15:00:30 <srwilkers> #startmeeting openstack-helm
15:00:30 <openstack> Meeting started Tue Jul 18 15:00:30 2017 UTC and is due to finish in 60 minutes.  The chair is srwilkers. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:31 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:33 <openstack> The meeting name has been set to 'openstack_helm'
15:00:39 <srwilkers> good morning o/
15:01:14 <dulek> o/
15:01:25 <lrensing> o/
15:01:27 <portdirect> oh hai \o
15:01:30 <xek> hi o/
15:01:56 <srwilkers> heres the link for today's agenda -- it's fairly light, probably in part to me not getting it out in a timely manner
15:01:57 <srwilkers> https://etherpad.openstack.org/p/openstack-helm-meeting-2017-07-18
15:02:02 <srwilkers> that'll be different for next week
15:02:24 <srwilkers> lets take a few minutes and see if we can add some more points to it
15:02:38 <v1k0d3n> o/
15:02:48 <srwilkers> hey v1k0d3n o/
15:02:59 <raymaika> hey guys o/
15:03:10 <v1k0d3n> on phone though so very limited. Still transitioning...moving this week.
15:03:19 <alraddarla> o/
15:05:00 <srwilkers> alright, we can get started and just open for discussion and see what else comes up after what's in the agenda
15:05:18 <srwilkers> #topic LMA spec
15:05:49 <srwilkers> i drafted a spec for an LMA approach for openstack-helm, and can be found here: https://review.openstack.org/#/c/482687/
15:06:22 <srwilkers> id really like some additional operations-focused feedback on what's there so far
15:07:34 <dulek> While I cannot provide much ops feedback on that, I sign up for reviews and testing of the implementation.
15:07:58 <srwilkers> awesome dulek. :)
15:09:42 <srwilkers> once we get some more feedback on the spec and can identify some solid operational requirements for configuration, it'll make tying up some of the services that are in flight much easier
15:10:12 <srwilkers> and also welcome those interested in helping get them to the finish line if anyones interested in contributing on that front
15:10:34 <alraddarla> \o/
15:11:59 <srwilkers> thats all ive got for this topic, unless there are other points someone would like to raise :)
15:12:49 <srwilkers> #topic Memcached resiliency
15:12:53 <srwilkers> floors yours xek
15:13:12 <xek> alanmeadows pointed me to an issue with memcached which I would like to work on
15:13:27 <xek> the issue is that memcached is not clustered in any way right now
15:13:47 <xek> I wanted to bring the blueprint to your attontion so that there is a consensus in the right solution
15:14:07 <xek> in short, I want to introduce mcrouter into the mix and change the defaults so that all services point to mcrouter
15:14:27 <xek> it can then be configured to talk to more then one memcached instance in any way which is desired by the operator
15:14:34 <xek> to meet distribution and ha requirements
15:15:17 <portdirect> sounds sane
15:15:35 <portdirect> reading bp now
15:15:40 <dulek> I can see that "official" chart uses just a service for load balancing: https://github.com/kubernetes/charts/tree/master/stable/memcached
15:15:54 <dulek> xek: Is what you're proposing a better idea?
15:16:16 <v1k0d3n> too bad we're not just using upstream for some things like this.
15:16:28 <portdirect> I would have thought so - a service really has limitations here
15:16:32 <v1k0d3n> but control can be an issue too. i see both sides.
15:16:47 <portdirect> and does not really do what we need
15:17:22 <dulek> Okay, I see no replication in this upstream chart, so it's rather poor
15:17:43 <xek> dulek, with default round-robin balancing you can wind up in a situation where you only use one memcached server or save in one instance and try to read from another
15:17:43 <portdirect> it replicates but relys on a 'headless service'
15:18:51 <portdirect> which doesnt allow reconfig/scaling etc after a pod that consumes it has started
15:18:51 <dulek> Also a funny thing is that Kolla has single replica hardcoded: https://github.com/openstack/kolla-kubernetes/blob/master/helm/microservice/memcached-deployment/templates/memcached-deployment.yaml#L33
15:19:14 <xek> memcached itself doesn't replicate key-values
15:19:59 <portdirect> xek: I'd be happy to approve this, but obviously would want consensus
15:20:26 <srwilkers> +1
15:22:55 <srwilkers> yeah, this blueprint is great.  id also approve moving forward with this approach
15:23:26 <lrensing> i'm good with you taking it on xek
15:23:42 <portdirect> so we have consensus?
15:23:45 <lrensing> i look forward to seeing memcached move forward :)
15:23:48 <lrensing> yes from me
15:24:28 <portdirect> nice one xek :) memcached has been a thorn in my oslo side too long :D
15:25:24 <srwilkers> any other comments or concerns here?
15:26:24 <srwilkers> #topic PTG planning
15:27:17 <srwilkers> so the PTG is less than two months away, and i think it'd be a good idea to start drafting up the items we'd like to target for those who will be present that week
15:27:37 <srwilkers> since we're a hosted project, we won't be allocated reserved space and we won't be included on the schedule
15:28:14 <srwilkers> but i've been told we're free to use the common spaces to meet, and we're responsible for advertising where we'll be hanging out while we're there
15:28:51 * portdirect votes for by the pool
15:29:40 <srwilkers> here's an etherpad we can use to add ideas/topics to between now and then:  https://etherpad.openstack.org/p/openstack-helm-denver-ptg-topics
15:29:59 <srwilkers> and i'll make sure to bring it up each weekly meeting as a friendly reminder
15:30:13 <lrensing> that'll be good srwilkers
15:30:41 <srwilkers> really looking forward to seeing some of you there :)
15:32:08 <srwilkers> but that's all i have wrt the PTG
15:32:24 <srwilkers> since we're out of topics, we've got ~30 minutes for any open discussion items
15:32:31 <srwilkers> #topic open discussion
15:32:47 <lrensing> i'll kick this one off since it's been a topic in irc this morning-
15:33:28 <lrensing> the ceph chart has been going through a few changes this week and some community members have had some issues getting it redeployed
15:33:52 <raymaika> o/
15:34:43 <lrensing> i believe current master is stable again, but if anyone finds any issues with it, bring it up in IRC so we can discuss, and if not much comes from that discussion please file a bug in launchpad so we can address it there
15:35:23 <portdirect> I've got 2 labs running from master (deployed in last 18 hours)
15:35:58 <portdirect> both seem stable - though I would not be surprised if there a few gremlins in there - dependant on network config
15:36:18 <lrensing> ++
15:36:32 <portdirect> its worth noting that we use ceph for all gate jobs - so it is tested for both pvc, and glance over 140 times a day :D
15:36:54 <lrensing> also, configuring ceph to use disks instead of directories is coming soon
15:36:55 <portdirect> but thats on the same hw config....
15:37:49 <lrensing> but anyways - thats all i had on that
15:40:16 <raymaika> yeah, I'm trying to find the root cause of my problem. I'll let you know if I manage to fix it or find something that points to a more specific problem than the logs I posted this morning.
15:42:44 <dulek> As the discussion died a bit - I've rebased Keystone fernet tokens patch again if anyone is interested: https://review.openstack.org/#/c/463730/
15:44:20 <srwilkers> awesome dulek.  we'll have a look at that today for sure
15:45:23 <portdirect> dulek: I 'm gonna enable cronjobs in the k8s api, and update docs accordingly today
15:45:41 <dulek> portdirect: That would be awesome.
15:45:46 <portdirect> at that point I'd really like to make this the default and merge
15:45:57 <portdirect> thanks for all your work on this dude :D
15:46:10 <srwilkers> agreed.  looking forward to getting it to the finish line
15:46:16 <dulek> :)
15:47:42 <srwilkers> any other topics?
15:48:42 <lrensing> none from me - given the long pause i think we can wrap this one up a little early
15:49:56 <srwilkers> alright.  ill go ahead and end the meeting.  thanks everyone :)
15:49:58 <srwilkers> #endmeeting