14:00:10 #startmeeting interop_challenge 14:00:13 Meeting started Wed Aug 3 14:00:10 2016 UTC and is due to finish in 60 minutes. The chair is shamail. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:15 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:17 The meeting name has been set to 'interop_challenge' 14:00:23 #link https://etherpad.openstack.org/p/interop-challenge-meeting-2016-08-03 14:00:24 o/ 14:00:28 #chair topol 14:00:29 o/ 14:00:34 Current chairs: shamail topol 14:00:38 o/ 14:00:40 o/ 14:00:42 o/ 14:00:44 eglute o/ 14:00:58 o/ 14:01:05 o/ 14:01:14 o/ 14:01:23 So we have the conf call open to see if folks need help getting on the irc channel 14:01:50 we have an agenda at https://etherpad.openstack.org/p/interop-challenge-meeting-2016-08-03 14:02:11 o/ 14:02:12 Greetings, my name is Hank Janssen and I work for Hewlett Packard Enterprises. I have been asked to join/lurk. I am on IRC (as you can see) do I need the cons call as well? 14:02:46 meeting will be on IRC. conf call just to help new folks onto irc 14:03:03 Than I am in business. Thank you 14:03:54 interop discussion here or on openstack-meeting / openstack-meeting-alt ? 14:04:33 Hi esker, it is here. 14:04:38 O/ 14:04:59 shamail: thx 14:05:19 np 14:05:49 it worked now. 14:05:51 had to quit limechat , then come back in. 14:05:56 lets get started 14:06:02 o/ 14:06:09 shamail you want to do the meeting commands for us 14:06:23 first topic on the agenda 14:06:46 topol: sure 14:06:46 #topic other tools 14:07:15 I dont think I have authority on this channel to set the topic 14:07:21 #topic other tools 14:07:24 You are chair 14:07:29 it just isn’t showing it 14:07:41 use a / 14:08:52 did anyone want to propose other tools for running the workloads 14:09:06 o/ 14:09:25 my view is if you have these tools so that everyone can try them I think thats what's necessary 14:09:27 have u run the "startmeeting" command? 14:09:49 Can we start with a brief overview of which tools have been submitted thus far? 14:09:58 agreed 14:10:08 hi leong: started it already as ‘interop_challenge’ 14:10:20 so far all we have is an example using terraform 14:10:45 Do we want to include the tools that the app group is using, such as libcloud, etc? 14:10:55 MarkBaker were you the one that mentioned this topic? 14:11:00 We have also discussed the possibility of using the Heat templates created as a part of the Reference Architecture for EWG 14:11:17 topol - I raised this agenda item. So I wanted to know if Red hat folks want to propose Ansible, or Ubuntu folks want to propose Juju, what is the process? 14:11:30 ok shamail 14:11:39 AFAIk terraform is not common to all distributions 14:11:49 MarkBaker I think the process is code talks. submit it for all of us to try? 14:11:52 My thinking is that any commonly-used tool with running code that's reasonably easy for folks to run should be fine...Heat, Terraform, Ansible, Fog all seem like reasonable candidates 14:12:03 +1 14:12:09 markvoelker +1 14:12:43 As topol notes, I think it's mostly a matter of "if we have a [terraform|heat|etc] workload, how do we submit it for the group to consider?" 14:13:00 markvoelker, agreed 14:13:02 hi brad. 14:13:07 E.g. are we just putting them in an etherpad somewhere? 14:13:07 markvoelker: +1. The requirements should be commonly-used tool, easily accessible by all, and documented instructions on how to use/deploy 14:13:36 so we have tongli to help try anything that anyone submits 14:14:01 Apps ecosystem have the "first app" running in shade and fog. 14:14:06 topol: So if I have a workload I think we might want to use for the challenge, just email it to tongli? 14:14:07 I think they also have heat. 14:14:11 https://github.com/kennjason/osops-tools-contrib 14:14:19 are we using this repository? 14:14:39 I thought we were going to use the OpenStack repo: https://github.com/openstack/osops-tools-contrib? 14:14:49 the EWG is working on a series of Workload Reference Architecture, with Heat as a sample code to bootstrap the "workload" 14:15:02 Ok, so submit a pull request to the osops-tools-contrib repo then? 14:15:02 shamail +1 14:15:10 leong that sounds cool 14:15:24 can we collaborate together and use those workload/heat template 14:15:40 leong that sounds good to me 14:15:47 so far we have 3-tier LAMP web app, ecommerce, big-data (hadoop) and media transcoding 14:15:51 I think the osops-tools-contrib has a lower bar for getting through the review process. 14:15:56 the 3-tier lamp is ready for testing 14:16:03 tongli: can you please type here so people on IRC are participating as well? 14:16:19 We are having two conversations (one on IRC and one on phone) 14:16:41 Or rather in gerrit: https://review.openstack.org/gitweb?p=openstack/osops-tools-contrib.git;a=summary 14:16:56 so action item document how folks can submit a workload 14:17:04 topol: yes please. =) 14:17:15 ok this is the url 14:17:16 https://github.com/openstack/osops-tools-contrib 14:17:20 The phone conversation is about who has +2 and +A on the repo. 14:17:56 #action Document how to submit requests for new tools (markvoelker, shamail) 14:18:00 Ok action item captured. Lets see those new workloads flow in 14:18:51 #info Disconnecting from the phone bridge and moving to IRC only 14:19:07 shamail Yay! 14:19:22 will be irc only for all meetings after this 14:19:30 where are the objectives for this work? 14:19:34 @topol, +1 14:19:47 feeling a bit confused and out of context 14:20:40 gema, we have a background deck we can send you. But in a nutsheel its showing all the OpenStack Clouds can run the same workloads. so its demonstrating portable workloads 14:21:13 #topic clouds readily available 14:21:16 topol: ack, a link to the deck in this meeting log would be good 14:21:20 Being the newbie here, the intent is to show workload interoperability, not cloud interoperability? 14:21:51 gema, https://drive.google.com/open?id=0B9udXv_O5_TsaktrSm5uNlAtVFZZWTZkck8xZ0ItcEprMlBj 14:21:56 MarkBaker: thanks! 14:21:59 #link https://drive.google.com/open?id=0B9udXv_O5_TsaktrSm5uNlAtVFZZWTZkck8xZ0ItcEprMlBj 14:22:01 * markvoelker adds that to the etherpad too 14:22:03 hmmm, I thought this is very much about interoperability. 14:22:13 so as of right now one workload fromalready exists in this repository (teraform/docker swarm) 14:22:13 tongli baby steps 14:22:43 so the outputs would be test scripts ? 14:22:48 #link https://github.com/openstack/osops-tools-contrib/tree/master/terraform 14:22:48 so please sets get those other workloads added and available for all to test 14:22:53 or only showing test results would be fine ? 14:23:08 To me this project is about cloud interop at the workload level ... 14:23:18 catherineD agreed 14:23:23 @catherineD +1 14:23:37 okay if we go on to second topic? 14:23:38 hj-hpe: My understanding is that this is interop at a user level (workload) vs API level 14:23:59 topol: ++, next topic 14:24:07 cloud interop would include API, Image,workload ... levels ..but this project's focus is for workload interop 14:24:11 we were asked if folks had public facing clouds that folks could use to test on 14:24:39 in our last meeting the assumption was everyone tested on their clouds whether they had an on-prem or public 14:24:51 That helps Dwalleck, thank you! 14:24:56 can someone explain workload level a bit, please? 14:24:57 catherineD: +1, this efforts focuses on workload mobility/portability using tools that leverage OpenStack-Powered clouds (e.g. passed DefCore) 14:24:58 Isn't it workload portability rather than workload interop? 14:25:10 topol: Yeah, I was planning on running the tests myself on our internal lab gear 14:25:20 tongli ++ 14:25:23 topol: +1 14:25:28 rockyg, +1 14:25:34 rockyg +1 14:25:48 We should still set #topic even if its not working :) 14:26:05 #topic public cloud avail 14:26:06 rockyg +1 14:26:27 @catherineD, can you tell a bit more about workload level, I do not really get what that means. 14:26:36 Ok. So, to zhipeng's question. Once we have the worloads/apps, do we also come up with some scenario tests to demonstrate they work? 14:26:46 So I dont believe we have public clouds for folks to use and everyone should test on their own infrastructures unless someone volunteeers that they have this availabke 14:27:08 can we ask OSIC for some resources? 14:27:30 you can put in the request to OSIC 14:27:37 but access to OSIC requires VPN 14:27:48 leong, very cool. thats good to know 14:27:59 https://osic.org/clusters 14:28:05 rockyg: do we actually need them? I was sort of under the impression that participants would run the tests on their own and report back... 14:28:15 shamail can you record that on the etherpad 14:28:24 #link OSIC Request Form https://osic.org/clusters 14:28:32 markvoelker your assumption is correct. 14:28:33 Sure 14:28:45 The objective is to run this on multiple clouds, showing interop 14:28:48 I think we need a "neutral body" for the basic check 14:28:51 Its not required but the canonical folks asked if the option was available 14:28:59 +1 markvoelker 14:29:22 markvoelker: for testing .. as understand from the last meeting , hogepodge: suggested that every clouds participated in this project should first run a complete RefStack API tests then run the tests in the interop workload (the only one exists today is teraform/docker swarm ..) 14:29:28 I'm willing to allow users to hit one of my test clusters if they want to run it as a "neutral body" 14:29:44 (works on OSIC) So would the task be for testing the scripts used to create the workloads? Or to test portability/iterop with one of the OSIC clusters? 14:29:54 So for Barcelona and being able to show demos of the workloads running on the different clouds was what we were shooting for 14:30:25 so new topic 14:30:29 dwalleck: I would think it would be the former (testing scripts) since the goal would be to eventually run the scripts on your own clouds/deployments 14:30:51 topol: +1, I think showing the same workload running on multiple clouds is where the value is here 14:30:54 we okay to move on to which openstack versions people are using? 14:31:05 markvoelker exactly 14:31:16 markvoelker: +1 14:31:26 should we collect the run results. 14:31:44 that will be kind nice if we collect these over time. 14:31:45 tongli if the worklods have those, sure 14:32:02 new topic 14:32:03 we need a place for people to download the workload scripts, and share/report back the results in a centralised place 14:32:31 * markvoelker glances at the clock 14:32:37 Not this ? https://github.com/openstack/osops-tools-contrib 14:32:54 leong, any suggestions on what that should be? could we use the same repository? 14:32:59 I agree we need some way to collect results, so how about if I volunteer to work with tongli/topol offline to come up with a plan and we'll present it next meeting? 14:33:03 leong: the plan is to submit all scripts to https://github.com/openstack/osops-tools-contrib (we should submit the EWG RA templates here too) 14:33:08 #topic OpenStack Version 14:33:13 then what bout the results? 14:33:16 markcoelker ++ 14:33:20 can we just open issues against that repo for results? 14:33:26 markvoelker: +1 14:33:42 #action Determine where to share results (markvoelker, topol, tongli) 14:33:54 So far I have seen folks saying they are using the Mitaka version of OpenStack. Anyone planning to use a different version? 14:34:06 and maybe want to define a "result template" so that people know what to fill in 14:34:08 Mitaka, please 14:34:15 topol: I think we'll probably do Mitaka and Kilo 14:34:18 leong: +1 14:34:56 But here again, I think there's actually value in showing the same workload running on multiple versions 14:34:58 Does it make sense to standardize on a base submission (e.g. Mitaka) and then let everyone submit other results too? 14:35:08 Or is the goal for everyone to run against all of the agreed upon versions? 14:35:13 ok with mitaka (optional kilo) 14:35:18 topol: atleast be Liberty 14:35:19 K, so action item. folks should record which version they plan to use. 14:35:27 shamail: I'd argue there's more value in showing a workload running on multiple clouds, multiple versions 14:35:32 the "result repo" can include multiple "folder" with different version? 14:35:53 that's probably part of the "result template" 14:35:56 markvoelker +1 14:35:59 markvoelker: +1, the only concern is the amount of time before the summit but this would be a more robust outcome (multiple clouds, multiple versions) 14:36:05 seems we are asking for a new project. 14:36:28 so lets see what everyone can get running. Ideally try and do Mitaka? 14:36:36 multiple clouds, multiple versions can all be added over the time. 14:36:43 starts with Mitaka. 14:36:46 shamail: I think it probably works in our favor to use the products people have on the market right now (and probably also have in their labs) 14:36:51 #action Please add which OpenStack versions you plan to run interop challenge tests against (everyone) 14:37:10 missed “in the etherpad” in the action :) 14:37:11 K, lets see what folks decide to use 14:37:12 should we only test against openstack powered platforms ? 14:37:14 I think for each workload, we need to create/provide a demo script. Every cloud runs through the demo and *those* results are what we use to show interop 14:37:14 fyi, the OSIC cluster is running Liberty at the moment 14:37:31 dwalleck. good to know 14:37:42 zhipeng: +1 imo 14:37:47 My thought is we are flexible on this so its easier for folks to be successful 14:37:54 topol: ++ 14:38:08 +1 topol 14:38:10 @topol, + + 14:38:11 markvoelker: +1 with the short time that we have ... the workloads should what exists today ... and submit to the interop repository 14:38:18 get started and then expand. 14:38:22 topol: +1 14:38:31 yep. ok topic change 14:38:32 can we talk about tools for this? 14:38:35 topol: +1 14:38:55 topol: -1 just to go rogue 14:39:04 jk :) 14:39:16 #topic openstack SV presentation, listing of partiicpants 14:39:40 topol: Is there a version of this presentation that will be shared with those attending the meeting for feedback? 14:39:47 any concern from anyone if I have a chart in my deck that lists all the folks participating in the interop challenge? 14:39:59 shamail, +1, oh and -2 just for the heck of it. 14:40:03 * markvoelker puts corporate hat on 14:40:08 there's an openstack sv presentation on this? 14:40:15 topol: No concerns from me. VMware would be happy to be included. 14:40:39 topol, fine by me 14:40:39 jkomg. no. I have a presnetation and was just going to include about 3 charts on this 14:40:56 * shamail just saw markvoelker put away the hat again 14:41:00 kk 14:41:14 the charts I will use will be a subset of the deck we sent all of you for the kickoff meeting 14:41:21 so no new material 14:41:33 I just did not want to surprise anyone 14:42:13 only action item is to update the existing chart with the new participants 14:42:24 unless someone had an objection 14:42:53 okay. topic change 14:43:02 ++ But you might want to keep that slide open until right before the summit 14:43:40 topol: do we have a deadline for folk to submit workload for this project? 14:43:48 rockyg, will do. especially with how slow my co-presenters are at getting their charts done.. 14:43:57 * topol topol throws some shade 14:44:01 #topic workload testing. 14:44:37 Has anyone tried the docker swarm workload. we have tongli to help folks if running into issues. 14:44:56 topol: I haven't yet...been traveling the past couple of weeks. =) 14:45:08 Will hopefully get to it next week 14:45:17 catehrD, ideally lets get workloads available for everyone to try by August 8 14:45:23 (15 minutes left) 14:45:47 #action Submit new workload scripts by August 8th (process TBD)… (everyone) 14:45:52 no hard deadline but everyone needs time to try and get things working 14:46:21 tongli is our roving assistance/coach/mentor 14:46:24 We should probably get the app ecosystem to toss in First App, also. 14:46:31 so we are going to use terraform, is that all agreed? 14:46:34 Hard deadline is for submitting the worload ... 14:46:36 reach out if you are getting stuck (thanks tongli) 14:46:55 tongli if other folks have other tools that are commonly used we are okay with thise 14:46:58 those 14:47:08 rockyg: Marcela (mbonnell) from App Eco will be contacting topol 14:47:10 @topol, ok. 14:47:15 let's see what comes in 14:47:21 we have "heat" template ready but not terraform 14:47:30 Tongli, it will need to auto install with the app.... 14:47:39 leong heat should be fine 14:47:46 leong: I think we would want to submit heat templates from EWG not terraform scripts 14:48:04 at this point the requirements is can everyone run it 14:48:10 This will allow our tool-base to expand to include Heat 14:48:11 any concerns with heat? 14:48:13 leong: +1 ...please submit the code ... 14:48:18 @rockyg, terraform can install stuff onto the VM just gets created. 14:48:23 yes, code talks 14:48:28 EWG = enterprise WG (for reference) 14:48:37 shamail, tongli I think the apps other than docker swarm that is already terraform would use what they already have. 14:48:48 rockyg: +1 14:48:51 rockyg +1 14:48:56 use what you have 14:49:09 lets see if we run into issues 14:49:40 ++ 14:49:43 so rockyg it sounds like you have an enterprise app and tongli wont have to build one? 14:49:56 also fyi apps.openstack.org also has a list of templates (heat, murano) that we can try/test, if we want to 14:49:57 rockyg what was your app again? 14:50:06 leong: great idea 14:50:07 tongli: is this the workload that's all good to be tested ? https://github.com/openstack/osops-tools-contrib/tree/master/terraform/dockerswarm-coreos 14:50:11 topol: 10 minutes left 14:50:18 shamail, thanks 14:50:28 @rohit404, yes 14:50:39 I'm just thinking the EWG ref arch. stuff and the app ecosystem stuff. 14:50:43 so actions item for us to try the EWG app as our nterprise app 14:51:07 @shamail, link to the code? 14:51:20 on the etherpad it looks like AT&T is trying to get us all an NFV app 14:51:32 so apps are looking like they are rolling in 14:51:52 I think I covered all the agenda items. did I miss any? 14:52:06 Otherwise the floor is open for any questions 14:52:23 #topic Open 14:52:49 Who plans to submit new workload scripts (just curious)? 14:52:56 from EWG, i will get our team to submit the 3-tier LAMP Heat template to the osops-tools-contrib repo 14:53:12 leong AWESOME!!! 14:53:19 MarkBaker: will you be submitting a request for another tool? 14:53:24 additional* 14:54:14 one question I have is during next weeks meeting is OpenStack SV Summit. I will not be able to attend due to that. should we cancel or should someone lead in my absence? 14:54:26 shamail, yes I will - just need to complete testing on other OpenStacks 14:54:32 awesome 14:54:41 shamail: I'll see if I can round some up. We've got some internally that was use for testing, just need to find out if they're ok to publicize. 14:54:46 shamail, hence my quesiton about testing resources 14:54:47 MarkBaker, excellent 14:55:00 Awesome! Thanks leong, MarkBaker, markvoelker 14:55:46 5 minutes left. any other questions or concerns? 14:56:09 and does someone want to lead the meeting next week or just cancel ? 14:56:48 I can do it topol 14:56:51 If we have the deadline for apps on 8/8, we should have a meeting 14:57:04 shamail +1 14:57:09 shamail THANKS 14:57:19 rockyg +1 14:57:56 ok. we made great progress in this meeting. Im glad we identified more workloads. anything else before I end the meeting? 14:58:34 everyone know how to reach tongli if they run into issues? 14:58:49 tongli post your email 14:59:07 THANKS everyone for participating today!!! 14:59:10 ahhhhhhh, ok. litong01@us.ibm.com 14:59:29 thanks all, till next time 14:59:32 and we have jkomg too 14:59:35 thanks. 14:59:40 if you have questions 14:59:48 Thanks 14:59:58 so I put the current test candidates in the etherpad line 14 - 16 ... new candidates should be add there and code should be submitted to the repository 15:00:09 #endmetting 15:00:16 #endmeeting