15:01:50 #startmeeting interop_challenge 15:01:51 Meeting started Wed Jan 18 15:01:50 2017 UTC and is due to finish in 60 minutes. The chair is topol. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:52 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:01:54 The meeting name has been set to 'interop_challenge' 15:01:56 topol: startmeeting interop_challenge 15:01:56 09:00 topol: Hi everyone, who is here for the interop challenge meeting today? 15:01:56 09:00 topol: The agenda for today can be found at: 15:01:58 09:00 topol: # https://etherpad.openstack.org/p/interop-challenge-meeting-2017-01-18 15:02:00 09:00 topol: We can use this same etherpad to take notes 15:02:02 09:00 dmellado: o/ 15:02:04 09:00 dmellado: hi topol 15:02:06 09:00 markvoelker: o/ 15:02:08 09:00 tongli: o/ 15:02:10 09:00 hogepodge: o/ 15:02:12 09:00 Rocky_g: o/ 15:02:14 09:01 topol: #topic Review last meeting action items 15:02:16 09:01 topol: #link http://eavesdrop.openstack.org/meetings/interop_challenge/2017/interop_challenge.2017-01-11-14.01.html 15:02:18 09:01 skazi: o/ 15:02:20 09:02 topol: tongli to check on eeiden about NVF workload 15:02:22 09:02 zhipengh: o/ 15:02:24 09:02 topol: tongli were you able to get a hold of eeiden? 15:02:26 09:03 tongli: @topol, brad, that was not done but I have a few more people who will be willing to work on NFV. a writeup about the workload will be done very quickly. 15:02:28 09:03 topol: tongli, that works :-) 15:02:30 09:03 tongli: I will have some details in the agenda. 15:02:32 09:03 topol: K, I put you on the agenda. you will see where 15:02:34 09:04 topol: zhipeng to add overview section to lessons learned doc 15:02:36 09:04 gema has joined (~Gema@host31-48-176-122.range31-48.btcentralplus.com) 15:02:38 09:04 tongli: when we go through the agenda, I can talk about them. 15:02:40 09:04 topol: this was done 15:02:42 09:04 topol: zhipeng pushed a fitst patch for folks to review 15:02:44 09:04 zhipengh: and updated with a second one :) 15:02:46 09:04 tongli: @topol. I feel the patch is a bit overlap with the conclusion section of the doc. 15:02:48 09:05 topol: #link https://review.openstack.org/#/c/421275/ 15:02:50 09:05 topol: a little overlap is okay. I'll re-review the whole doc 15:02:52 09:05 tongli: also it stated that ansible + shade is the recommended approach, do we want to say that? 15:02:54 09:06 tongli: I am not really sure about it. but we are clearly doing it. 15:02:56 09:06 topol: well nsible + shade is what we had the most success with across all the cloud environments 15:02:58 09:06 tongli: ok. 15:03:00 09:06 topol: That I think is a fair statement 15:03:02 09:06 topol: so in aany case folks please review this patch 15:03:04 09:06 tongli: sure. 15:03:06 09:07 markvoelker: I think it's fair to say we found it to be a good choice, but I'd be careful to word it such that we don't look like we're saying we did an exhaustive evaluation of many options 15:03:08 09:07 topol: tongli to create an etherpad for PTG topics real fast so we can put the link in this meeting <-- This was completed and on our agenda to review 15:03:10 09:07 dmellado: +1, I'd say that it's an approach that worked fine for us so far 15:03:12 09:07 topol: markvoelker +++ I agree, please make a comment on the pacth with that suggestion 15:03:14 09:08 topol: dmellado feel free to comment on the patch as well 15:03:16 09:08 tongli: @markvoelker, @dmellado, you guys said what I could not say it well. 15:03:18 09:08 zhipengh: markvoelker agree 15:03:20 09:09 topol: #topic Patches that need review #link https://review.openstack.org/#/q/status:open+project:openstack/interop-workloads,n,z 15:03:22 09:09 tongli: I did not want to make that strong a statement. basically that is the reason I did not +1. 15:03:24 09:09 topol: we have three patches for folks to review 15:03:26 09:09 daniela_ebert has joined (c2191e09@gateway/web/freenode/ip.194.25.30.9) 15:03:28 09:09 topol: so if you have some time please review. Great to see folks making updates! 15:03:30 09:10 topol: #topic Update from tongli on interlock on interop with OpenStack China companies 15:03:32 09:10 topol: tongli, update us please. whats happening in Beijing 15:03:34 09:10 garloff has joined (~kgarloff@160.44.193.111) 15:03:36 09:11 tongli: there were total 10 parties/companies attended their interop challenge wg kick off meeting. 15:03:38 09:11 tongli: I have listed these companies/parties in the etherpad. 15:03:40 09:11 tongli: some I do not have their English names. 15:03:42 09:12 tongli: please take a look at the etherpad. 15:03:44 09:12 tongli: I gave an introduction session talked about our effort and what we have done and what we have planed to do. 15:03:46 09:13 tongli: guys from ZTE, Huawei, China Mobile will start working on NFV workload. 15:03:48 09:13 topol: list of companies ooks strong 15:03:50 09:13 topol: tongli very nice to see folks start on NFV workload! 15:03:52 09:13 tongli: on the second day of the meeting, I went over the LAMPStack workload, and talked about the structure of the new repo. 15:03:54 09:14 tongli: how the new workload should look like and how to deal with different targets etc. 15:03:56 09:14 tongli: these companies have started running LAMPStack on their cloud, few have run the workload successfully but most of them have various issues. 15:03:58 09:15 topol: tongli can any of the folks attend our irc interop meeting? is the time too late for them? 15:04:00 09:15 zhipengh: for the second round, should we provide use case descriptions for each workload to be tested ? 15:04:02 09:15 tongli: two companies will also have resources to help with Kubernetes workload 15:04:04 09:15 tongli: some of them will attend this IRC meeting, 15:04:06 09:16 topol: tongli excellent 15:04:08 09:16 garloff: It would seem reasonable for them to pass LAMP first before moving to more advanced workloads, no? 15:04:10 09:16 tongli: they want to use our time Tuesday night 10:00pm at this channel to do their own meeting since the meeting will be conducted in Chinese. 15:04:12 09:16 topol: garloff, if they have enough resources theymight be able to both in parallel 15:04:14 09:16 garloff: ok 15:04:16 09:16 tongli: @garloff, yes, they are doing exactly that. 15:04:18 09:17 topol: they are all doing lampstack first? 15:04:20 09:17 tongli: they also want to develop some workloads from their own use cases. but that will come later. 15:04:22 09:17 garloff: My worry otherwise would be is that we test interop at container level where possibly the lower layers (OpenStack) are not interoperable at all ... 15:04:24 09:17 tongli: but all agreed to put their workload in the same repo. 15:04:26 09:17 tongli: so we will have to review more workloads. 15:04:28 09:18 topol: tongli, so putting the workloads in the same repo is key. Ideally we end up with a broad set of workloads for demonstrating interop 15:04:30 09:18 topol: tongli, reviewing more workloads is a good thing! 15:04:32 09:18 tongli: @topol, I am hoping that they can officially use this channel at our time Tuesday night 10:00pm. 15:04:34 Would we be interested in having both workloads if we can get ours up soon? 15:04:34 09:19 tongli: and their conversations will be also recorded just like ours. 15:04:36 09:19 topol: tongli, dont they just have to reserve the slot? 15:04:38 09:19 tongli: everything (meetings, workloads) will be all at the same place. 15:04:40 09:19 tongli: yes, I do not know how to reserve that. 15:04:42 09:20 topol: tongli shamail knows how to do this 15:04:44 09:20 topol: I believe 15:04:46 09:20 tongli: ok, let me ask him to help. 15:04:48 09:20 topol: tongli thanks 15:04:50 09:20 tongli: or he can point me to the right place. 15:04:52 09:20 tongli: yes, please add an action for me. 15:04:54 09:20 garloff will not try to read a conversation recorded in chinese though 15:04:56 09:21 topol: #action tongli to work with shamail to reserve evening timeslot on meeting5 for china interop 15:04:58 09:21 tongli: @garloff, some of the conversation probbably will be in English. 15:05:00 09:21 kbaikov: o/ 15:05:02 09:21 topol: garloff I will struggle with that as well 15:05:04 09:21 tongli: they were a bit afraid using English will take them 3 times longer to do the same thing in Chinese. 15:05:06 09:21 topol: tongli, this is excellent progress. Any more updates on the china company interop initiative? 15:05:08 09:22 tongli: they are excited and ready to contribute, i am very happy about that. 15:05:10 09:22 topol: tongli thats fine, perhaps a meetign sumamr or action items can be added in english? 15:05:12 09:22 tongli: they will have a summit in April and Interop show off is part of it. 15:05:14 09:23 tongli: I think they have allocated 10 minutes for a demo. 15:05:16 09:23 topol: tongli you will attend their meeting correct? So you can keep us updated? 15:05:18 09:23 tongli: most likely they will use LAMPStack, maybe a new one. 15:05:20 09:23 tongli: I will try to at least attend first few ones. 15:05:22 09:23 topol: tongli ok 15:05:24 09:24 tongli: but they will be on Chinese New Year next week for about two weeks. 15:05:26 09:24 topol: tongli or at least read their meeting minutes? 15:05:28 09:24 tongli: so their first will be in Feb. 15:05:30 09:24 tongli: @topol, I will attend. 15:05:32 09:24 tongli: and report back next day in our meeting. 15:05:34 09:25 topol: how many folks do we have who speak chinese? You cdiep, zhipeng? perhaps you can split the translation duties? 15:05:36 09:25 tongli: sure. that will be nice. 15:05:38 09:25 tongli: or use google translate? 15:05:39 o/ 15:05:40 09:25 topol: I know catherine (cdiep) speaks chinese 15:05:42 09:26 gema: google translate won't work for chinese, I don't think 15:05:44 09:26 gema: plus what do you guys want to translate? 15:05:46 09:26 tongli: @gema, it does. 15:05:48 09:26 gema: surely just the main points? 15:05:50 09:26 topol: gema+++ Just need a sumamry or main points or issues 15:05:52 09:26 tongli: haha, yeah, I can help, no worries. 15:05:54 09:26 gema: tongli: great news :D 15:05:56 09:26 topol: ok, great progress, tongli is that all on the china meetings? 15:05:58 09:27 dmellado: please, chinese it's not my greatest skill 15:06:00 09:27 dmellado: xD 15:06:02 09:27 topol me either 15:06:04 09:27 Rocky_g: No, it works kind of ok.the ordering of clauses can get weird, but lots of folks keep it simple on irc, so that helps 15:06:06 09:27 tongli: @topol, that is all. 15:06:08 09:27 topol: #topic Check on possible day/time for our half day session at PTG 15:06:10 09:28 topol: markvoelker, any clarity on when we might squeeze the half day session in? 15:06:12 09:28 tongli: @topol. I have put the china trip update on the etherpad before this meeting. you can also check it out if you like 15:06:14 09:28 markvoelker: I have an email out to catherine to start banging on the schedule--hopefully we'll get that settled in the next few days. 15:06:16 09:29 topol: also I was hoping I could fly in monday morning (one less day away from the family). So if its not first thing monday morning I wont cry 15:06:18 09:29 topol: markvoelker, that sounds good 15:06:20 09:30 topol: markvoelker, our flight from rdu is only like an hour. YAY! 15:06:22 09:30 markvoelker: =) 15:06:24 09:30 Rocky_g: Topol, just don't snow ;-) 15:06:26 09:30 topol guessing that markvoelker could use coming in monday morning too 15:06:28 09:31 dmellado: I'd try my best to squeeze there in too, I'll be attending some another sessions too 15:06:30 09:31 topol: Rocky_g snow is an evil word around here 15:06:32 09:31 markvoelker is in face flying in on Monday morning 15:06:34 09:31 dmellado: (also let's see how the jet lag treats me after my 10 hour flight xD) 15:06:36 09:31 markvoelker: *fact 15:06:38 09:31 Rocky_g: Oops. I meant snore. Sorry to swear. 15:06:40 09:31 topol: markvoelker what time? 15:06:42 09:32 tongli: I have not booked the flight yet. 15:06:44 09:32 markvoelker: topol: I land at 7:30am in ATL. Should be at the PTG shortly thereafter. 15:06:46 09:32 topol: I have not booked flight yet either but will soon 15:06:48 09:32 Rocky_g: Ouch 15:06:50 09:33 topol: Rocky_g I agree, that seems a little early 15:06:52 09:33 topol: #topic Review current PTG Agenda #link https://etherpad.openstack.org/p/interop-challenge-meeting-2017-01-11 15:06:54 09:33 gema: markvoelker: you are too efficient 15:06:56 09:33 topol: any updates? 15:06:58 09:34 topol: gema, efficient would be he leaves at a time where he can take the kids to school monday morning then head to the airport. Im channeling Mark's wife :-) 15:07:00 09:34 gema: :) 15:07:02 09:34 tongli: haha. 15:07:04 09:34 topol: gema, at least thats my plan 15:07:06 09:35 topol: so folks take a look at the ptg agenda. any updates? Additions? 15:07:08 09:36 gema: topol: do we need to spend time of the morning organizing a dinner? 15:07:10 09:36 tongli: Boston activities? 15:07:12 09:36 gema: topol: cannot we do that in advance? 15:07:14 09:36 tongli: what we should do as a group for Boston summit? 15:07:16 09:36 topol: gema, as far as Im concerned either way. a dinner should not take long to organize 15:07:18 09:37 tongli: @topol, should not take more than 2 minutes. I do put there so that we know we will have a dinner, no? 15:07:20 09:37 Rocky_g: ++ 15:07:22 09:37 gema: ack 15:07:24 09:37 topol: lets see how many people we are. did folks add their name on the etherpad so we know who will be there 15:07:26 09:37 gema: NFV is behind organising a dinner 15:07:28 09:37 eeiden: A quick NFV note since it's on the agenda -- OPNFV had a 2 week delay due to lab issues, so we're still trying to verify that we got the workload fixed. I'll keep you all posted as it progresses, but we're working to get it up and in the repo asap. 15:07:30 09:37 gema: that's how I read it xD 15:07:32 09:38 tongli: @gema, that list is not the priority list. :-) 15:07:34 09:38 topol: eeiden did you see we have some companies in china interested in adding an nfv workload as well 15:07:36 09:38 Rocky_g: Uh, Gigli, when was the last time you tried to get even ten engineers to are on food choices? 15:07:38 09:38 gema: tongli: great, thanks for clarifying 15:07:40 09:38 eeiden: ooh no I did not! 15:07:42 09:38 Rocky_g: Gigli should be tongli. Dang phone auto correct 15:07:44 09:38 topol: eeiden connect with tongli for details 15:07:46 09:39 eeiden: Thanks! 15:07:48 09:39 topol: lots of restaurant in ATL. we'll find somehting great Im sure 15:07:50 09:39 tongli: @eeiden, look at my updates on the etherpad can also help. 15:07:52 09:40 topol: #action all, please review and update PTG agenda with topics #link https://etherpad.openstack.org/p/interop-challenge-meeting-2017-01-11 15:07:54 09:40 topol: #topic Open discussion 15:07:56 09:40 topol: any other topics for today? 15:07:58 09:41 tongli: Brad, I added an item for discussion. 15:08:00 09:41 tongli: I have started working on Kubernetes workload. 15:08:02 09:41 topol: tongli go ahead 15:08:04 09:41 tongli: got one question though, 15:08:06 09:41 zhipengh: @tongli could you first provide a spec describing the use case for k8s workload ? 15:08:08 09:41 tongli: compile kubernetes normally will take about 1 hour. 15:08:10 09:42 tongli: so that is not really good. 15:08:12 09:42 topol: tongli, my compile takes about 18 minutes. Are you generating a release as well? 15:08:14 09:42 tongli: so I have decided to use the prebuilt kubernetes 15:08:16 09:42 topol: tongli, prebuilt seems okay to me 15:08:18 09:42 tongli: you can download from here. 15:08:20 09:43 topol: tongli, you running in a VM? 15:08:22 09:43 tongli: hmmm. mine took way longer, I am not sure even 18 minutes is good enough. 15:08:24 09:43 tongli: yes. running in a VM. 15:08:26 09:43 tongli: it has to. 15:08:28 09:43 topol: tongli how much memory, cpus and diskspace did you allocate? we should compare envs 15:08:30 09:44 topol: in any case prebuilt I think is fine 15:08:32 09:45 topol: Im guessing most folks would appreciate the stability of not compiling Kube everytime 15:08:34 09:45 topol: Is everyone ok with prebuilt Kube? 15:08:36 09:45 hogepodge: Magnum can have a kube cluster up in minutes 15:08:38 09:45 tongli: if we do compile , will it buy us anything ? 15:08:40 09:45 tongli: the VM I have has 8 GB of memory, 10 GB diskspace. 15:08:42 09:45 tongli: used 4 GB, actually the compiler run out memory and died. 15:08:44 09:45 tongli: ok, if no one object, I will use the prebuilt kubernetes. 15:08:46 09:45 tongli: use prebuilt image? 15:08:48 09:46 topol: tongli Ha Ha Ha. I did that too. Also didnt provide enough diskspace. An even more painful mistake 15:08:50 09:46 tongli: @hogepodge, if not use prebuilt image, I would really like to know how they did it. 15:08:52 09:46 topol: 8GB memory and 80GB disk space is my rule 15:08:54 09:46 topol: tongli, as ton ngo 15:08:56 09:46 topol: he is magnum core 15:08:58 09:46 hogepodge: They use prebuilt images. 15:09:00 09:47 tongli: @topol, I have already talked to him. 15:09:02 09:47 topol: hogepodge +++ 15:09:04 09:47 tongli: he said it normally will take about 30 minutes to 1 hour to build it 15:09:06 09:47 topol: tongli sounds about right 15:09:08 09:47 tongli: he suggested to use the binary and gave me the url 15:09:10 09:47 topol: tongli, I think using the binary makes a lot of sense 15:09:12 09:48 tongli: @hogepodge, I would love to know how Magnum does it within minutes. 15:09:14 09:48 topol: tongli, my guess is they have it on an image and provision the image 15:09:16 09:48 topol: cooking show, cake is already baked :-) 15:09:18 09:49 tongli: if yes, I do not know if we want to go that route. 15:09:20 09:49 hogepodge: They use an image like coreos. They just have to boot then wire everything together 15:09:22 09:49 tongli: do we want to do that? 15:09:24 09:49 tongli: our dockerwarm workload uses coreos image already. 15:09:26 09:49 topol: my preference would be to do it how real operators would actually do it 15:09:28 09:50 tongli: @topol, that is my preference as well. 15:09:30 09:50 topol: my guess that is something probably more on the faster side 15:09:32 09:50 tongli: @zhipeng, the spec in one sentence will be create a kubernetes cluster on OpenStack cloud. 15:09:34 09:51 topol: tongli, what did ton ngo recommend 15:09:36 09:51 topol: I would start with that 15:09:38 09:51 zhipengh: @tongli and to do what ? 15:09:40 09:51 tongli: use binary. 15:09:42 09:51 tongli: not to use image 15:09:44 09:51 zhipengh: I think we should have scenarios other than just purely testing stuff 15:09:46 09:51 tongli: if use the coreos image, then there isn't a whole lot need to be done. 15:09:48 09:51 topol: zhipengh. thats fine 15:09:50 09:52 topol: can you think of stuff to add. we can view getting kube cluster up as the crawl stage 15:09:52 09:52 tongli: @zhipeng, propose what you like to do, and do it. 15:09:54 09:52 tongli: patch will be really nice. 15:09:56 09:52 Rocky_g: ++ 15:09:58 09:53 topol: So Im okay with crawl, walk, run approach with kube 15:10:00 09:53 topol: get something going and then build a bigger scenario with it 15:10:02 09:53 tongli: @topol. I was thinking once we have basic stuff working, then we can add/remove worker nodes, and ensure cluster still working. 15:10:04 09:53 tongli: apps are still running. 15:10:06 09:54 topol: so tongli go the binary route and then lets build on top of it 15:10:08 09:54 tongli: right. 15:10:10 09:54 topol: scenario suggestions and code are welcome 15:10:12 09:54 tongli: create a basic working workload, then add more things to it. 15:10:14 09:54 topol: sounds like a good approach to me 15:10:16 09:55 topol: I have some friends in kubernetes land that do Kube apps. So if we need suggestions there we can follow up 15:10:18 09:56 tongli: ok. 15:10:20 09:56 topol: everyone ok with the approach we are taking with Kube? 15:10:22 09:56 Rocky_g: Yeah. But we do eventually need the app on top 15:10:24 09:57 topol: Rocky_g agreed 15:10:26 09:57 Rocky_g: Kewl 15:10:28 09:58 Rocky_g: I'm dropping off. In China and have an early morning. Bye! 15:10:30 09:58 topol: Kubernetes has a whole SIG dedicated to Apps. The WG lead is very nice. I just have to goto slack to contact her 15:10:32 09:58 topol: OK. 15:10:34 09:58 topol: #action tongli to use binary to do kube, will add an app/scenario on top. looking for suggestions 15:10:36 09:58 tongli: @Rocky_g, I am in china as well. 15:10:38 09:58 Rocky_g has left IRC (Quit: Bye) 15:10:40 09:58 topol: 1 min left. anthing else? 15:10:42 09:59 tongli: we are good I think. 15:10:45 09:59 topol: K, great meeting everyone! Thanks for partiicpating 15:10:46 09:59 topol: #endmeeting 15:10:48 10:00 daniela_ebert has left IRC (Quit: Page closed) 15:10:52 #end meeting 15:10:54 #endmeeting