15:00:53 <mattmceuen> #startmeeting openstack-helm
15:00:54 <openstack> Meeting started Tue Mar 13 15:00:53 2018 UTC and is due to finish in 60 minutes.  The chair is mattmceuen. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:55 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:58 <openstack> The meeting name has been set to 'openstack_helm'
15:01:04 <mattmceuen> #topic rollcall
15:01:13 <jayahn> o/
15:01:23 <pcarver> hi
15:01:31 <seungkyua> hi
15:01:47 <jayahn> seungkyua: you made it. :)
15:01:51 <mattmceuen> https://etherpad.openstack.org/p/openstack-helm-ptg-rocky
15:01:57 <mattmceuen> That's not the agenda
15:02:06 <portdirect> o/
15:02:12 <seungkyua> yes. ha
15:02:22 <mattmceuen> https://etherpad.openstack.org/p/openstack-helm-meeting-2018-03-13
15:02:26 <mattmceuen> This is the agenda :)
15:02:29 <mattmceuen> thanks for joining guys!
15:02:34 <jayahn> since we have "review PTG highlights" on the agenda, the first link can also be agenda
15:02:36 <mattmceuen> Feel free to add anything to the agenda
15:02:39 <jayahn> lol
15:02:44 <mattmceuen> good save jayahn :D
15:03:03 <mattmceuen> I am just back from vacation and trying to catch up / get in the swing of things so please bear with me
15:03:21 <mattmceuen> #topic PTG Highlights
15:03:36 <mattmceuen> A big thanks to everyone who participated in the PTG
15:04:01 <jayahn> fighting with blizzard
15:04:02 <mattmceuen> We had some really valuable discussion, and was good to meet some more of the team face to face too
15:04:18 <mattmceuen> aaaaaaaaand blizzard.  that was fun! :)
15:04:22 <mattmceuen> https://etherpad.openstack.org/p/openstack-helm-ptg-rocky
15:04:34 <mattmceuen> ^ here are our notes from the sessions
15:04:50 <mattmceuen> There is much in there that we will be driving over the next 6 months and the rocky cycle
15:05:25 <mattmceuen> Please everyone take a look at your leasure and inwardly digest.  For folks who attended it will be a good refresher and for folks who did not a good learning
15:05:48 <mattmceuen> For now, I want to just touch on a small group of takeaways to make sure they're communicated well with the team
15:06:15 <mattmceuen> Please interrupt me if you'd like to briefly discuss; in depth discussion should probably become agenda items of their own.
15:06:37 <jayahn> sure thing
15:06:54 <mattmceuen> First:  we refined our general versioning approach.  The  general direction is documented in the etherpad.
15:07:15 <jayahn> squre sugar presentation was great :)
15:07:36 <mattmceuen> We hope to get our 1.x gated version support all the way to master (non-voting).  We will not branch OSH to 2.0 till we need to, likely for e.g. kubernetes upgrades (not OpenStack upgrades unless we find a real reason)
15:08:06 <mattmceuen> We came up with a list of criteria for OSH v.1.0.  I will author a spec for this and socialize it with y'all for feedback.
15:09:04 <mattmceuen> Destructive testing:  we would liek to add cookiemonster testing as a 3rd party gate, and would also like to explore powerfulseal
15:09:26 <mattmceuen> Docs: we need ops and troubleshooting guide that describes how to handle the sharper edges of OSH ops
15:09:55 <mattmceuen> Docs: we need sane defaults for the LMA stack, or a well-documented way to inject operator-specific things (I added an agenda item for this one)
15:10:10 <mattmceuen> Docs:  Helm-Toolkit and chart authoring should be documented
15:10:30 <mattmceuen> Docs:  all our executable documentation should be gated, following the pattern established already
15:10:51 <mattmceuen> (notice the trend?  we have a lot of documentation needs!  that's not the full list either - would love folks to volunteer for these)
15:11:23 <mattmceuen> Storyboard:  we want to follow the community goal for this assuming it's established, and will plan to follow the critical mass of folks migrating to storyboard rather than blazing the trail
15:12:04 <jayahn> one after-feedback: tagging with major openstack version caused some confusion with few people when I explain to team.  we might want to document that tagging policy well.
15:12:27 <mattmceuen> +1
15:12:36 <mattmceuen> Mailing List:  I will begin leveraging it to call out any big items from the team meetings that should be mailed out (and then potentially discussed).  Anyone else can feel free to do the same of course, not exclusively me :)
15:13:10 <jayahn> we will !
15:14:13 <mattmceuen> Mailing list:  will be leveraged in situations where the discussion participants prefer it, likewise with IRC.  There is no wrong way to start or continue a conversation, and different folks prefer different means - we will try to meet in the middle and all be flexible.
15:15:05 <jayahn> who is blue one on etherpad? Roy?  hi roy :)
15:15:52 <jayahn> sorry matt, i did not mean any interruption.
15:16:14 <mattmceuen> nope this one's just long :-D
15:16:55 <mattmceuen> Reviewers:  we need more reviewers and more core reviewers.  We need more substantive reviews given the volume of patchsets that are coming through, not just +1's w/o comments.    We will be watching for folks to prove their value as reviewers over time with an eager eye toward adding cores, and truing up our current core list to accurately reflect reality.
15:17:19 <srwilkers> o/
15:17:21 <jayahn> +1
15:17:38 <jayahn> you are great summarizer... matt.
15:18:37 <mattmceuen> Cores should probably also in general be announcing the direction for the project ,with input from the entire team of course.  W.r.t. the item above about announcing big outcomes from our meetings.  This wasn't something we discussed in the PTG but is something I'm connecting the dots on as I type.  We can discuss more later if desired.
15:19:08 <mattmceuen> YAML ordering:  Jay & team will author a spec for this, defining an ordering for top-level values keys, and having an alphabetical ordering under that
15:19:13 <mattmceuen> whew
15:19:40 <mattmceuen> That's not the full list, not even all the big hitters, but are some of the important bits to make sure we're all clear on.  Please see the etherpad for more.
15:19:49 <mattmceuen> Any discussion / comments?
15:20:19 <portdirect> we should mention the upgrade gate
15:20:35 <portdirect> as its kinda the big-hitter for any 1.0.x release
15:20:36 <mattmceuen> also please add in anything I missed you'd like to communicate -- just like ^ portdirect thanks :)
15:20:49 <srwilkers> sorry, reading scrollback as i lost track of time
15:21:02 <portdirect> story of your life srwilkers ?
15:21:14 <srwilkers> portdirect: so it seems
15:21:15 <mattmceuen> Yep we need to gate our upgrades as part of 1.0, and "supporting an  OS version" is synonymous with "a gated upgrade path"
15:21:25 <portdirect> ++
15:21:44 <portdirect> mattmceuen: "the summarizer" :D
15:21:44 <mattmceuen> anything else?
15:21:50 <mattmceuen> my fingers are so tired
15:22:04 <jayahn> you need AI to write down for you.
15:22:40 <mattmceuen> one more thing on our list of features needed :D
15:24:02 <mattmceuen> #topic YAML Values Ordering
15:24:17 <mattmceuen> Speak of the devil - there's already a spec in progress! https://review.openstack.org/#/c/552485/
15:24:27 <mattmceuen> (did I mention I'm still catching up :) )
15:24:31 <jayahn> This is just starting, pls put your opinion on the reviews, and let's get dicussion going
15:24:37 <mattmceuen> that's outstanding
15:24:55 <mattmceuen> yes let's please get eyes on this.
15:24:59 <jayahn> I was yelling this morning to put this spec on line. :)
15:25:22 <mattmceuen> haha just in time!
15:25:34 <portdirect> is powerds0111 here?
15:25:44 <portdirect> looks like a great start :)
15:25:54 <jayahn> hm.. good question.
15:26:42 <mattmceuen> that was also something we brought up at the PTG - potentially moving the time of the meeting to something that works better for the majority of team members
15:26:50 <jayahn> he was 7am armada weekly this morning (daylight saving mistake), probably sleeping.
15:27:09 <mattmceuen> ah daylight savings time, america's gift to the world
15:27:14 <mattmceuen> don't worry, we don't like it either
15:27:23 <mattmceuen> (speaking for all americans)
15:27:57 <mattmceuen> #topic [openstack-dev] [release] Relaxed rules for cycle-trailing projects
15:28:10 <srwilkers> can confirm: hate DST
15:28:14 <mattmceuen> jayahn I have not read this one yet, can you please give us the gist
15:29:00 <jayahn> we just found this with openstack-helm filter on mailing list.
15:29:14 <jayahn> I think Roy(?) seems knows better
15:30:10 <jayahn> long stroy short, it was after-two-week" requirements for any packaging/deployment project to be included in coordinated release
15:30:21 <jayahn> now they relaxed that to any time in 3 month.
15:30:44 <jayahn> according to this specific mail from Thierry Carrez
15:31:02 <jayahn> just FYI
15:31:09 <srwilkers> the tl;dr being that we could still be considered part of the Queens release if we release before the end of rc2 this cycle i think?
15:31:16 <jayahn> yeap
15:31:19 <jayahn> exactly
15:32:19 <mattmceuen> We are technically release-independent, this appears to be part of the confusion.
15:32:24 <srwilkers> rc2 might be unrelated, seems that's the cutoff for any deliverables that arent packaging
15:32:24 <portdirect> ^^
15:32:42 <portdirect> We are release independent
15:32:45 <mattmceuen> The idea is that we are starting with Newton, and working our way to Master as quickly as is feasible :)
15:32:56 <jayahn> yeap. that was our goal
15:33:43 <jayahn> we are release independent. but it seems like there is rule to be included in so called "coordinated openstack release"
15:34:01 <jayahn> anyhow, i think we can go to the next one
15:34:52 <mattmceuen> agree.  I think our discussion at the PTG helped crystallize in our own minds how the OSH versioning/releasing should work, and we'll just need to get that documented to clear up any confusion for others
15:35:10 <mattmceuen> #topic LMA stack defaults
15:35:20 <mattmceuen> srwilkers we missed you~
15:35:46 <mattmceuen> So it was brought up that we should have one or both of a couple things w.r.t. LMA in OSH
15:35:48 <srwilkers> i was there in spirit
15:35:53 <srwilkers> i was that snow ruining everyones day
15:36:00 <mattmceuen> there was plenty of irish spirit there, I think I saw you
15:36:48 <mattmceuen> First:  a meaningful out-of-box configuration for all the LMA components; usable dashboards that operators can use without knowing too much about LMA configuration
15:37:06 <mattmceuen> Second:  a documented way for operators to inject their configurations in
15:37:17 <mattmceuen> What are your thoughts srwilkers on these, current state & future state?
15:37:39 * mattmceuen has never configured the lma stack except by pushing some buttons
15:38:22 <srwilkers> First point -- yes, that'd be great.  I think we'll find more meaningful out of box configurations as more end users start kicking the tires and seeing what makes sense
15:38:39 <portdirect> I think this ties in with the rework of the gate - to follow the same format as other osh components (ie kill armansible)
15:38:46 <srwilkers> yep
15:39:09 <mattmceuen> +1 and also +1 for armansible
15:39:48 <jayahn> pls explain me armansible
15:39:55 <jayahn> ㅠ. ㅠ
15:40:19 <mattmceuen> armada + ansible :)  clever!
15:40:20 <pcarver> sorry to interrupt, but what is LMA?
15:40:32 <jayahn> logging, monitoring, alarm
15:40:33 <mattmceuen> no interruption, o/ pcarver!
15:41:24 <mattmceuen> anything else on this topic before we move onwards?
15:41:30 <srwilkers> current state for first point is that it's mostly followed upstream references for the services charted out, but future state needs to be targeting more real things.  examples that come to mind:  multiline configured in fluentbit to successfully capture python stack traces, more targeteted log filtering perhaps
15:42:00 <jayahn> +1 on that
15:42:44 <srwilkers> second:  there's a patchset in flight in osh that's a bit stale now that tries to outline some of the configuration entities in the charts, but would also benefit from some feedback with anyone with more ops experience than i have currently.  particularly whether it makes sense  to serve as a foundation to build on
15:43:29 <srwilkers> as i think writing an ops guide for the LMA stuff would benefit from collective input, as opposed to just my own
15:44:16 <mattmceuen> there's probably an 80/20 rule at play here -- 80% of the things that an OSH operator might want to tune could be tuned via 20% of the configuration, so if we document (e.g. gates) that 20% and point folks toward the projects' own documentation, that might be valuable?
15:44:42 <mattmceuen> (maybe even 90/10)
15:45:38 <srwilkers> yep. the patchset i have in flight makes a point of directing users to the projects' documentation for clarification on what the exposed configurations target
15:45:47 <mattmceuen> I think a place to start in any case is to gate with the tunables that we most expect to be tweaked, and build from there
15:45:52 <mattmceuen> perfect
15:45:52 <srwilkers> yep
15:46:15 <mattmceuen> cool beans
15:46:19 <srwilkers> thats it from me
15:46:22 <jayahn> i will put it in my todo list
15:46:38 <mattmceuen> awesome, thanks for volunteering jayahn!
15:46:39 <d|k> srwilkers: because i'm lazy, could you give a link to the patch you're talking about?
15:46:51 <jayahn> more yelling for me. :)
15:46:58 <jayahn> yeap, link would be great
15:47:08 <srwilkers> i too am lazy d|k
15:47:11 <srwilkers> but here you go
15:47:12 <srwilkers> https://review.openstack.org/#/c/528488/
15:47:21 <d|k> SLOTHS UNITE!
15:47:26 <d|k> thx!
15:48:11 <mattmceuen> HA
15:48:22 <mattmceuen> #topic can we modify kolla-build
15:48:31 <mattmceuen> jayahn can you elaborate on that one?
15:48:35 <jayahn> i might know the answer but
15:48:48 <jayahn> since we are getting kolla image not from kolla, but from our own
15:49:05 <jayahn> will it be possible to modify official kolla build?
15:49:20 <jayahn> for instance, adding lbaas stuff on newton kolla.
15:49:45 <jayahn> i remember portdirect said kolla did not want us to have our own version of kolla image.
15:50:15 <jayahn> it is just question to clarify what we can do here. :)
15:50:38 <mattmceuen> srwilkers portdirect what are your thoughts - anyone can surely propose a PS to kolla, so I guess it's a question of liklihood of acceptance :)
15:51:17 <jayahn> since newton is end of life, we cannot propose ps to kolla. it is all done deal.
15:51:38 <jayahn> otherwise, we will do ps on kolla side.
15:52:02 <jayahn> for sure
15:52:33 <mattmceuen> cool - if anyone has addn'l input on that it would be great, and we can continue in the chat room if so!
15:52:43 <mattmceuen> #topic SKT work planning
15:53:03 <jayahn> that was our way of sharing
15:53:12 <mattmceuen> jayahn shared some work his team at SKT as a group are planning to focus on this month -- https://etherpad.openstack.org/p/openstack-helm-meeting-2018-03-13
15:53:22 <jayahn> I will share at least once a month about our plan for the month.
15:53:36 <jayahn> anyone can give us feedback on priority or suggest  things to do
15:53:38 <mattmceuen> thank you for sharing jayahn, that's very valuable and is probably something that we need to do more of overall
15:53:45 <jayahn> or do question on the topic
15:54:14 <jayahn> regard this as a sort of backlog for us.
15:54:23 <jayahn> you can influence it. :)
15:54:34 <mattmceuen> team, please take a look at what jayahn has in that backlog and discuss feedback in irc/mailing list/etc
15:54:48 <mattmceuen> really appreciate you taking the initiative there
15:55:06 <mattmceuen> this dovetails into another topic :)
15:55:16 <mattmceuen> #topic specs for big changes
15:55:47 <jayahn> this is just our quick opinion.
15:55:52 <mattmceuen> Jay notes a couple of PS in the agenda that could have benefitted from discussion via a spec or otherwise first
15:55:56 <mattmceuen> I agree
15:56:30 <mattmceuen> Full disclosure:  some of us got a fire lit under our butts with one of those
15:56:50 <jayahn> understood. just say it then. :)
15:57:02 <mattmceuen> :)
15:57:04 <jayahn> we will support
15:57:18 <mattmceuen> we'll strive for good communication as well
15:57:29 <mattmceuen> #topic roundtable
15:57:32 <ttx> re: rc2 above -- actually that would be before rocky-2 milestone
15:57:38 <mattmceuen> sorry for rushing thru this, 2.5 mins left :)
15:57:48 <ttx> the new wording says "3 months after the coordinated release"
15:58:02 <ttx> so you'd have until May basically. The sooner the better though
15:58:07 <jayahn> okay. :)
15:58:13 <jayahn> thanks for clarification
15:58:16 <ttx> np!
15:58:37 <portdirect> ttx: we are going to remain release independent
15:58:44 <portdirect> how does this impact that?
15:58:49 <mattmceuen> Yep that is the part that's confusing me too
15:59:59 <mattmceuen> can we continue the convo in #openstack-helm?  I want to make sure we're on the same page but we're outta time!
16:00:03 <mattmceuen> #endmeeting