite. some will just work around the problem on their end, and never engage upstream16:17
mugsiesure, and I get that. e.g. our team can get requests like that with 30 days notice, and that is nowhere near enough time to do the upstream thing, and by the time we are done the next fire is burning its way down the mountain16:19
mugsie(we also work via a vendor, so trying to engage upstream can be a lot more problematic)16:20
scasyeah. that's also probably some of the point about deploy projects being products in foss clothing16:20
scasfor the overarching stuff, letting 'the community' figure things out is so far proving to be pretty sad times16:21
scasi'm trying not to be too critical. it's a balance, because it's about being inclusive without being too exclusive16:22
scasif the answer to 'what is openstack' means that automation/lifecycle focused projects are 'not-openstack', there needs to be some sort of cushion16:25
scasin retrospect, the big tent concept likely should have been a more tempered sig-based grouping, but it's pretty far down the road to veer off in a different direction16:27
scascentos does a similar concept with sigs16:28
scasthe success can be either grand, or not at all. i joined a centos sig some time ago, but the mission got out there in a mixed way which drove away all the contributors16:29
scasnobody wanted to repackage eol software, when the upstreams themselves didn't even do it16:29
*** e0ne has quit IRC16:31
scasthat, to me, harkens to the benevolent dictator practice in open source -- not 'for life', but 'for now'16:31
scasthe esr definition is implied here, natch16:34
*** diablo_rojo has joined #openstack-tc16:47
mnasertc-members: if i can take a minute of your time, https://review.openstack.org/#/c/584047/ is an amendment to the charter which requires 9 approvers.  if anyone out of those who havent checked it has a few minutes to review it, as we'll need 4 days after we have our last minimum approval :)17:38
pabelangermnaser: thanks, voted17:39
*** zbitter has joined #openstack-tc17:40
scasdare i say, rocky is in the final stages, and i'm on the current development focus17:41
scasfor my effort, that's something noteworthy17:41
*** zaneb has quit IRC17:42
*** e0ne has joined #openstack-tc17:43
*** cdent has quit IRC18:09
*** cdent has joined #openstack-tc18:09
*** zbitter is now known as zaneb18:26
*** purplerbot has quit IRC18:27
*** purplerbot has joined #openstack-tc18:28
*** dtantsur is now known as dtantsur|afk18:31
*** diablo_rojo has quit IRC19:07
*** cdent has quit IRC19:14
*** cdent has joined #openstack-tc19:57
*** cdent has quit IRC20:13
openstackgerritMerged openstack/governance master: Add vice-chair to check-review-status  https://review.openstack.org/58706220:22
openstackgerritMerged openstack/governance master: Fix crash with None value when no topic is set  https://review.openstack.org/58706320:22
*** masayukig has quit IRC21:13
*** masayukig has joined #openstack-tc21:15
*** e0ne has quit IRC22:39
*** hongbin has quit IRC22:52
fungii haven't directly compared to previous ptl elections, but it doesn't seem like we're significantly outside the usual envelope of last-minute nominations yet23:19
fungiquite a few incumbents delay nominating to encourage others in the community to do so23:20

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