19:01:32 <hogepodge> #startmeeting refstack
19:01:33 <openstack> Meeting started Mon Jun 15 19:01:32 2015 UTC and is due to finish in 60 minutes.  The chair is hogepodge. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:01:34 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:01:34 <davidlenwell> o/
19:01:37 <openstack> The meeting name has been set to 'refstack'
19:01:48 <pvaneck> o/
19:02:06 <hogepodge> Agenda #link https://etherpad.openstack.org/p/refstack-meeting-15-06-15
19:02:21 <hogepodge> Please take a look and add any items. Most are based off of existing code reviews.
19:02:23 <sslypushenko__> o/
19:03:07 <hogepodge> #topic Infra Hosting
19:03:45 <hogepodge> Afaik, what's left if +1 reviews from the team then working with infra to get the +2s so we can start the process.
19:03:54 <hogepodge> s/if/is
19:03:54 <davidlenwell> https://review.openstack.org/#/c/188207/8
19:04:07 <davidlenwell> we need to manually rebase ..
19:04:28 <hogepodge> because of the index?
19:04:34 <davidlenwell> don't know if we should wait for more upvotes.. it will wipe the vote history
19:04:35 <davidlenwell> yes
19:05:07 <hogepodge> Rebasing faster stils leaves the comments, and needs to be done anyway
19:05:40 <davidlenwell> it will come in as a new patchset
19:06:41 <davidlenwell> which will put the comments on the previous patchset ..
19:07:05 <hogepodge> Can we rebase today and get the team to upvote, then ask for review from infra? Plus integrate any changes from previous comments. Feels like we're close to being done.
19:07:24 <sslypushenko__> +1
19:07:50 <davidlenwell> yes.. I can do that..
19:08:37 <hogepodge> The puppet deployment script will need to be modified to deploy to two servers rather than the one (which I think is what it does right now)
19:09:08 <krotscheck> o/
19:09:24 <hogepodge> Any further items on this topic?
19:09:39 <krotscheck> does it need to live on two servers?
19:09:39 <davidlenwell> krotscheck:  did you catch hogepodge's last message
19:09:51 <davidlenwell> yes
19:10:02 <krotscheck> Because....
19:10:07 <davidlenwell> if you read scroll back in #refstack you can see the reasoning
19:10:14 <davidlenwell> or i can go into it again here if you like
19:10:37 <krotscheck> My scrollback is transient, I don't have what you're referring to.
19:11:12 <davidlenwell> ahh .. well long story short .. we want to be able to scale the api seperately from the web front end..
19:11:28 <davidlenwell> the web front end can be scaled with a cdn.. but the api needs an lb and multiple servers to scale
19:12:05 <sslypushenko__> davidlenwell: I still can't get why we really need this)
19:12:13 <krotscheck> That feels like premature optimization. Do you have any kind of usage metrics that indicates that you need that?
19:12:25 <davidlenwell> its not an optimization
19:12:32 <davidlenwell> they can be on the same server now
19:12:47 <davidlenwell> just want seperate urls .. so they are easier to decouple later
19:13:10 <krotscheck> davidlenwell: That requires two different SSL certs. Infra's going to ask some questions about that.
19:13:33 <krotscheck> Unless they're on different domains.
19:13:35 <krotscheck> Sorr
19:13:38 <krotscheck> Only if they're on different domains.
19:13:42 <sslypushenko__> davidlenwell We can decouple Refstack without hosting it on 2 uls
19:13:50 <krotscheck> if they're on the same domain, then you can selectively LB the /api/ path
19:13:50 <davidlenwell> api.refstack.org and www.refstack.org
19:14:19 <davidlenwell> see that sounds more complex to me..
19:14:32 <sslypushenko__> +1 for refstack/api
19:14:43 <davidlenwell> it doesn't require two certs .. a single cert can work on both domains.. I do it with lenwell.org and home.lenwell.org
19:15:05 <krotscheck> Ok, so I think we all agree that separation of concerns - API vs. Client - is a good thing.
19:15:26 <krotscheck> The question is: Should we host on different domains or on the same one?
19:15:50 <krotscheck> With an easy-to-use CORS middleware accessible, there's no technical concern about the browser sandbox.
19:15:52 <davidlenwell> I think seperate domains are easier to decouple later ..
19:16:20 <krotscheck> Load balancing is very doable by either domain or by uri
19:16:28 <krotscheck> So those aren't concerns either.
19:16:43 <pvaneck> wasn't there talk about perhaps moving to refstack.openstack.org at some point?
19:16:46 <krotscheck> So the only real question is: One or two SSL certs? (Infra doesn't do wildcard, best of my knowledge)
19:16:59 <krotscheck> That too.
19:17:41 <krotscheck> On the server side, it's really just a question of vhost configuration.
19:18:01 <hogepodge> If I understand the history correctly refstack.org was conceived of at a board meeting specifically to host this service, so it has board backing.
19:18:32 <hogepodge> I don't have a strong opinion of one domain vs two.
19:18:48 <davidlenwell> I don't think it will be a problem to get a wild card cert.. but we can defer to next week when we have catherine here to make a call
19:18:56 <krotscheck> So, if we're ok with two SSL certs, then I would say we go with two domains, because I really don't want to write conditionals into the puppet module to detect whether it needs to munge everything into one vhost or provide two.
19:19:02 <hogepodge> Is this something that we would like to vote or, or further code review including infra input?
19:19:35 <krotscheck> Well, isn't this discussion what the spec is for?
19:19:38 <davidlenwell> I think we need input from infra.. right now they have not commented on this issue at all
19:20:18 <sslypushenko__> +1 for discussion with infra
19:20:45 <hogepodge> Ok, leave for discussion in the cert. Can everyone state their case in the review and we can solicit infra feedback? If they have a strong stance I would tend to defer to their judgment. I'd also like Catherine to have a say too. She should be back next week.
19:20:59 <davidlenwell> +1
19:21:08 <krotscheck> +1
19:21:25 <hogepodge> Ok, moving on to the next topic.
19:21:36 <hogepodge> #topic OpenStack ID Integration
19:21:44 <hogepodge> #link https://review.openstack.org/#/c/176335/
19:21:48 <hogepodge> sslypushenko__: can you update us?
19:22:25 <sslypushenko__> I finished unittests for this path
19:22:47 <sslypushenko__> I need Docker patch to be merged for some addtitional testing
19:23:00 * krotscheck has a call, will be partially available
19:24:45 <hogepodge> Let's merge the next topic into this one then, the docker patch
19:24:54 <hogepodge> #link https://review.openstack.org/#/c/182266/
19:25:02 <sslypushenko__> +1)
19:26:05 <pvaneck> sslypushenko__: I commented on some things. But once that is complete, I think it is good to go
19:26:11 <davidlenwell> we had upvoted it but you patche dit again
19:26:22 <hogepodge> sslypushenko__: is it still a wip?
19:26:38 <sslypushenko__> I resolved that issue, pvaneck
19:26:47 <pvaneck> the issue in start.sh?
19:26:56 <pvaneck> one line needs to be updated to fit with master
19:26:57 <sslypushenko__> I think it is ready
19:27:43 <davidlenwell> I will review it again
19:27:51 <sslypushenko__> pvaneck I meant issue with Refstack url
19:28:45 <sslypushenko__> oh! I see new review
19:29:28 <sslypushenko__> I will update  code ASAP
19:29:43 <pvaneck> yea, I tested this morning, and it works well once that cd line is fixed
19:30:26 <hogepodge> Great. Glad to see this moving forward. Anything else on these two topics?
19:30:49 <sslypushenko__> Nothing ftom my side
19:31:04 <hogepodge> #topic Schema Version Handling
19:31:18 <hogepodge> pvaneck: you've mostly been working on this.
19:31:24 <hogepodge> Any items or issues?
19:31:33 <pvaneck> tested with 2015.next.json
19:31:44 <pvaneck> works from what I can tell
19:32:03 <hogepodge> pvaneck: I wanted to point out these reviews:
19:32:18 <hogepodge> #link https://review.openstack.org/#/q/status:open+project:openstack/defcore,n,z
19:32:34 <hogepodge> These are the open defcore reviews, a bunch of them reintroduce flags using the new format.
19:33:12 <hogepodge> pvaneck: If you want to see how flagging is being handled. Basically if the flag key appears in a test it will have a value which is a dictionary of date, reason, and action
19:33:48 <hogepodge> So this, for example: https://review.openstack.org/#/c/189927/
19:33:59 <hogepodge> https://review.openstack.org/#/c/189927/1/2015.next.json,cm
19:34:18 <pvaneck> hogepodge: yep, I made sure v1.3 schema handling handled that format
19:34:31 <hogepodge> pvaneck: ok, great. Just wanted to make you aware of it.
19:34:38 <pvaneck> are any flags merged yet, so i can test to see if it works as intended?
19:35:01 <pvaneck> I'm pretty sure it does, but just in case
19:35:06 <hogepodge> pvaneck: not yet, flag merging is going to take a bit of time while we sort out what can be fixed and what can't
19:35:20 <pvaneck> alright
19:35:22 <hogepodge> (by fixed I mean fixed quickly)
19:35:51 <hogepodge> Anything else on this topic?
19:35:59 <pvaneck> all good here
19:36:19 <hogepodge> #topic Auto-Watch Configuration Changes
19:36:28 <hogepodge> #link https://review.openstack.org/#/c/188835/
19:36:43 <hogepodge> krotscheck: this one is from you. Looks like a straight forward feature addition.
19:36:47 <pvaneck> that should be an easy merge
19:37:00 <krotscheck> Yep
19:37:32 <hogepodge> Ok, so just needs reviews.
19:37:37 <hogepodge> Figured that would be a quick one.
19:37:55 <hogepodge> moving on
19:38:06 <hogepodge> #topic Client: current and future work
19:38:24 <hogepodge> The client hasn't seen a bunch of development lately. One main patch outstanding to add https support
19:38:30 <hogepodge> #link https://review.openstack.org/#/c/191129/
19:38:53 <hogepodge> I was wondering if anyone had a vision of the direction the client would move in. New features or fixes?
19:40:17 <sslypushenko__> In terms of future plans, I can say that we need a mechanism for preventing multiple uploads of same test results
19:40:49 <hogepodge> sslypushenko__: I would imagine that would be a server-side api update with client support
19:41:02 <sslypushenko__> I think it will be part of work on migration to subunit2sql
19:41:17 <sslypushenko__> hogepodge Not really
19:41:57 <pvaneck> I think catherine had some ideas as well
19:42:08 <sslypushenko__> I thought about it and I didn't find way how to do this without client
19:42:55 <sslypushenko__> Actually, It this issue can be easily resolved with test result ids  generated by client
19:43:40 <pvaneck> I think eventually we may need a new way to identify clouds as well
19:43:46 <pvaneck> we use the keystone service id now
19:44:01 <pvaneck> but you need admin rights to view that I believe
19:44:45 <sslypushenko__> I think it is sounds reasonble
19:44:51 <hogepodge> pvaneck: that's problematic
19:44:53 <pvaneck> if we want non-admin users to test a cloud, then I dont think the client would be able to get the keystone Id
19:45:11 <pvaneck> if the tempest.conf doesnt have the admin credentials listed
19:45:13 <hogepodge> pvaneck: having non-admin users test clouds is essential.
19:45:58 <hogepodge> pvaneck: can you work up a spec for how that could be worked around?
19:46:42 <pvaneck> catherine comes back i think thursday. I'll brainstorm with her
19:46:54 <hogepodge> pvaneck: thanks
19:47:10 <sslypushenko__> I think that we will have to work  keystone team in terms of this issue
19:47:33 <hogepodge> sslypushenko__: that would be a much longer term effort that can't help us with existing deployments
19:48:06 <sslypushenko__> okay. Let's try to think about it
19:49:11 <hogepodge> Anything else?
19:49:30 <sslypushenko__> About client?
19:49:36 <hogepodge> yeah
19:50:00 <sslypushenko__> Nothing from me
19:50:09 <hogepodge> #topic Open Discussion
19:50:48 <hogepodge> One that hit me from earlier in the meeting, should we add openstack eavesdrop and other bots to the #refstack channel?
19:51:07 <Rockyg> ++
19:51:21 <hogepodge> This was just based on davidlenwell and krotscheck chatting about room history
19:51:43 <davidlenwell> I thought it already was
19:51:45 <Rockyg> And OpenStack standards...
19:51:56 <davidlenwell> at least it used to be
19:52:11 <hogepodge> I see gerrit is plugged in
19:53:04 <sslypushenko__> About Docker patch -  I rebased it on current master, fixed it and do some testing. I think its ready to be merged
19:53:25 <hogepodge> Room isn't logged.
19:53:31 <hogepodge> #link http://eavesdrop.openstack.org/irclogs/
19:53:50 <sslypushenko__> +1 for room logs)
19:54:13 <hogepodge> I can poke infra to see if they can plug it in for us
19:54:21 <hogepodge> assuming no objections otherwise
19:54:36 <pvaneck> +1 for logging
19:54:44 <hogepodge> This is my last week running the weekly meetings. Catherine should be back next week.
19:54:55 <sslypushenko__> davidlenwell Please, look at Docker patch and press the button if is ok)
19:54:57 <davidlenwell> no objection here
19:55:04 <davidlenwell> yep ..
19:55:06 <davidlenwell> on it
19:55:17 <pvaneck> sslypushenko_: did you push your latest patch?
19:55:22 <hogepodge> Thanks for everyone's patience with the unusual situation.
19:56:23 <pvaneck> sslypushenko__ ^^
19:56:45 <sslypushenko__> done)
19:58:05 <hogepodge> I'll leave the meeting open for a couple more minutes. Thanks everyone. Hope you had a fun weekend, and have a fantastic week.
19:58:12 <Rockyg> Hey, vacations happen.  You can't always predict them
19:59:10 <hogepodge> #endmeeting