03:00:07 #startmeeting zun 03:00:08 Meeting started Tue Feb 14 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_2017-02-14_0300_UTC Today's agenda 03:00:17 #topic Roll Call 03:00:22 hello 03:00:25 Shubham 03:00:27 o/ 03:00:28 lakerzhou 03:00:31 Namrata 03:00:31 Madhuri Kumari 03:00:34 kevinz 03:00:59 thanks for joining the meeting pksingh shubhams sudipto lakerzhou Namrata mkrai kevinz 03:01:06 #topic Announcements 03:01:12 1. Select a mascot representing Zun 03:01:16 #link https://www.openstack.org/project-mascots/ 03:01:39 i was received an email from openstack foundation to ask the zun team to select a logo 03:02:11 basically, we select a animal or an item, someone will make a logo for zun team 03:02:31 What is the meaning of zun ? 03:02:33 I forgot 03:02:41 zun is a container 03:02:44 That is a chinese word, right? 03:02:50 I guess that means a container? 03:03:04 it seems to be an old chinese world 03:03:11 yes 03:03:32 #link https://en.wikipedia.org/wiki/Zun 03:03:32 hongbin By when we have to reply? 03:03:39 hi 03:03:48 mkrai: i guess one week or two 03:03:52 hi eliqiao 03:04:08 ok, any idea? 03:04:16 or want to discuss it offline? 03:04:28 hongbin: any specific theme for this that we have to follow? 03:04:36 Need to think 03:04:47 * eliqiao lurk 03:04:49 shubhams: i don't think there is any specific theme 03:04:57 ok, let me send an email for that 03:05:16 #action hongbin send a ML to discuss the team mascot 03:05:25 ok, next one 03:05:30 #topic Review Action Items 03:05:34 1. eliqiao_ create a bug for polling issue (DONE) 03:05:41 #link https://bugs.launchpad.net/zun/+bug/1662415 03:05:41 Launchpad bug 1662415 in Zun "Improve sandbox creation polling" [Medium,Triaged] 03:05:51 thanks eliqiao for creating the bug 03:06:00 maybe a barrel, :) 03:06:02 https://en.wikipedia.org/wiki/Barrel 03:06:08 i guess anyone can assign it to himself/herself if he/she want 03:06:15 Qiming: hey 03:06:22 o/ 03:06:28 interesting 03:06:46 Cool 03:06:49 #topic Cinder integration (diga) 03:06:55 #link https://blueprints.launchpad.net/zun/+spec/cinder-zun-integration The BP 03:06:59 #link https://review.openstack.org/#/c/417747/ The design spec 03:07:11 diga told me he won't be able to come today 03:07:22 he said he was working on this bp 03:07:32 he will submit the patches by thursday 03:07:43 that is all 03:07:47 any comment ? 03:08:22 ok, next one 03:08:27 #topic Kuryr integration (hongbin) 03:08:32 #link https://blueprints.launchpad.net/zun/+spec/kuryr-integration The BP 03:08:47 the spec is up for review for about one week or two 03:09:02 #link https://review.openstack.org/#/c/425883/ 03:09:16 hongbin, i reviewed it yesterday. Had a few concerns... 03:09:19 several people have provided feedback, and hte spec was revisioned a few times 03:09:31 sudipto: ack, will address your comments later 03:09:38 hongbin, sure. 03:10:12 i would ask everyone to review the spec again if you have a chance 03:10:22 will review it today 03:10:23 it seems it is close to merge soon 03:10:30 I will review 03:10:33 thanks 03:10:44 ok, next one 03:10:50 #topic Support interactive mode (kevinz) 03:10:55 #link https://blueprints.launchpad.net/zun/+spec/support-interactive-mode The BP 03:11:00 #link https://review.openstack.org/#/c/396841/ The design spec 03:11:06 kevinz: ^^ 03:11:46 Hi hongbin, https://review.openstack.org/#/c/417681/ this patch has been merged 03:12:09 I will add some test cases here and investigate the security problem 03:12:32 kevinz: ack 03:13:15 kevinz: i tried the attach command, it works very well 03:13:49 kevinz: i think the next step is to implement a run command that does attach automatically, like docker run -it 03:14:18 hongbin: Yeah Thanks , I will do it 03:14:31 kevinz: also, it seems the container will die after a attach and deattach 03:14:51 kevinz: it seems there is a way to keep the container alive after a de-attach 03:15:00 kevinz: those are good to support 03:15:16 OK I will find out the reason. 03:15:36 kevinz: yes, those are minor tweaks to follow up 03:15:47 kevinz: however, the work is amazing 03:16:01 kevinz: thanks for the great work, i knew it is not easy 03:16:23 kevinz: but you make it ! 03:16:34 hongbin: My pleasure 03:16:46 that is all from me 03:16:50 Thanks for your comments on that patch 03:16:58 all, any comment for kevinz ? 03:17:07 np 03:17:20 really great work !! 03:17:44 kevinz: i guess you can close the bp ? 03:17:50 Thanks for pksingh, your valuable advice for this bp 03:18:17 np 03:18:18 hongbin:OK I will close it after add zun run 03:18:24 kevinz: ack 03:18:57 ok, advance topic 03:19:05 #topic Introduce host capabilities and cpusets 03:19:10 #link https://review.openstack.org/#/c/427007/ The spec 03:19:17 sudipto: lakerzhou ^^ 03:19:30 I had minor comments about the NUMA support 03:20:04 I seemed to have replied to them. 03:21:01 I think it should be addressed in another feature. The current design cover one use case. It can be extended later it is necessary. 03:21:16 Do you any more concerns for now lakerzhou or can we take the comments during the code reviews? 03:21:37 no I don't have concerns. 03:21:53 yes, i agree that the current design is well covered for one use case 03:22:12 You're more than welcome to review the patches :) 03:22:13 we can always improve it later 03:22:27 hongbin: about host capabilities, I think we could support cache allocation later. 03:22:50 sudipto, I certainly well, thanks for the great work :) 03:22:53 yeah, thought of that. 03:22:55 eliqiao: could you elaborate? 03:23:36 eliqiao: perhaps sudipto can comment on your question 03:24:59 hongbin, this is something that eliqiao is driving in nova: https://blueprints.launchpad.net/nova/+spec/cat-support (i am guessing this is what it is?) 03:25:07 sudipto: yes. 03:25:34 i see 03:25:39 it looks like a good to have 03:25:53 sudipto: That's for VM support in nova, though libvirt. 03:26:20 eliqiao, but i think the l3-cache would apply to containers too... 03:26:22 hongbin: for container support, we could leverage other software, like intel-cat-cmt .. 03:26:39 is it supported by all the servers of intel? 03:26:58 sudipto: yes, only on xeon cpu. 03:27:15 in order to support this, we would have to do quite a bit of plumbing in the zun code, since we don't have libvirt to our helm 03:27:34 but yeah we could look at this maybe in the second iteration once we have all the pieces in? 03:27:45 sudipto: right, I will talk with my colleage for zun support. 03:28:10 they are working on an agent which will provide cache monitor/allocation. 03:28:18 later we can integrate it to zun. 03:28:46 eliqiao: ack 03:29:11 eliqiao: feel free to create a bp/spec if you want to lead this work 03:29:21 I will find my time to review host-capabilities in zun and to see if we can support cat in zun 03:29:27 hongbin: okay. thx. 03:30:08 ok, any other comment? 03:30:46 hongbin, should we merge the spec, if everyone is on the same page? 03:31:05 sudipto: i am ok with that (already gave +2) 03:31:52 it seems lakerzhou also agreed to merge it 03:31:55 i agree too 03:31:59 i think it is ready to go 03:32:32 Hongbin, I agree 03:33:05 ok 03:33:20 sudipto: any other comment you want to make? 03:33:37 hongbin, just would like to thank people who took the time out to review the spec. :) 03:33:57 :) 03:34:22 thanks sudipto for this work 03:34:33 ok, next topic 03:34:36 #topic Update image API to support multi-compute scenario (Wenzhi) 03:34:42 #link https://review.openstack.org/#/c/432857/ 03:35:01 it looks wenzhi is not here 03:35:11 however, he proposed an interesting patch 03:35:39 i have reviewed his patch, perhaps, i can explain it a bit 03:35:58 what he proposed is an API to schedule a container 03:36:23 this api basically write down the host to the container object in db 03:36:41 then, the image api can select host based on the 'host' field in the container object 03:37:04 hongbin, i went through the patch, thought i saw jenkins fail on that yesterday. 03:37:47 sudipto: it seems he uploaded another patchset, perhaps, it fixed the jenkins 03:38:01 hongbin, sure. 03:38:31 i wanted to discuss the idea here, but it seems it is better to leave it offline 03:38:47 then, everyone has a chance to review it 03:39:14 Yes I will review it and post my comments on the patch 03:39:22 mkrai: ack 03:39:32 ok, then let's enter open discussion 03:39:34 #topic Open Discussion 03:39:45 anyone has a topic to bring up? 03:40:03 Yes I have one 03:40:09 mkrai: go ahead 03:40:19 #link https://review.openstack.org/#/c/431442/ 03:40:34 this is support to add image-search API in zunclient 03:40:59 But seems the response object for image drivers, glance and docker are completely different 03:41:15 So we need to think of some common response object for the same 03:41:38 yes, possibly, we need to define an object for image 03:41:58 docker search returns 'stars', 'official' etc which glance doesn't have 03:42:30 mkrai: one way to handle that is to define a metadata field 03:42:57 hongbin: that includes driver specific fields? 03:43:07 mkrai: possibly 03:43:22 mkrai: another way to handle that is to simply hide those fields 03:43:56 hongbin: Do you think just the name and driver would be enough? 03:44:04 with metadata field 03:44:31 mkrai: personally, i would like to see more fields that are common for both docker and glance 03:44:52 i.e. size, location, etc. 03:45:18 hongbin: docker search don't show size and all 03:45:24 location is important IMHO 03:45:35 only common field is 'name' I guess 03:45:38 IIUC, that should tell you if it's a public registry or a private one. 03:46:07 NAME DESCRIPTION STARS OFFICIAL AUTOMATED 03:46:19 These are the fields for docker 03:46:22 i see 03:46:24 but i thought there was debate around that in glance, whether the location should be admin only. 03:46:41 sudipto: ack 03:47:22 mkrai: then, i am ok to start with name and driver 03:47:33 I will try to create a common response object and post it on review link 03:47:51 You all can post your suggestion on that 03:48:03 hongbin: ack 03:48:16 thanks mkrai 03:48:20 hongbin: sudipto Thanks 03:48:52 hongbin:I am working on container resource.the patch is ready I am testing it. 03:48:52 That's all from my side. 03:49:02 will upload it today 03:49:27 Namrata: awesome 03:50:03 Namrata: did the heat team give you enough comment for this feature? 03:50:55 i havent joined the team meeting 03:51:07 i will do this week and ask for reviewing the patches 03:51:22 Namrata: that is fine, i will help you to ping them 03:51:32 sure .thanks hongbin 03:52:18 thanks Namrata for working on this feature in heat 03:52:34 hongbin:thanks for sugesting the same 03:53:16 ok, it seems there is no more topic to bring up 03:53:37 then, i will end the meeting a bit earlier, all , thanks for joining the meeting 03:53:45 #endmeeting