Tuesday, 2017-11-14

*** hongbin has quit IRC00:23
*** sdague has quit IRC00:55
*** kumarmn has joined #openstack-tc01:37
*** gcb has joined #openstack-tc01:42
*** mriedem has quit IRC01:54
*** kumarmn has quit IRC02:09
*** kumarmn has joined #openstack-tc02:09
*** kumarmn has quit IRC02:30
*** kumarmn has joined #openstack-tc02:30
*** kumarmn has quit IRC02:35
*** rosmaita has quit IRC04:15
*** kumarmn has joined #openstack-tc04:38
*** kumarmn has quit IRC04:43
*** mtreinish has quit IRC06:55
*** mtreinish has joined #openstack-tc06:55
*** kumarmn has joined #openstack-tc07:31
*** kumarmn has quit IRC07:36
openstackgerritRabi Mishra proposed openstack/governance master: Add heat-tempest-plugin repo  https://review.openstack.org/51957808:17
*** jpich has joined #openstack-tc08:57
* johnthetubaguy waves hello to folks who have returned from syd (and those who are still here from before)08:58
* gcb waves09:10
*** zaneb has quit IRC09:54
*** sdague has joined #openstack-tc10:32
*** dtantsur|afk is now known as dtantsur11:20
*** kumarmn has joined #openstack-tc11:31
*** kumarmn has quit IRC11:36
*** rosmaita has joined #openstack-tc13:01
*** kumarmn has joined #openstack-tc13:32
*** kumarmn has quit IRC13:35
*** mriedem has joined #openstack-tc13:59
*** kumarmn has joined #openstack-tc14:33
* smcginnis is back home and catching up14:57
*** zaneb has joined #openstack-tc14:58
*** hongbin has joined #openstack-tc15:20
*** zaneb has quit IRC15:21
*** chandankumar is now known as chkumar|upgrade15:37
*** zaneb has joined #openstack-tc15:43
*** dtantsur is now known as dtantsur|brb15:46
openstackgerritEmilien Macchi proposed openstack/governance master: Remove stable:follows-policy from Kolla  https://review.openstack.org/51968515:51
*** chkumar|upgrade is now known as chandankumar16:10
*** zaneb has quit IRC16:25
*** zaneb has joined #openstack-tc16:25
dimsfolks, please peek at https://etherpad.openstack.org/p/LTS-proposal17:15
pabelangerwill do17:16
dimsthanks pabelanger17:16
pabelangeralso, I've heard LTS might not be the right name a few times.  Specifically the 'support' wording in LTS17:17
TheJuliadims: already open on a tab :)17:18
johnthetubaguyLTS = Long Term Stable-branch ?17:21
dtroyeragreed on the 'maybe don't use the term "LTS" here', it's heavily influenced in prople's minds by the distro's usage17:21
*** jpich has quit IRC17:21
dtroyerZR = zombie release, the occasional release that is kept alive 'forever', but without new releases, just 'alive'17:22
dtroyers/alive/not dead/17:22
SamYaplewell in most things 'LTS' just means youll get security and critical bug fixes for a long time. that counts as 'support'17:27
SamYaplei like the way ceph does it. with an LTS once a year, and a new 'stable' every 6 months17:28
SamYaplewith the intermediate 'stable' release not having long support at all (it basically EOLs as soon as the new LTS lands)17:28
pabelangerdtroyer: yah, zombie release is more the feedback I hear in hallways17:29
*** dtantsur|brb is now known as dtantsur17:45
*** kumarmn_ has joined #openstack-tc19:33
*** kumarmn has quit IRC19:36
*** dtantsur is now known as dtantsur|afk19:45
dtroyerSamYaple: it feels like that is our best bet for a small step forward, to get explicit about extending only specific releases, every 2nd or 3rd one or so20:04
SamYaplei like year long releases, all our cycles start to match up20:05
SamYaple*and* it lets us kind of target a general "bug fixy" release20:05
SamYaplemore experimental stuff can goin the inbetween releases20:05
SamYaplegive it some time to beused and stable-up20:05
dtroyerand all those clamoring for more rapid innovation will… ???   Will projects like Nova that are already fairly slow to change slow down more or will they feel like they have a chance to catch up?20:07
SamYaplewell forthe "rapid innovation" nothing will change right? well be in the same situation, 2 releases a year20:08
dtroyerI think we have a perception that "more shorter release cycles" is that it helps drive things to completion, even multi-cycle things.  is that still the case?20:08
pabelangerdtroyer: good question20:08
SamYaplei suspect the _perception_ might shift a bit, but the reality would be the same20:09
smcginnisThose that want things faster can deploy from master. (rhyme not intended)20:10
SamYapleor latest stable (which may ormay not be an LTS)20:10
smcginnisBut I do think the shorter cycles do help drive things.20:10
smcginnisBut at the same time, also causes us to compromise on some things to try to not make something a multi-release process.20:10
dtroyersmcginnis: careful, that'll become a catch phrase20:11
SamYaplei would like to avoid "no feature merges before LTS release" type attitude for sure. but weve talked about bugfix/stabilization cycles before. just the name LTS kind of gets people in that stabilization mindset without having to explicitly define one20:12
dtroyerit does feel to me that OpenStack is past the stage where rapid releases are useful and a one-year cadence of primary releases that are directly upgradeable would be welcome, with intermediate secondary releases in between to keep the bug fix flow going.20:13
smcginnisSamYaple: That certainly is a risk.20:13
SamYapledtroyer: master with words, you are20:14
smcginnisdtroyer: I was thinking that earlier. At early stages, more rapid releases are probably more necessary just by the nature of a newly developed and rapidly changing project. But once things get to a point where they are mostly "stable", it is probably less of an issue.20:14
dtroyerSamYaple: it took me since that last session to put that together :)20:14
SamYaplei would suggest looking at the ceph release cycle and community perception. it strongly matches what weare discussing20:15
SamYapleplus ceph already sync'd its releases to openstack. so now wewould be syncing LTS releases :)20:15
dtroyerremember the odd/even minor version thing that the kernel did for a while?  is it similar to that? (modulo the timing parts)20:16
SamYaplehmm probably best explained by this chart20:16
SamYaplethough we wont need to neccesarily follow the deprecation stuff they do20:17
SamYaplei like our deprecation stuff mostly20:17
SamYapleluminous, jewel, hammer, firefly were the LTS releases inthat chart20:17
dtroyerok, at a quick look the idea looks very similar, if not identical20:18
SamYapleexcept the deprecation stuff. thats an important thingto note20:18
SamYaplethey EOL the stable-but-not-LTS brnach as soon as the newLTS is out20:19
dtroyerto put it into our terms, let's say the spring release (Queens now) is the primary release with a longer shelf life and the fall release (Rocky is next) is shorter.  yes20:19
SamYaplewould "shorter" be a 6 month deprecation vs the 12 month we have now?20:20
dtroyerthe goal that I would expect as an operator then is to be able to upgrade spring to spring (Q -> S),  or spring to fall (Q -> R) and fall to next spring (R -> S)20:20
dtroyerthat cuts the number of releases we need to upgrade test between20:20
SamYaplehmmm. thats not bad, but that wouldjust makeevery release LTS again20:21
dtroyerSamYaple: I think so, yes20:21
SamYapleignore me, i misread20:21
dtroyerno, the fall release EOLs earlier, like you said20:21
SamYapleyea. ok then yes this wouldbe almost identical20:22
SamYapleand frankly, that cycle has worked pretty well overthe last 4 years or so of ceph20:22
SamYaplei sometimes move to a non-LTS, sometimes not depending on my needs20:22
dtroyerthis puts no conditions on the contents of any given release (feature, bug-fix-focus, whatever), only on what we target for testing and expect for $UPSTREAM to manage over time20:22
SamYaplethere is also no "conditions" on the ceph releases, but as it happens humans get involved and the work turns much more bug-fixy towards the end before an LTS release20:23
SamYaplethats not abad thing, but something to consider20:23
dtroyerI wondered if that would be a natural outcome20:24
SamYaplegood chat, i gotta go fora bit. ill read scrollback to catch back up on any additional conversations regardingthis20:24
dtroyerthanks SamYaple, it helps me congeal the stuff still swirling around in my brain :)20:25
dimsSamYaple : please throw a line proposal in https://etherpad.openstack.org/p/LTS-proposal if you wish. let's see how many folks will go for it.20:35
openstackgerritMatthew Treinish proposed openstack/governance master: Update Python PTI for tests to be specific and explicit  https://review.openstack.org/51975120:43
*** kumarmn has joined #openstack-tc20:48
*** kumarmn_ has quit IRC20:51
openstackgerritMatthew Treinish proposed openstack/governance master: Update Python PTI for tests to be specific and explicit  https://review.openstack.org/51975122:16
SamYapledims: cool ill give it a look see what i can add22:35
dimsThanks SamYaple !22:36
*** kumarmn has quit IRC23:37

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