03:00:07 #startmeeting zun 03:00:08 Meeting started Tue Sep 5 03:00:07 2017 UTC and is due to finish in 60 minutes. The chair is hongbin. Information about MeetBot at http://wiki.debian.org/MeetBot. 03:00:09 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 03:00:12 The meeting name has been set to 'zun' 03:00:13 #link https://wiki.openstack.org/wiki/Zun#Agenda_for_2017-09-05_0300_UTC Today's agenda 03:00:19 #topic Roll Call 03:00:37 o/ 03:00:54 fengshengqin 03:01:05 o/ 03:01:08 0/ 03:01:47 thanks for joining the meeting kevinz FengShengqin spn Shunli 03:01:54 let's get started 03:01:58 #topic Announcements 03:02:04 1. New BPs for Queens cycle 03:02:09 #link https://blueprints.launchpad.net/zun/+spec/zun-policy-in-code Register and Document Policy in Code 03:02:15 #link https://blueprints.launchpad.net/zun/+spec/zun-split-tempest-plugins Split Tempest Plugins into Seperate Repos/Projects 03:02:26 these two bps are openstack community goal 03:02:37 welcome to pick it if you interest 03:03:12 o/ sorry, i'm late. 03:03:18 any other announcement from our team members? 03:03:28 hi kiennt , np, thanks for joining 03:03:56 ok, move forward 03:03:58 #topic Cinder integration (hongbin) 03:04:18 for this one, i was working on several patches last week 03:04:50 #link https://review.openstack.org/#/q/status:open+project:openstack/zun+branch:master+topic:bp/direct-cinder-integration 03:05:14 there are a few patches that are ready for review 03:05:36 there are a few WIP patches, which is mainly for missing unit tests 03:05:47 i will continue to work on those 03:06:07 with all the patches above, the cinder integration feature is ready 03:06:25 that is all from me 03:06:30 any question for this topic? 03:06:33 we still need prepare cinder volume for container? 03:06:52 Thanks hongbin! This is great feature to zun 03:06:52 FengShengqin: yes, users need to create a cinder volume in before 03:07:09 FengShengqin: which is a "cinder create ...." command 03:07:30 kevinz: welcome 03:08:04 any other question? 03:08:25 #topic Introduce container composition (kevinz) 03:08:30 #link https://blueprints.launchpad.net/zun/+spec/introduce-compose 03:08:34 kevinz: ^^ 03:09:22 kevinz: there? 03:09:24 Hi all, actually I didn't get much time last week due to Arm offsite meeting. I will work on a doc about capsule this week 03:09:34 hongbin: yeah 03:09:40 kevinz: :) 03:09:55 kevinz: get that 03:10:08 hongbin: And rebase some code to latest commit. Then I think the basic function of capsule will finish 03:10:23 kevinz: awesome 03:10:45 https://review.openstack.org/480455 and this patch need review 03:10:45 kevinz: i am looking forward to it, the k8s connector bp will leverage the capsule feature 03:11:08 hongbin: That's great 03:11:28 core reviewers, pls help to review this patch 03:11:30 #link https://review.openstack.org/480455 03:11:39 thanks kevinz 03:12:02 That's all from me. Next week I will go to Denvor for PTG 03:12:14 kevinz: have a good trip 03:12:24 kevinz, Have a great trip :) 03:12:35 hongbin: spn: Thanks! 03:12:46 #topic Add support for clear container (mkrai) 03:12:54 mkrai: hi madhuri, you there? 03:13:00 mkrai is not well 03:13:02 and she is out 03:13:08 sorry to hear 03:13:13 But last week she was available for a brief time 03:13:25 and we discussed some of the specific of the implementation 03:13:35 https://etherpad.openstack.org/p/zun-clearcontainers 03:13:43 I captured our discussion over there 03:13:49 #link https://etherpad.openstack.org/p/zun-clearcontainers 03:14:42 * hongbin is reading the etherpad 03:15:02 Kevinz, I am going to PTG too. We should meet there. 03:15:31 lakerzhou1: Wow ,that's great! 03:15:52 lakerzhou1: hope to meet you there! 03:16:56 spn: the outline in the etherpad looks good to me 03:17:11 spn: thanks for working on that 03:17:27 hongbin, thank you. I trying to add sort of diagram just to clear my mind on how it all works 03:17:50 we will sit down on implementation specifics this week and add notes over there. 03:18:03 thanks spn 03:18:05 spn: so the basic direction is that cc and runc will on different host, right? 03:18:18 Shunli, yes 03:18:25 thx, ack 03:18:30 what was your discussion 03:18:42 earlier about running zun containers and kvm vms on same machine? 03:18:56 zun containers with regular runc engine 03:19:50 ok, i think we should move forward 03:20:01 #topic NFV use cases (lakerzhou) 03:20:03 no, just to confirm that i understand the etherpad correctly. 03:20:07 #link https://etherpad.openstack.org/p/zun-nfv-use-cases 03:20:30 for this topic, there are a few progress last week 03:20:47 Shunli has uploaded a few patches for the pci support in db/objects 03:21:11 https://review.openstack.org/#/c/500418/ 03:21:27 yes, this is the big patch 03:21:34 working on the comments and the ut this week. 03:22:00 Shunli: ack, thanks for working on that 03:22:35 that is for the pci scheduling support 03:23:03 in addition, i am working on a few patches in kuryr/kuyr-libnetwork for the sriov support 03:23:25 #link https://review.openstack.org/#/c/500420/ 03:23:31 #link https://review.openstack.org/#/c/500436/ 03:23:48 later maybe will start the pci tracker as the base code need by pci tracker is almost ready. 03:24:22 yes, what is missing is the pci tracker and scheduler filter 03:24:31 pci scheduler filter in particular 03:25:04 after that, we will look for an integration between zun and kuryr 03:25:16 sure 03:25:47 i believe the integration is not hard, zun just needs to populate the neutron port with information about which VF to bind 03:26:09 it's seems a long way to go. pci feature is too complicate and too much code,:-) 03:26:23 yes :) 03:26:54 Shunli: i saw you ported a lot of code from nova to zun already 03:27:15 which is great 03:27:44 Shunli: thanks for your hard work on this 03:27:54 np my pleasure. 03:27:59 all, any remark on this topic? 03:28:41 for demo, pci scheduler can be optional if it is too complicated 03:29:28 lakerzhou1: not just for scheduler, zun need some basic support code for zun need to port from nova. 03:30:08 yes, it looks a lot of codes are for tracking the pci resources in all nodes 03:30:14 such as the db to store pci device, compute node to collect pci device, etc.. 03:30:41 shunli, yes, understood. 03:31:11 lakerzhou1: eventually, zun need to pick a VF, and populate the binding:profile in the neutron port 03:31:35 lakerzhou1: then, kuryr will pick up the port and read information from binding:profile 03:32:11 lakerzhou1: then, kuryr will know which VF Zun has picked, and bind the interface of the VF 03:32:58 lakerzhou1: that is how it should work in theory :) 03:33:28 hongbin, yes, that is correct 03:33:36 ok 03:33:49 all, any other remark on this topic? 03:34:25 #topic Mutli-host problems (kiennt) 03:34:25 I have located three physical servers, but it is not easy to get VPN access 03:34:50 lakerzhou1: we can sync with that offline 03:35:12 kiennt: there? 03:35:29 hongbin, sure, I will sync up with you tomorrow 03:35:41 hi 03:35:50 lakerzhou1: ack 03:36:13 kiennt: hey, do you have anything to bring up about this topic? 03:36:59 hongbin: About the gate, the fix with docker cluster store ip fix the prev problem 03:37:31 kiennt_: ack 03:37:40 but, as you know, multi test still failed. I will work on this to figure out what was wrong. 03:38:28 that's all from me, will work on it this week. 03:38:35 kiennt_: ack 03:38:48 kiennt_: thanks for working on that 03:39:06 #topic Open Discussion 03:39:07 hongbin: my pleasure :) 03:39:27 all, any topic to bring up for discussion? 03:39:39 none from me 03:39:59 none from me 03:40:42 ok, all, thanks for joining the meeting, see you next week 03:40:47 #endmeeting