15:00:50 <srwilkers> #startmeeting openstack-helm
15:00:50 <openstack> Meeting started Tue Jun 27 15:00:50 2017 UTC and is due to finish in 60 minutes.  The chair is srwilkers. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:51 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:53 <openstack> The meeting name has been set to 'openstack_helm'
15:00:59 <srwilkers> morning everyone o/
15:01:14 <jayahn> morning
15:01:35 <portdirect> o/
15:01:38 <dulek> o/
15:01:40 <portdirect> dulek: https://etherpad.openstack.org/p/openstack-helm-meeting-2017-06-27
15:01:54 <srwilkers> thanks portdirect
15:01:56 <dulek> portdirect: Oh, why me? :D
15:01:58 <jayahn> really sorry about this stupid auto-translation.
15:02:08 <srwilkers> jayahn, no worries :)
15:02:12 <jayahn> normally i don't use it.
15:02:23 <portdirect> oh - lol sorry dulek :)
15:02:38 <srwilkers> we've got a few items on the agenda. if anyone wants to add to it, feel free to do so
15:02:38 <portdirect> hers trouble - hey anticw o/
15:02:57 <srwilkers> just in time anticw
15:03:08 <srwilkers> #topic ceph secrets generation
15:03:26 <srwilkers> anticw, portdirect, lrensing: floors yours
15:04:04 <portdirect> o/
15:04:07 <lrensing> o/
15:04:22 <portdirect> so have a big scary ps in flight: https://review.openstack.org/#/c/474317/
15:04:45 <portdirect> would really appreacate fb - though I'm aware its rough round the edges still :)
15:04:58 <portdirect> but it should greatly simplify ceph deployment
15:05:17 <portdirect> removing the secret gen stuff from helm-toolkit
15:05:30 <portdirect> and making the ceph chart standalone
15:05:43 <lrensing> yep - something that will definitely need to be added (sometime) is a cleanup secrets job on deletion
15:06:06 <portdirect> we can do that via lifecycle hoosk really easily :)
15:06:11 <portdirect> *hooks
15:06:13 <lrensing> since they are generated in a roundabout way that normal helm delete can't detect
15:06:16 <lrensing> exactly :)
15:06:43 <portdirect> this also move the dev env default to ceph
15:07:07 <portdirect> and I've tested it on a number of platforms: ubuntu, centos, fedora, and atomic
15:07:30 <srwilkers> portdirect, thats awesome. means we can get rid of nfs then, yea?
15:07:34 <anticw> it's a fairly large patch, it conflicts with some of the other patches out there so having as many comments sooner will be useful
15:07:44 <portdirect> ++ anticw
15:08:14 <portdirect> srwilkers: it means we could - though I'd like t keep it - as ceph is not for everyone
15:08:24 <anticw> nfs is still useful
15:08:30 <portdirect> low power hardware etc
15:08:40 <anticw> sometimes you need to test things end-end quickly and nfs is maybe 3x or more faster than ceph
15:08:44 <portdirect> ++
15:09:22 <portdirect> v1k0d3n know this is an area that interests you a lot - so would value any input you have?
15:09:25 <anticw> portdirect: off the top if my head all i see remaining are some nova tweaks and docs?
15:09:31 <portdirect> yeah
15:09:45 * portdirect he says confidently
15:09:54 <anticw> oh, and cleanup on remove
15:10:15 <portdirect> yeah - I think most of the cleanup I can get done between calls today
15:10:54 <portdirect> the ceph.conf gen itself leavs somthing to be desired
15:11:06 <portdirect> but I'd really like to have that at a follow on ps
15:11:29 <portdirect> as this thing is already huge.
15:11:40 <srwilkers> agreed
15:11:58 <anticw> i wondered if it could be done in smaller parts, but it's not clear that it easily can be
15:12:53 <portdirect> I thought and tried to work out a way to do it - but any other approach would have either left master temp broken, or had a chain of ps's all highly dependent on each other
15:12:55 <v1k0d3n> sorry, was distracted. looking over...looks really good. it's a pretty large PS to go over.
15:13:21 <v1k0d3n> really keen on putting ceph issues behind us.
15:13:36 <portdirect> I hope the gate log dumps make it easier :)
15:14:08 <v1k0d3n> ++
15:14:29 <srwilkers> any other concerns/points for the ceph changes?
15:15:04 <anticw> there are other ceph cleanups that would be useful to make things robust on restart, https://review.openstack.org/#/c/474011/ for example
15:15:10 <portdirect> if not - I'll try and get it in a merge ready state by EOD
15:15:21 <portdirect> anticw: that ones a good un
15:15:25 <anticw> i don't know what alanmeadows is planning long-term with that but presumably that will beed to be refactored
15:15:30 <portdirect> i think its the only one in flight right now
15:15:35 <anticw> portdirect: i have the 'nova ceph fixes' as well to refactor
15:15:48 <anticw> i'm going to leave both of those until the dust settles on the secrets stuff
15:16:11 <portdirect> cool - it should be possible for your nova work to slide in really nicely
15:16:30 <portdirect> I was looking at it while writing some of this - so a rebase should not be too hard
15:16:45 <srwilkers> portdirect, nice
15:16:55 <anticw> i think we'll be fine
15:17:09 <srwilkers> going to move on since we've got a full agenda -- we can carry any remaining ceph stuff to osh channel
15:17:12 <srwilkers> #topic PTG
15:17:24 <srwilkers> jayahn brought up some points for the PTG
15:17:30 <srwilkers> before we dive into those though
15:18:06 <srwilkers> i reached out to ttx about arranging for presence for openstack-helm at the PTG.  since we're a hosted project, they cant schedule us in a room proper or put us on the agenda, however
15:19:01 <srwilkers> he did say we're free to use the drop in rooms and space for openstack-helm, given that we handle the messaging and arrangements ourselves. ill be firing out emails related to this over the next few days
15:19:09 <srwilkers> jayahn, floors yours :)
15:20:12 <jayahn> not much to say. we are looking into making monasca chart, and found monasca-helm repo. then thought it would be a good cross-project topic at PTG
15:20:54 <jayahn> since we can get monasca guys working together on helm. i used to know monasca ptl from HPE. I will try  to find out he is still PTL.
15:21:09 <jayahn> then, suggest some communication over helm chart collaboration
15:21:17 <srwilkers> jayahn, i started looking at monasca-helm. is it specific to gathering openstack metrics?
15:21:35 <jayahn> yes. gathering and analyzing
15:21:56 <jayahn> it seems like monasca-agent has plugin to prometheus as well
15:22:00 <dulek> jayahn: Roland Hochmuth is currently Monasca's PTL from HPE.
15:22:14 <jayahn> yes, same guy then.
15:22:19 <srwilkers> yeah, i saw the prometheus bits there
15:22:58 <jayahn> i will try to talk to him in this week
15:23:01 <portdirect> are they on irc?
15:23:09 <portdirect> would be good to have sync up before?
15:23:17 <jayahn> yes, there is #openstack-monasca channel
15:23:27 * portdirect joining :)
15:23:44 <srwilkers> jayahn, thatd be awesome. ive got a helm chart in progress for the openstack-exporter by canonical for prometheus, but would be interested in looking at monasca-helm as an alternative to that exporter
15:24:20 <jayahn> yeap.
15:25:01 <srwilkers> any other PTG-related questions/points?
15:25:42 <jayahn> not much. i really need to get travel fund secured, but seems difficult.
15:26:10 * jayahn sad. the company is not happy about multiple international travel.. summits and ptgs..
15:26:10 <srwilkers> jayahn, i think they provide travel assistance, but honestly not sure how that process works :(
15:26:19 <srwilkers> they being the foundation
15:26:40 <srwilkers> #topic neutron-onos -> openstack-helm-infra
15:26:40 <jayahn> i know. i usually get travel assistance for the summit,
15:26:48 <portdirect> I've used it in the past - they were very helpful - Id really recommend reaching out jayahn
15:26:54 <jayahn> since i am an ambassador. :)
15:27:20 <srwilkers> was neutron-onos one of your topics as well jayahn?
15:27:29 <jayahn> anyway, i just wanted to say we will move neutron-onos to openstack-helm-infra.
15:27:36 <srwilkers> nice :)
15:27:39 <jayahn> pordirect: when you have some time, pls look over.
15:28:01 <portdirect> jayahn: will do - I've been very time poor - but promise to do it by wed
15:28:10 <portdirect> I have a lab setup that I plan to deploy it on
15:28:12 <jayahn> we internally tested and it is working, but needs some documentation and guide
15:28:37 <korzen> so neutron-onos is copy-paste te neutron chart?
15:28:38 <jayahn> i will try to provide basic documents by end of this week
15:28:51 <srwilkers> jayahn, documentation would be awesome. we'd really like to see proper docs accompanying new charts/features coming in
15:29:07 <jayahn> korzen: yes, one engineer from my team is doing hard work copy & paste every changes in neutron chart. :)
15:29:24 <jayahn> from neutron chart to neutron-onos
15:29:34 <korzen> I was thinkin about small addition to neutron chart to support onos
15:29:47 <jayahn> yeap. that is also our final goal.
15:29:47 <portdirect> jayahn: lets work out how to take what you need from neutron with out having to be so wet :)
15:29:51 <korzen> and onos-related chart that would not need to copy everything from neutron
15:29:55 <portdirect> korzen: sounds intersting
15:29:59 <jayahn> sure.
15:30:28 <korzen> I am also short on time, but I will commend on the review about the idea
15:30:46 <jayahn> korzen: thanks!
15:30:50 <srwilkers> korzen, thatd be great
15:31:10 <korzen> I would do it tomorrow I hope ;)
15:31:14 <portdirect> I've been wondering about adding the ability to disable/enable each manifest in a chart - would allow real granularity of control if needed
15:31:33 <jayahn> next one is also mine. just letting know we are putting tempest addition. pls review and give us feedback.
15:31:56 <portdirect> jayahn: looked at it this morning - looks good so far
15:32:01 <srwilkers> #topic Adding Tempest
15:32:21 <srwilkers> jayahn, sounds good. will do :)
15:32:26 <jayahn> internally at skt, we are using upstream rally test for individiual chart testing, and then use full tempest testing with full openstack installation.
15:32:45 <jayahn> i was wondering how community upstream will use this tempest.
15:33:30 <srwilkers> jayahn, thats something we can discuss. i think theres a blueprint on launchpad currently. we could use that for a base for our discussions regarding tempest moving forward
15:33:52 <jayahn> sure. i will find out blueprint
15:34:31 <srwilkers> sounds good
15:35:00 <srwilkers> #topic LCOO JIRA issues
15:35:23 <srwilkers> dulek, i think this is yours
15:35:28 <dulek> Hi! I wanted to grab some work that isn't assigned in LCOO JIRA (https://openstack-lcoo.atlassian.net/projects/OSH)
15:35:56 <dulek> I wonder how up to date are OSH-27 (Common approach for post-resource Tenant injection in Keystone) and OSH-101 (Configure Keystone to use LDAP as authentication backend.)
15:36:01 <dulek> Any pointers?
15:36:10 <portdirect> dulek: i think we really need to go though that list
15:36:29 <dulek> portdirect: What do you mean?
15:36:37 <portdirect> OSH-27 and OSH-101 are done i think
15:37:27 <dulek> Oh, cool then. :D
15:37:36 <srwilkers> portdirect, agreed. we should review that list and see what's really remaining to work on
15:38:02 <portdirect> yeah - soon as we have it in a happy shape then I think it would be great to use it along with launchpad to co-ordinate our activities
15:38:09 <dulek> srwilkers, portdirect: That would definitely be helpful. I want to grab some work, but I need to know what's on the project's roadmap.
15:38:11 <portdirect> (at least for LCOO peeps)
15:38:47 <dulek> So I'll use my remaining time to remind my Keystone Fernet tokens (https://review.openstack.org/#/c/463730/) and credential tokens (https://review.openstack.org/#/c/468867/) patches.
15:38:51 <srwilkers> dulek, agreed. ive been working the last few days to tie things to blueprints in launchpad that exist currently. i think another diligent pass through the JIRA instance should result in an accurate portrayal of work for the roadmap
15:39:25 <portdirect> dulek: nice - though was there a reason we did not just use a pv for the rotation?
15:39:56 <portdirect> fair points about the atutomatic cron jobs
15:40:07 * portdirect shakes fist a cron :)
15:40:26 <srwilkers> cron >:(
15:40:29 <dulek> portdirect: PV would need to be read many, write one. That isn't supported by RBD.
15:40:43 <portdirect> where possible id really like to minimise interaction with the k8s api
15:41:00 <portdirect> could use another provider, or cephfs?
15:41:13 <portdirect> thats the approach some other projects have taken
15:41:39 <srwilkers> ++
15:41:43 <dulek> portdirect: Plese see initial comments from alanmeadows on patchset #2. I would be very glad if core team could come to a consistent decision there.
15:42:19 <dulek> Kolla-k8s is using secret to store keys. Same goes with fuel-ccp.
15:42:57 <portdirect> is that a recent change for kolla ? last i looked they were using cephfs
15:43:07 <dulek> And yes, cephfs would work in that case, but alanmeadows was argumenting that our primary choice is RBD.
15:43:54 <dulek> portdirect: Hm, I've looked a while ago - patch is 1,5 month old now. Maybe they've switched.
15:44:56 <portdirect> dulek: actually looks like they have moved the other way
15:45:04 <dulek> portdirect: Okay, they're using secrets for sure: https://github.com/openstack/kolla-kubernetes/blob/master/helm/microservice/keystone-fernet-rotate-job/templates/keystone-fernet-rotate-job.yaml#L107
15:45:30 <portdirect> I'll re-evaluate my position - but something does not sit right with me about this approach :)
15:46:18 <dulek> Yeah, PVC was much more simple and elegant, but this compatibility issue came into the picture. :)
15:46:42 <dulek> Okay, I don't want to take more meeting time, we have two more items left.
15:46:53 <srwilkers> dulek, sounds good
15:47:15 <srwilkers> #Topic announcements
15:47:30 <jayahn> Not in this meeting, but i would like to have some level of discussion regarding more visible way for openstack-helm blueprints to represent community roadmap, especially on deciding priorities.
15:47:34 <jayahn> ah,, announcement
15:47:43 <srwilkers> jayahn, i think this is your announcement :)
15:47:47 <jayahn> yeap
15:48:10 <jayahn> every info is there. :) july 13-14 we are having openstack days korea
15:48:21 <srwilkers> thats really awesome
15:48:22 <jayahn> and skt is doing openstack-helm hands-on tutorial at there.
15:48:39 <portdirect> thats awesome ! nice one jayahn
15:49:12 <jayahn> and I am personally trying to reach out others in korea to talk about "potential collaboration" on openstack-helm
15:49:20 <srwilkers> jayahn, nice
15:49:43 <srwilkers> #topic open-discussion
15:49:53 <jayahn> since i am the main organizer of the event. i can do whatever i want to do. :) kk
15:50:06 <srwilkers> figured the last ten minutes can be open discussion. jayahn, i agree about community roadmap and priorities
15:50:08 <jayahn> kidding.
15:50:26 <jayahn> ah, yes.
15:50:42 <jayahn> i would like to have that discussion some time soon.
15:50:59 <portdirect> srwilkers and I are planning on getting launchpad and the LCOO jira much more aligned this week
15:51:00 <srwilkers> the current list of blueprints should be mostly accurate. portdirect and i are sitting down this week to go through jira and what's on launchpad to verify the community representation is in order
15:52:07 <srwilkers> from that point, any chats regarding the community path forward for the project should take place in IRC/launchpad
15:52:32 <srwilkers> makes it easier for all of us when we can use launchpad for what it was meant to be used for :)
15:53:44 <srwilkers> any other comments/concerns?
15:54:01 <jayahn> not for the moment. :)
15:55:05 <srwilkers> cool, ill go ahead and end the meeting then and give you all some time back
15:55:10 <srwilkers> see you in openstack-helm
15:55:16 <srwilkers> #endmeeting