21:00:32 <strigazi> #startmeeting containers
21:00:33 <openstack> Meeting started Tue Jan 22 21:00:32 2019 UTC and is due to finish in 60 minutes.  The chair is strigazi. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:34 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:36 <openstack> The meeting name has been set to 'containers'
21:00:39 <strigazi> #topic Roll Call
21:00:44 <strigazi> o/
21:01:34 <strigazi> ping eandersson lxkong colby_ brtknr
21:01:41 <jakeyip> o/
21:01:44 <lxkong> o/
21:02:00 <eandersson> o/
21:02:09 <colin-> imdigitaljim otw
21:02:32 <strigazi> #topic Announcements
21:03:02 <strigazi> thanks to eandersson and jakeyip, we have the ci runnig with python3 by default.
21:03:12 <eandersson> \o/
21:03:19 <colin-> nice
21:03:35 <strigazi> #topic Stories/Tasks
21:04:32 <strigazi> From my side, I have only the patch to add tiller for usage within magnum, not for end users unless they really try to use it:
21:04:35 <strigazi> https://review.openstack.org/#/c/612336/
21:05:15 <strigazi> It deploys tiller only on master nodes, in a separate namespaces, with tls, and access to kube-system.
21:05:51 <strigazi> To better understand how it is deployed take a look in this section:
21:07:17 <strigazi> https://docs.helm.sh/using_helm/#example-deploy-tiller-in-a-namespace-restricted-to-deploying-resources-in-another-namespace
21:07:54 <strigazi> and this one: https://docs.helm.sh/using_helm/#using-ssl-between-helm-and-tiller
21:08:33 <strigazi> My colleague Diogo is working on consuming it to deploy the metrics server which deprecates heapster and the prometheus operator
21:09:01 <strigazi> https://review.openstack.org/#/c/629577/
21:09:27 <colin-> have you been using prometheus much internally? curious how you find it. obviously it's wildly popular and has a lot to offer
21:09:35 <strigazi> If you are not familiar with k8s, the patch for tiller might seem complicated.
21:09:39 <colin-> i've had positive experiences with it in the past
21:10:12 <strigazi> colin-: limited experience so far without any issue but also no big challenges
21:10:40 <imdigitaljim> sorry im late! had workstation reimaged and had to reinstall IRC
21:10:56 <strigazi> colin-: we work on a solution that an external service (probably another central prometheus) will consume the cluster prometheuses
21:11:16 <strigazi> imdigitaljim: welcome
21:11:31 <strigazi> colin-: makes sense?
21:11:33 <colin-> understood. when it starts to take shape i am curious what KPIs from the clusters themselves will be collected
21:12:48 <lxkong> hi guys, i have been working on octavia-ingress-controller integration(https://review.openstack.org/#/c/631330/) and fixing a bug that deletes fip of the load balancer vip in octavia.
21:13:00 <strigazi> colin-: cpu/ram/disk/network and when use cases become more serious, user should expose user-defined metrics
21:14:50 <openstackgerrit> Lingxian Kong proposed openstack/magnum master: [k8s_fedora_atomic] Delete floating ip for load balancer  https://review.openstack.org/630820
21:14:58 <strigazi> lxkong: I will take a look to ingress too
21:15:31 <lxkong> strigazi: thanks, i will write some doc this morning in that patch
21:17:18 <strigazi> lxkong: thanks
21:17:32 <cbrumm_> lxkong - We maybe looking at that soon also
21:17:52 <lxkong> cbrumm_: sounds great :-)
21:18:24 <lxkong> cbrumm_: if you have any question relating to the octavia-ingress-controller, please let me know
21:18:37 <strigazi> FYI, at CERN we will write out own "ingress" to complement traefik
21:19:12 <strigazi> s/out/our/
21:19:23 <cbrumm_> lxkong - Are you sure you want that? colin- can be persistent
21:19:49 <lxkong> strigazi: as more and more add-on components adding to magnum, i think we should find a plugable way for the implementation
21:20:34 <lxkong> so give the users the option that they could do their own thing if it's not acceptable for the upstream
21:20:52 <lxkong> cbrumm_ :-)
21:20:52 <strigazi> lxkong: helm can help :)
21:21:36 <lxkong> strigazi: i mean, deploy all the things during the cluster creation
21:21:59 <lxkong> rather than leaving the setup things to the magnum user
21:22:05 <strigazi> we have a cern chart for our internal components, so we can just point to that chart to deploy at creation time
21:22:17 <lxkong> that's nice
21:22:29 <strigazi> not for the user. for us
21:23:21 <strigazi> you can deploy this in ingress or anything else, packaged in a <corporate> downstream chart
21:23:31 <strigazi> you can deploy this ingress or anything else, packaged in a <corporate> downstream chart
21:23:59 <imdigitaljim> lxkong: were working towards a complete driver that has high customizability, low operational costs, extra features, and improvements running on centos
21:24:01 <lxkong> is the helm/charts mechanism supported in magnum aleady?
21:24:22 <imdigitaljim> looking to upstream it when we get to a satisfiable point
21:24:37 <strigazi> lxkong: 612336 629577
21:24:41 <lxkong> strigazi: thanks
21:24:44 <lxkong> will take a look
21:24:57 <strigazi> adding a hook for this is very easy
21:27:49 <strigazi> I think the approach with a chart maintained in its repo to add extra components is very maintainable
21:28:15 <strigazi> Or even a list of components deployed *on* kubernetes
21:28:23 <lxkong> strigazi: agree
21:29:08 <lxkong> strigazi: but can i stick with the current octavia-ingress-controller patch and maybe use the helm way for the future add-on?
21:29:16 <strigazi> sure
21:29:19 <lxkong> cool
21:32:01 <strigazi> Anything else you want to bring up?
21:32:19 <lxkong> not from me
21:32:32 <eandersson> I got a couple of follow up patches to the py3 patches, but nothing critical. https://review.openstack.org/#/c/631083/ https://review.openstack.org/#/c/631331/
21:33:11 <strigazi> +2
21:33:55 <eandersson> Ideally I would like to move to dump_as_bytes for everything, but one step at a time.
21:34:02 <eandersson> thanks strigazi
21:35:02 <strigazi> eandersson: https://review.openstack.org/#/c/631331/10/magnum/drivers/common/templates/swarm/fragments/make-cert.py
21:35:33 <eandersson> It was original on purpose as I changed import json
21:35:41 <eandersson> but can revert that change to make it cleaner
21:36:20 <eandersson> (I prefer two spaces but it's super nit)
21:36:26 <eandersson> *new lines
21:36:28 <strigazi> you can leave it
21:36:51 <colby_> strigazi: yes sorry, I was out of office till today. I just spun up a cluster with that tag and will try again with cinder
21:37:46 <strigazi> colby_: brtknr is bringing queens really close the current state. but with 1.11.5-1, it shoult work.
21:40:58 <strigazi> If there is anything else to discuss we can end the meeting
21:41:08 <strigazi> objections?
21:41:37 <colby_> strigazi: ok great. Will that backport he linked to me be merged into a future release? I saw 6.3.0 just dropped.
21:42:18 <strigazi> colby_: yes
21:44:07 <colby_> thanks for responding back!
21:44:18 <colby_> Ill let you know how my tests go
21:44:51 <strigazi> cool
21:47:31 <strigazi> It seems we are out of topics, thanks for joining everyone
21:47:41 <strigazi> see you next week
21:47:44 <eandersson> Thanks!
21:47:52 <lxkong> have a good day/night, everyone
21:48:05 <strigazi> #endmeeting