15:00:42 <mattmceuen> #startmeeting openstack-helm
15:00:42 <openstack> Meeting started Tue Oct 31 15:00:42 2017 UTC and is due to finish in 60 minutes.  The chair is mattmceuen. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:43 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:46 <openstack> The meeting name has been set to 'openstack_helm'
15:01:10 <mattmceuen> #topic rollcall
15:01:25 <mattmceuen> Happy Halloween everyone :)
15:01:38 <jayahn> ah.. yeap. it is halloween. :)
15:01:53 <jayahn> mostly, it means party time in here. :)
15:02:05 <mattmceuen> Here's the agenda -- I'll give folks a couple mins to join and add agenda items -- https://etherpad.openstack.org/p/openstack-helm-meeting-2017-10-31
15:02:42 <lamt> Happy Halloween o/
15:02:46 <mattmceuen> @jayahn same!  Adults get to party and kids get candy, everyone wins
15:03:35 <srwilkers> o/
15:04:11 <portdirect> \o\
15:05:08 <srwilkers> ./o/
15:05:32 <srwilkers> :me thinks that looks an awful lot like a tie fighter doing a barrel roll
15:05:41 <jayahn> we are doing irc text dancing? :)
15:05:48 <mattmceuen> #topic comparison of log-based alerting software used in OSH
15:06:12 <mattmceuen> I don't think we have @mateuszb here today (correct me if wrong mateuszb)
15:06:42 <mattmceuen> Torelay his thoughts though:  <mateuszb> I just wanted to let you know that I prepared patchets for ElastAlert (https://review.openstack.org/#/c/516629/ ) and fluent-plugin-prometheus (https://review.openstack.org/#/c/514938/ and https://review.openstack.org/#/c/515061/ ). I'll be more than happy to hear some feedback - I'd like to make a decision  which one of the solutions is preferable so that I may
15:06:43 <mattmceuen> continue to improve it
15:06:49 <jayahn> i think you are right. mateuszb is not here today
15:07:39 <mattmceuen> I'll move his agenda topic to the next meeting, but if you guys can take a look and get him feedback on the reviews, that would be awesome
15:08:31 <portdirect> will do
15:08:34 <jayahn> yeap
15:08:53 <portdirect> we really need to get the review que back under control this week as well
15:09:07 <mattmceuen> Yeah, it's been a rough week all around
15:09:17 <mattmceuen> #topic Sydney Summit - social gathering?
15:10:06 <mattmceuen> It's a short summit!  I can say that I'm planning on joining the WoO outing to tour Sydney and hear ghost stories on Monday night, and am very open to other events as well :)
15:10:06 <jayahn> just wanted to know, since i am planning for "evenings". :)
15:10:22 <srwilkers> i will join you all in spirit
15:10:31 <jayahn> Monday night, there will be LCOO social gathering.
15:10:33 * mattmceuen halloween humor
15:10:41 <srwilkers> ;)
15:10:42 <portdirect> jayahn there are a *few* bars in sydney
15:10:50 <jayahn> I am thinking doing skydiving on Sunday
15:10:53 <srwilkers> nothing like austin im sure
15:11:00 <jayahn> :).. that is sad
15:11:09 <mattmceuen> yeah - I'll be at LCOO too, and will go straight to the ghost thing from there
15:11:19 <portdirect> jayahn: I may be up for jumping out of a plane if you are game?
15:11:22 <jayahn> ah-ha
15:12:03 <mattmceuen> half the OSH team jumping out of a plane, why are we tempting fate, people!
15:12:04 <jayahn> portdirect: you can join. i am thinking 11am schedule
15:12:07 <srwilkers> madness
15:12:24 <portdirect> jayahn: will check my flights
15:12:43 <portdirect> mattmceuen: sorry for once again de-railing a perfectly good meeting
15:12:46 <srwilkers> dont worry mattmceuen -- ill be safe in the US as a contingency plan
15:13:07 * mattmceuen looks up armed guard services
15:13:13 <mattmceuen> ok
15:13:35 <mattmceuen> #topic newton release cut-off
15:14:02 <mattmceuen> I have done a poor job of preparing a list, but an excellent job of procrastinating.
15:14:28 <mattmceuen> Speaking with portdirect, though, it sounds like I may have been overestimating the effort of OSH release alignment
15:14:29 <jayahn> great! you are exactly same as me.
15:14:36 <portdirect> 1) support node/cluster reboot fully
15:14:56 <portdirect> 2) get the charts in correct repos
15:15:18 <portdirect> 3) get voting gates
15:15:30 <jayahn> regarding 1, i assume we only need to solve galera cluster stuff? right?
15:15:43 <portdirect> as of today i think so
15:16:31 <mattmceuen> if we can make OSH release-agnostic (rather than branching by OS release), then it's a much lighter-weight activity to support newton and ocata (and ...) in parallel, and (key point) to continue to develop and gate in parallel.
15:16:37 <portdirect> SamYaple has been doing some work there for his own salt work
15:16:45 <mattmceuen> So that's awesome.
15:16:53 <jayahn> we did some sort of extensive test on node/pod/network failure. i think galera is only thing not recovering properly after "all-nodes down".
15:17:00 <portdirect> I'd like to loop him into the mariadb discussion - as hes got a really good grasp on it
15:17:08 <jayahn> i saw SamYaple's comment here.
15:17:50 <jayahn> any upgrading requriement?
15:18:30 <portdirect> mattmceuen: I fully agree with making osh release agnostic if possible
15:19:09 <portdirect> jayahn: I don't fully follow your last point
15:19:23 <mattmceuen> jayahn - I think upgrades should be included as first-class, do you have any concerns with it we need to account for?
15:20:02 <jayahn> no. it was a question asking if upgrades should be included as first-class. (thank you for saying that)
15:20:09 <jayahn> my personal opinion is no.
15:20:34 <jayahn> FYI, while we are testing upgrade, most of things are okay. i found out that if we upgrade libvirt image, then the existing VMs are gone while we are doing upgrade via helm.
15:20:54 <portdirect> jayahn: could you file a bug?
15:21:05 * portdirect bets he has already...
15:21:16 <jayahn> we are discussing if that is bug or expected behaviour.
15:21:22 <portdirect> even if we dont have a solution - I'd like to get this documented
15:21:30 <jayahn> okay. i will do that
15:21:36 <jayahn> bug report.
15:21:40 <portdirect> so that we can make upgrades as painless as possible
15:21:43 <mattmceuen> Agree - I'd like that to be fixed too.  Ease of upgrades is one of the big selling points of OSH IMO
15:22:14 <jayahn> if we are upgrading only with configuration changes. there is no problem at all, even for libvirt.
15:22:15 <portdirect> I think its ok to say : you can upgrade nova, but not libvirt (ideally we'd work out the version delta that is supported)
15:22:24 <jayahn> it is only when we actually upgraded image itself.
15:23:00 <jayahn> i will do bug report as portdirect asked.
15:23:18 <mattmceuen> So portdirect, I guess that means we will want newton gates, ocata gates, and newton->ocata gates? :)
15:23:31 <portdirect> yup
15:23:48 <mattmceuen> wunderbar
15:23:58 <portdirect> ...pike, queens and master
15:24:05 <portdirect> for kolla and loci
15:24:37 <portdirect> and for loci we should also gate python3 for pike (poss queens) and above
15:24:58 <portdirect> jayahn: are you guys still running centos?
15:25:14 <jayahn> we are currently running ubuntu.
15:25:16 <portdirect> I really need help with the non-ubuntu gates to get them voting in osh-infra
15:25:23 <jayahn> however, we need to test centos again soon
15:26:12 <portdirect> the funny thing is i run centos and fedora in my personal labs - everything works perfectly - but somthing happens in zuul that causes issues with calico
15:26:24 <jayahn> hm..
15:26:45 <portdirect> we could move to flannel on those platforms - but I'd really like them to support network policy via cni as we begin to secure things
15:28:01 <mattmceuen> agree -- we are targeting OSH for broad use, and should gate accordingly
15:28:09 <jayahn> i cannot say we will be able to help on that right now. I have to see after the summit.
15:29:08 <mattmceuen> I think we can say that we're targeting OSH for centos, and it's just a matter of time till we get voting gates for it
15:29:27 <portdirect> It would be great to get some addtional eyes on the gate work I've done in osh-infra
15:29:53 <portdirect> it redoes the kubeadm aio image entirely to be something a lot more solid
15:30:23 <portdirect> and *almost* (famous last words) production ready
15:30:45 <portdirect> I need to get the docs worked for it properly
15:31:12 <portdirect> but i think a few people (lamt, srwilkers ?) are using it for their work as well
15:31:31 <srwilkers> yep, running it as we speak
15:31:33 <srwilkers> works like a dream
15:31:51 <mattmceuen> speaking of osh-infra and returning to the point about moving projects to the right repos - segueway!
15:32:02 <mattmceuen> #topic moving charts to the right repos
15:32:05 <portdirect> lol
15:32:17 <mattmceuen> What all charts need to move where
15:32:18 <portdirect> so the 1st step is getting the gate solid in osh-infra
15:32:31 <portdirect> as we will use this as the base for all other gates
15:33:29 <portdirect> please put your eyes on this https://review.openstack.org/#/c/514892/ as i could really do with feedback
15:33:49 <portdirect> but once thats in then we should move the charts as follows:
15:34:45 <portdirect> osh-infra: ceph, nfs-provisioner, memcached, mariadb, etcd, rabbitmq, libvirt, ovs, and Prometheus/fluentd
15:35:40 <portdirect> the last two may be controversial - but i feel that they are essential for us to have a comprehensive LMA stack in OSH and thus form part of the essential infra for a real-world deployment
15:36:05 <portdirect> addtionally we can scrape the results from them in the gate to use for debugging and dev work
15:36:38 <portdirect> oh and ONOS/opencontrail as well in osh infra ;)
15:36:53 <portdirect> the osh repo will then contain only openstack projects
15:37:02 <portdirect> and anything else will move to addons
15:37:09 <mattmceuen> I don't hate the idea of moving those things into osh-infra -- I think we kinda return to the mission statement of osh-infra that way.  Any dissenting thoughts?
15:38:03 <alanmeadows> shouldn't we stretch the envelope with an ocata->newton gate?
15:38:19 <alanmeadows> (revert to db snapshot yields working system)
15:38:19 <portdirect> yes
15:38:32 <mattmceuen> great point alan
15:38:38 <portdirect> but thats somthing i think we should add later
15:38:46 <alanmeadows> sure, prioritized last
15:39:15 <mattmceuen> We have in-flight patchsets for some of the stuff targeted for osh-infra above (e.g. Ceph and LMA).  Should we try to get them merged before moving to osh-infra, or abandon/recreate PS, or depends?
15:39:16 <portdirect> osh - the first project thats lets you go back in openstack time :)
15:39:42 <alanmeadows> I think its a great tagline that we are serious about openstack version mobility
15:39:47 <alanmeadows> anyways, digression sorry
15:40:22 <portdirect> can we call it flux capacitor? RDO already have delorean.
15:40:41 <alanmeadows> If the intention is to line things into the proper repositories it seems like the goal should be a scramble to get them into infra?
15:41:03 <alanmeadows> or am I misstating what was said above
15:41:31 <mattmceuen> scramble:  yes
15:41:38 <mattmceuen> I just want to define scramble
15:41:39 <portdirect> mad dash: no
15:41:49 <mattmceuen> scramble like sports or like eggs
15:42:16 <jayahn> scramble to get them into infra >> can you guys explain what it mean exactly? :)
15:42:17 <portdirect> but yup alanmeadows we are all aligned there
15:42:30 <portdirect> jayahn: push/rush
15:42:30 <alanmeadows> Little legs scurrying but in an orchestrated fashion and with purpose
15:42:31 <jayahn> sorry about not getting english..
15:42:58 <mattmceuen> we only barely speak english ourselves jayahn :)
15:42:59 <portdirect> alanmeadows: almost like hamsters in a wheel?
15:43:11 <alanmeadows> portdirect: I just said little legs...
15:43:27 <portdirect> i knew what you were thinking...
15:43:35 * alanmeadows blushes
15:43:48 <jayahn> okay. direct question. what would we do with LMA patchsets in openstack-helm repo?
15:44:05 <mattmceuen> so if a PS is very close to ready, we should push to get it merged into OSH prior to moving to OSH Infra.  Like, a next-couple-days exercise.    If it still needs substantial work, the PS should be abandoned and redone against osh-infra.  Agree?
15:44:07 <jayahn> waiting for osh-infra ready, then move. (hold more patchset works meanwhile?)
15:44:22 <jayahn> ah.
15:44:23 <jayahn> got it
15:44:24 <portdirect> jayahn: I'd keep working there untill we have basic cross repo gating in place
15:44:57 <alanmeadows> is cross-repo gating a target for a release cut?
15:45:04 <alanmeadows> or will that need to follow
15:45:04 <portdirect> alanmeadows: yes
15:45:23 <portdirect> i think it really defines the point at which we have a release
15:45:37 <portdirect> everything is int its correct place
15:45:55 <jayahn> agreed on that
15:46:01 <alanmeadows> it also indicates when folks could safely take things as PS to OSH and resubmit them safely (with tests indicating it doesnt break things)
15:46:13 <portdirect> ++
15:47:05 <mattmceuen> are we good with a timeline of "thursday" to move the projects to infra, or is that too ambitious?
15:47:24 <alanmeadows> sounds like it all boils down to when cross-repo gating is a lock
15:47:44 <alanmeadows> if its at the tail end of things, doesn't sound like much movement will occur
15:48:18 <mattmceuen> ok.  Call to action: try to get open PS against those items closed soon if possible!
15:48:43 <mattmceuen> (indluding the new gate foundation too)
15:48:58 <mattmceuen> next:
15:49:14 <portdirect> mattmceuen: Thursday is ambitious - but possible
15:49:38 <mattmceuen> "shoot for the stars, make it to the top of the stairs"
15:49:45 <mattmceuen> #topic next two team meetings
15:50:37 <mattmceuen> Several of us will be in Sydney next week and will be pretty busy.  v1k0d3n has volunteered to lead the meeting next week for those who are able if folks are interested -- ?
15:50:57 <srwilkers> i think with most of the team in sydney/travelling, we should just cancel the next two meetings
15:51:06 <srwilkers> that's typically what happens for the summit anyway
15:51:13 <jayahn> two meetings? not only next one?
15:51:29 <mattmceuen> I am on vacation the week after next :)  not sure about everyone else though
15:51:44 <srwilkers> i think some people are staying in sydney a bit later after the summit.  would give everyone time to travel and readjust
15:52:14 <portdirect> I'll be back for the one after the summit
15:52:31 <srwilkers> and i imagine the discussions at the summit will tie into the work we do moving forward -- would be nice to have everyone present if possible
15:52:48 <portdirect> srwilkers: would you be up for leading a little one then?
15:53:16 <portdirect> so those who were there can give a recap of what happened - and then the next one we'll be back up to full steam
15:53:27 <srwilkers> whatever works
15:53:28 <portdirect> as i think only mattmceuen and alanmeadows will be ooo
15:53:46 <mattmceuen> If just alan & me are gone, let's meet in our absence
15:54:03 <mattmceuen> v1k0d3n, srwilkers, if you guys can make that happen it would be appreciated
15:54:08 <mattmceuen> and we call off next week
15:54:18 <alanmeadows> sounds like just a single skip is in order
15:54:35 <jayahn> quick question, you guys are working on 28th Nov? when is thanksgiving holiday?
15:54:47 <portdirect> jayahn: I will be
15:54:55 <mattmceuen> I'll be working too
15:54:59 <srwilkers> 23rd is thanksgiving
15:55:09 <jayahn> ah.. got it.
15:55:15 <mattmceuen> #topic Needs Review
15:55:22 <alanmeadows> @jayahn: this will be portdirect on thanksgiving http://www.seriouseats.com/images/2014/11/20141128-twase-1.jpg
15:55:29 <v1k0d3n> Yup! On it mattmceuen
15:55:36 <jayahn> nice
15:55:52 <srwilkers> alanmeadows: ;)
15:56:00 <mattmceuen> https://www.irccloud.com/pastebin/8BqUQnTX/
15:56:26 <jayahn> just want to have your eyes on "nova placement" and "fluent logging".
15:57:14 <mattmceuen> Yeah, let's please get reviews on those and get things as ready as possible before the summit (esp the fluent piece given the osh-infra discussion)
15:57:19 <portdirect> nova-placement is essential for >newton so lets get that in asap
15:57:47 <alanmeadows> I will take a look at placement today - we have been doing a lot of ocata uplift work and had to tinker with the outstanding PS to get things going
15:58:08 <alanmeadows> (especially being considerate of it doing *both* newton and ocata)
15:58:41 <jayahn> we are also doing ocata uplift work.
15:59:09 <mattmceuen> alright folks, we're about out of time!  last call!
15:59:20 <jayahn> srwilers: can you time to look at fluent-logging?
15:59:29 <jayahn> srwilkers
15:59:36 <jayahn> 1min left. :)
15:59:36 <srwilkers> yep, ill look at it now
15:59:40 <mattmceuen> #endmeeting