03:00:08 <hongbin> #startmeeting zun
03:00:09 <openstack> Meeting started Tue Dec  5 03:00:08 2017 UTC and is due to finish in 60 minutes.  The chair is hongbin. Information about MeetBot at http://wiki.debian.org/MeetBot.
03:00:10 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
03:00:13 <openstack> The meeting name has been set to 'zun'
03:00:14 <hongbin> #topic Roll Call
03:00:22 <kiennt26> o/ hi
03:00:52 <kevinz> o/
03:00:59 <Shunli> shunli
03:01:25 <hongbin> thanks for joining kiennt26 kevinz Shunli
03:01:55 <hongbin> ok, let's get started
03:02:05 <hongbin> #topic Announcements
03:02:14 <hongbin> i have an announcement
03:02:35 <hongbin> i will be on vacation starting from next week, until the end of this year
03:02:49 <Shunli> .........
03:02:54 <hongbin> mkrai will serve as a temp PTL while i am on vacation
03:03:18 <hongbin> any other announcement from our team member ?
03:03:22 <kevinz> hongbin: admire you :-)
03:03:29 <hongbin> kevinz: :)
03:03:33 <kevinz> hongbin: have a good vacation
03:03:41 <hongbin> kevinz: thank you
03:04:15 <hongbin> ok, next topic
03:04:18 <hongbin> #topic Container composition & k8s connector (kevinz)
03:04:23 <hongbin> kevinz: ^^
03:04:57 <kevinz> I'm working on investigation for k8s connector
03:05:20 <kevinz> some commit to my github
03:05:28 <kevinz> https://github.com/kevin-zhaoshuai/openstack-connector-k8s
03:05:50 <kevinz> Still need to do more about connect to K8s client
03:06:15 <kevinz> This week I will implement Volume support for capsule first
03:06:28 <hongbin> sure
03:06:59 <hongbin> btw, glad to see a good progress on the k8s connector
03:08:07 <hongbin> kevinz: any other update from your side ?
03:08:14 <kevinz> hongbin: Thanks, this features need to go through the create-pod -->create capsule process first
03:08:28 <kevinz> That's all from my side
03:08:44 <hongbin> kevinz: ack
03:08:53 <hongbin> thanks kevinz
03:09:15 <hongbin> all, any question / comments on this topic before moving to another topic ?
03:09:21 <kevinz> my pleasure
03:09:42 <hongbin> #topic Add support for clear container (mkrai)
03:09:58 <hongbin> for this one, i am helping madhuri to update the devstack patch
03:10:15 <hongbin> #link https://review.openstack.org/#/c/499534/
03:10:47 <hongbin> manage to the gate to sucess lask week
03:11:05 <hongbin> still need some testing on clear container runtime before removing the WIP
03:11:23 <hongbin> that is all on this one
03:11:28 <kevinz> hongbin: I've question
03:11:35 <hongbin> kevinz: go ahead
03:11:59 <kevinz> do we can use Clear container with zun now?
03:12:19 <kevinz> I'd like to try it in Arm server:-)
03:12:39 <hongbin> if cc is installed, we can use cc in zun
03:13:00 <hongbin> the key is using the --runtime option to specify cc as the runtime (default is runc)
03:13:34 <kevinz> This command in zun client?
03:13:41 <hongbin> kevinz: yes
03:13:41 <kevinz> I mean --runtime
03:13:51 <hongbin> kevinz: zun create --runtime ...
03:14:07 <kevinz> OK. I'll try
03:14:10 <kevinz> Thanks hongbin
03:14:28 <hongbin> kevinz: welcome
03:14:43 <hongbin> ok, move on to the next topic
03:14:45 <hongbin> #topic Introduce package manager for Zun application (hongbin)
03:14:51 <Shunli> hongbin, what should do on compute node for cc?
03:14:51 <hongbin> #link https://blueprints.launchpad.net/zun/+spec/zun-app-package-manager
03:15:13 <Shunli> the community docker will works for cc?
03:15:23 <hongbin> Shunli: yes, cc is a runtime in docker
03:15:35 <Shunli> ok, thx.
03:15:43 <Shunli> I like to try the cc.
03:15:51 <hongbin> Shunli: docker default runtime is runc, so if you don't do special configuration, zun will use runc
03:16:13 <hongbin> Shunli: the devstack patch is for configuring docker to enable cc as the alternative runtime
03:16:32 <Shunli> ack. thx hongbin.
03:16:39 <hongbin> no problem
03:16:54 <hongbin> ok, for the package management bp
03:17:07 <hongbin> i am writing the spec last week
03:17:23 <hongbin> i was trying to clarify the idea
03:17:33 <hongbin> in general, the following will be proposed
03:17:40 <hongbin> 1. a packaging format
03:17:47 <hongbin> 2. a package repository
03:18:00 <hongbin> 3. package management server
03:18:45 <hongbin> for #2, it will be the same as helm, we can use any http server to serve as the image repository, as long as it serves http
03:19:08 <hongbin> the package repository is mainly for serving a few tar file (the package) to download
03:19:26 <kevinz> it is maintained by user or upstream?
03:19:36 <hongbin> in openstack, we can use swift for this purpose, alternative is github pages, aws s3 ...
03:19:55 <hongbin> kevinz: we will have a devstack script to set that up
03:20:08 <kevinz> hongbin: ack
03:20:29 <hongbin> kevinz: it is maintained by the administrator in my first thought
03:21:53 <hongbin> for #3, the question from reviewers is how to store the data (i.e. package)
03:22:05 <kevinz> hongbin: OK, I just wonder it is a prebuilt or user built
03:22:12 <hongbin> the proposal is to store data in zun-api endpoint
03:23:45 <hongbin> kevinz: my primitive thought is to have everyone has privilege to upload their own package to their tenant, but only admin can publish package across tenants
03:24:21 <kevinz> hongbin: +1. good approach
03:25:01 <hongbin> however, still need to investigate the details and clarify it in the next patchset
03:25:23 <hongbin> i guess that is all from my side
03:25:57 <hongbin> appreciate reviews if you have time
03:26:04 <kevinz> Huge work I think:-)
03:26:17 <kevinz> I will review
03:26:52 <hongbin> yes, but i will try to get a minimum functionality working at the first iteration
03:27:24 <hongbin> and build up the whole things steps by steps
03:27:35 <hongbin> #topic Open Discussion
03:28:07 <hongbin> Shunli: kiennt26 : anything you want to discuss at the meeting ?
03:28:13 <kiennt26> hi
03:28:50 <kiennt26> Regarding patch 524045, it don't work as expect.
03:28:59 <kiennt26> #link https://review.openstack.org/#/c/524045/
03:29:29 <kiennt26> here is the log, it raises PCIdeviceNotfound exception
03:29:31 <kiennt26> #link http://logs.openstack.org/45/524045/3/check/zun-tempest-docker-etcd/452823b/logs/screen-zun-compute.txt.gz#_Dec_04_09_47_28_783378
03:30:16 <hongbin> i see
03:30:17 <kiennt26> I'm working on it to figure out what was wrong.
03:30:36 <hongbin> thanks for that kien
03:30:45 <Shunli> seems the update_pci_devices not work as expected.
03:31:28 <Shunli> pci device created in etcd in update_pci_device.
03:31:36 <hongbin> that is fine, the etcd gate was always failing in before
03:32:42 <hongbin> kiennt26: thanks for reporting that
03:33:18 <Shunli> thanks kien for working on that.
03:33:42 <kiennt26> hongbin, Shunli: yrw :) I'm working to make etcd gate green
03:34:07 <hongbin> great, thanks for that
03:34:12 <Shunli> great.
03:34:48 <hongbin> kiennt26: btw, thanks for the api version patch in tempest, that is great work!
03:35:20 <hongbin> it fixes a bunch of problem, and make us enable to write further test cases
03:35:54 <kiennt26> my pleasure o/
03:36:05 <hongbin> ok, all, anything else to discuss before closing the meeting ?
03:36:25 <kiennt26> About patch https://review.openstack.org/#/c/525036/ , what is your opinion? Should we have a note as my comment? Or simply use KURYR_CAPABILITY_SCOPE=global.
03:36:56 <hongbin> kiennt26: i don't have strong opinion for that
03:37:14 <hongbin> kiennt26: i will take the approach you prefer
03:37:57 <kiennt26> hongbin: yeah. How about you, Shunli , kevinz? WDYT?
03:38:32 <Shunli> watching the patch
03:40:47 <Shunli> i do not know the context, can not give opinion.
03:41:28 <hongbin> i think having a note is always good
03:42:17 <hongbin> it will get readers understand the this config better, so kiennt26 , i think your suggestion is good
03:42:30 <kiennt26> Ok, let's do this way. That's all from me.
03:42:38 <hongbin> thanks kiennt26
03:42:48 <kiennt26> hongbin: Have a good vacation
03:42:56 <hongbin> all, thanks for joining the meeting see you next time
03:42:58 <hongbin> kiennt26: thanks
03:43:04 <hongbin> #endmeeting