15:01:09 <mattmceuen> #startmeeting openstack-helm
15:01:10 <openstack> Meeting started Tue Apr  3 15:01:09 2018 UTC and is due to finish in 60 minutes.  The chair is mattmceuen. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:11 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:13 <mattmceuen> #topic Rollcall
15:01:14 <openstack> The meeting name has been set to 'openstack_helm'
15:01:17 <mattmceuen> o/ all!
15:01:23 <srwilkers> o/
15:01:23 <gmmaha> o/
15:01:40 <gagehugo> o/
15:02:03 <mattmceuen> Agenda - please add, we have a light agenda today so far: https://etherpad.openstack.org/p/openstack-helm-meeting-2018-04-03
15:02:44 <mattmceuen> And the one thing we carried over from last week may need to get pushed again because key folks are travelling today :)
15:04:23 <mattmceuen> #topic Multiline LMA functionality -- we are tabling till next week
15:04:33 <mattmceuen> #topic Team meeting time
15:05:07 <mattmceuen> It has been a very busy week folks (plus holiday) - I apologize but have not been able to get a survey type thing out to you all to determine a more optimal meeting time.  I will plan to do that this week.
15:05:22 <portdirect> o/
15:05:46 <mattmceuen> #topic Values-driven config files
15:06:17 <mattmceuen> portdirect has been doing some great work on this on a chart by chart basis, so I just wanted to make sure folks on the team were looped in on these changes
15:07:12 <mattmceuen> the idea being that going forward, we want to make all config files injectable through values overrides, rather than statically defined in the chart
15:07:41 <mattmceuen> doing this consistently will make it really easy to configure the charts in ways that were unforseen at the time of chart creation -- no code changes needed
15:08:17 <mattmceuen> So one takeaway is - as we add more config files, let's please add them via the values file-based approach that we're moving to!
15:08:30 <mattmceuen> Any questions / comments on this topic?
15:09:44 <mattmceuen> #topic Ceph split
15:10:19 <mattmceuen> As we discussed last week (or was it two weeks ago) -- we are moving to split up ceph into a chart for the mons, one for osds, and one (or more) for other ceph services
15:10:42 <mattmceuen> Renis and Randeep are working on this -- there's a PS in progress, so feedback is welcome!  https://review.openstack.org/#/c/558326/
15:10:43 <rwellum> Matt sorry slow here - one qn about the values-driven overrides?
15:10:51 <mattmceuen> hey rwellum sure thing
15:11:27 <rwellum> Just I think this is a great idea, will it help with upgrading to newer OS version? Because with the correct logic the same charts can be used?
15:12:56 <portdirect> yes
15:13:10 <mattmceuen> We're actually planning to use the same charts, no matter what :)
15:13:12 <portdirect> a great example of this is rootwrap changes between os versions
15:13:22 <portdirect> and this change really faciliates that
15:13:36 <portdirect> eg things like this: https://review.openstack.org/#/c/545743/
15:13:43 <portdirect> can be soly driven via values now :)
15:13:50 <portdirect> without having to edit the charts :)
15:14:02 <portdirect> *solely
15:14:23 <mattmceuen> I didn't mention this, but there is support for templating in the config files as embedded in the overrides.  So very flexible.
15:14:23 <mattmceuen> example: https://review.openstack.org/#/c/557882/3/ironic/values.yaml
15:14:33 <rwellum> I actually thought you guys had this functionality a long time back.
15:14:45 <rwellum> Ah right maybe this is what I am remembering.
15:15:11 <portdirect> we have had some varitions on this in the past
15:15:33 <portdirect> but this much simpler, and more powerful than we had previously
15:17:09 <mattmceuen> thanks portdirect.  Anything else on either config or ceph front guys?
15:17:38 <gmmaha> on ceph front, as renis and randeep are taking the lead, i will be the soundboard on this and test the patches out as they flow.
15:17:44 <gmmaha> synced with renis yesterday on this.
15:17:50 <gmmaha> thanks mattmceuen
15:18:02 <rwellum> Why the ceph split? Charts too complicated?
15:18:25 <mattmceuen> the ceph split is to facilitate upgrades
15:18:50 <mattmceuen> to guarantee an ordering so that the mons can be upgraded first, then the osds, then anything else, per ceph upgrade best practices
15:19:16 <mattmceuen> Thanks gmmaha!!
15:19:29 <mattmceuen> Testing the upgrade process IRL will be incredibly valuable to validating this work
15:19:42 <gmmaha> ++
15:19:55 <rwellum> So this is more of a micro-architecture break down then?
15:20:28 <mattmceuen> I would call it a right-sizing micro-architecture break down, yes :)
15:20:47 <mattmceuen> It doesn't imply we want to split up every chart or anything - just when needed to do what we need to do
15:22:07 <mattmceuen> #topic PS needing review
15:22:08 <rwellum> Ok ty
15:22:39 <mattmceuen> What do we need eyeballs on, folks -- I already mentioned the first (of multiple) ceph split PS:  https://review.openstack.org/#/c/558326/
15:22:52 <portdirect> rwellum: its to model the same deployment as ceph-deploy
15:23:08 <portdirect> as ceph is like openstack, a collection of applications
15:23:19 <portdirect> mons, osds, clients etc
15:23:44 <portdirect> so here a single chart is somewhat analogous to having a dev-stack chart
15:23:54 <rwellum> portdirect: but you can't do ceph with no mons or no osd's etc? Which is why I was curious.
15:24:32 <portdirect> yup
15:24:46 <portdirect> this is also part of the shift towards us moving ceph out of osh
15:24:51 <portdirect> and into ceph-helm
15:26:01 <rwellum> Oh yeah that actually makes complete sense.
15:28:04 <mattmceuen> https://www.irccloud.com/pastebin/SSrJtyZS/
15:28:32 * mattmceuen needs to look before he clicks on buttons in irccloud
15:28:45 <mattmceuen> Other PS that anyone is waiting on?  I wanted to get this on our radars as well, since it's important that we provide quick reviews to new contributors to our project:
15:28:45 <mattmceuen> https://review.openstack.org/#/c/557586/
15:30:27 <mattmceuen> #topic Roundtable Discussion
15:30:54 <mattmceuen> Open forum folks - what would you like to discuss, learn, or vent? :)
15:31:17 <rwellum> Qn: you guys doing official releases yet?
15:32:20 <rwellum> As kolla-k8s enters it's terminal state - I will have time to get involved, so I'll try not to be too annoying :)
15:32:39 <mattmceuen> We had a lot of good disucssion on this at the PTG!
15:32:41 <portdirect> not yet - though mattmceuen has a spec up for the critieria we are aiming for
15:32:52 <portdirect> and I'm working on one for the release versioning strat
15:33:08 <portdirect> this is the major focus we have atm
15:33:24 <portdirect> with a goal of our 1st official release in the next 6-8 weeks
15:33:38 <mattmceuen> agree, we will get release strategy well defined and communicated soon
15:34:14 <mattmceuen> also rwellum - welcome! :)
15:35:16 <rwellum> That's great news. yeah I missed the PTG (changing jobs).
15:35:21 <rwellum> Thanks Matt.
15:36:01 <mattmceuen> Here are our notes - the rough sketch of release strategy can be found there, although it of course would require some squinting if you weren't present:  https://etherpad.openstack.org/p/openstack-helm-ptg-rocky
15:36:53 <mattmceuen> Ok guys - I know a lot of team member are very heads down or travelling this week, so I won't keep y'all longer than needed.  Have a great week, thanks for all your hard work, and see you in the team chat
15:36:58 <mattmceuen> #endmeeting