20:00:25 <david-lyle> #startmeeting Horizon
20:00:26 <openstack> Meeting started Wed Mar 11 20:00:25 2015 UTC and is due to finish in 60 minutes.  The chair is david-lyle. Information about MeetBot at http://wiki.debian.org/MeetBot.
20:00:27 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
20:00:29 <openstack> The meeting name has been set to 'horizon'
20:00:36 <TravT> o/
20:00:40 <mrunge> o/
20:00:40 <rhagarty> hey
20:00:41 <gary-smith_> hi
20:00:47 <doug-fish> hi all
20:00:48 <crobertsrh> Hello/
20:00:49 <matt-borland> o/
20:01:01 <mattfarina> hello
20:01:04 <tqtran_> o/
20:01:16 <robcresswell> Evening all
20:01:20 <wchrisj> hello
20:01:29 <sqchen> hello
20:01:53 <lhcheng> o/
20:01:55 <david-lyle> Hello everyone
20:01:56 <jgravel> hi
20:02:03 <absubram> hi
20:02:51 <david-lyle> Ok, FPF is upon us, code should be up for patches for K-3 items
20:02:52 <vishwanathj> HI
20:03:28 <esp> o/
20:03:59 <doug-fish> is this the part of the meeting where we make confessions regarding code that isn't going to make k3?
20:04:23 <absubram> :)
20:04:25 <wchrisj> lol
20:04:32 <david-lyle> doug-fish: yes, good segue
20:04:47 <david-lyle> #link https://launchpad.net/horizon/+milestone/kilo-3
20:05:03 <david-lyle> 10 Good progress, 1 Beta Available, 12 Needs Code Review, 10 Implemented
20:05:18 <esp> haha
20:05:36 <mrunge> what's the difference here?
20:05:50 <mrunge> I mean, between good progress, beta and code review?
20:06:31 <david-lyle> beta, unsure, radomir set that
20:06:42 <mrunge> (I'm just curious how to set that right)
20:07:00 <david-lyle> good progress means most pieces are ready for review, but not all, or WIP
20:07:23 <david-lyle> Needs Code Review should mean just that, ready to review and merge as far as the author is concerned
20:07:42 <mrunge> ah, ok. thanks!
20:08:15 <david-lyle> ideally we should only have Needs Code Review on that list
20:08:33 <mrunge> yepp, agreed
20:08:38 <david-lyle> but there are a few high priority items that we can be flexible for
20:09:53 <wchrisj> david-lyle: I had a bugfix merged a couple of nights ago that isnt on this list. Should I pursue getting it on this list? How?
20:10:11 <david-lyle> bug fixes are separate
20:10:16 <wchrisj> ok
20:10:24 <david-lyle> we can land those up to RC
20:10:29 <wchrisj> cool
20:10:30 <david-lyle> and then even critical ones
20:10:39 <david-lyle> this is new features
20:11:46 <david-lyle> status on the blueprint is a mess
20:11:53 <david-lyle> was just looking through them
20:12:24 <david-lyle> if you have multiple patches on your bp and some have landed, put "-- merged" next to the patch in the whiteboard
20:13:05 <mrunge> oh yes please!
20:13:05 <vishwanathj> david-lyle, what steps do I need to execute next to get the https://blueprints.launchpad.net/horizon/+spec/fwaas-router-insertion approved for K-3, I have already uploaded a WIP patch set
20:14:25 <david-lyle> vishwanathj: I'm willing to consider it, but the dependencies haven't merged. also get amotoki on board to review it
20:14:47 <vishwanathj> The neutron side of the blueprint will merge either late this week or early next week, all our pending pieces for horizon will be complete in the next couple days
20:14:56 <david-lyle> trouble is we're very overloaded and this is coming very late
20:15:14 <vishwanathj> david-lyle, completely understand
20:15:29 <david-lyle> next week will be the end of k-3
20:15:35 <vishwanathj> got some great help from absubram to get started in a short time
20:16:05 <david-lyle> once it lands ping me and we can talk about it
20:16:17 <vishwanathj> david_lyle, thanks a bunch
20:17:41 <david-lyle> I'll start walking through the patches, at this point WIP is not enough, and reviewers need to focus on items that don't have a bunch of work left to complete
20:18:30 <david-lyle> unless the priority is high, but the WIP needs to move to Ready for Review quickly
20:19:53 <david-lyle> There were no items added to the agenda, so I'll open it up
20:20:01 <david-lyle> #topic Open Discussion
20:20:47 <wchrisj> I'd like to thank folks, especially cores for the quick turnaround of late on reviewing/approving stuff.
20:20:48 <doug-fish> there was dicussion about better error messages/exception handling today in IRC
20:21:10 <doug-fish> I know we've had a blueprint open for that for a while, but it's still an issue
20:21:13 <robcresswell> https://blueprints.launchpad.net/horizon/+spec/improve-error-message-details-for-usability
20:21:19 <robcresswell> ^^ relevant
20:21:21 <doug-fish> maybe worth a summit session?
20:21:26 <doug-fish> yeah, that's it robcresswell
20:21:37 <mrunge> yes, we need to talk about that
20:21:49 <mrunge> or even a smaller work item for the summit
20:21:52 <david-lyle> sure, add it
20:21:59 <robcresswell> Would definitely like to follow up on this as we move into Liberty. At the very least, I think a good start would be to start properly handling different exceptions and throw specific errors per exception
20:22:14 <robcresswell> With generic fallbacks
20:22:15 <david-lyle> we won't solve for K anyway
20:22:16 <doug-fish> mrunge: you have a suggestion to divide it off smaller?
20:22:25 * mrunge adding that now
20:22:36 <doug-fish> I guess I partly wanted revisit just to make sure we are all still onboard with no exception showing
20:22:51 <david-lyle> the main reason for limited progress on that is lack of options
20:23:02 <mrunge> doug-fish, I would assume, it doesn't need all attendees, but should come up with a rule or a conclusion
20:23:11 <mrunge> to put it down somewhere in the wiki
20:23:21 <doug-fish> sure - we could do that in short order I think
20:23:28 <david-lyle> yes
20:23:38 <doug-fish> coming up with a direction for improvement - there's the challenge
20:24:15 <david-lyle> mrunge: can you share the wiki you started again for the summit?
20:24:15 <mrunge> at least, we have a few ideas there :D
20:24:36 <mrunge> david-lyle, YOU started: https://etherpad.openstack.org/p/horizon-liberty-summit
20:24:44 <mrunge> and I added stuff ;-)
20:24:45 <david-lyle> oh sure, blame me
20:24:48 <david-lyle> :)
20:24:59 <mrunge> will do
20:25:06 <david-lyle> thanks!
20:25:50 <mrunge> david-lyle, for the summit, we should come up with some kind of plan, what to discuss and whom to pull in
20:26:11 <mrunge> at least some companies might want to know, who to send there
20:26:22 <david-lyle> mrunge: yes, I don't have final word on the allotment of slots we ended up with
20:26:31 <mrunge> ah, I see
20:26:35 <david-lyle> maybe all that I requested
20:26:40 <david-lyle> not sure
20:26:45 <mrunge> would be very cool
20:26:52 <TravT> when do you find out?
20:26:56 <david-lyle> either way, we can make time
20:27:03 <david-lyle> TravT: not sure that either
20:27:07 <mrunge> on the other side, not every topic needs all people involved
20:27:18 <david-lyle> schedule actually has been earlier than years past
20:28:02 <david-lyle> but with the changing format of sessions created need earlier for room configurations
20:28:16 <david-lyle> usually the newly elected PTL makes all those calls
20:28:26 <tqtran_> david-lyle: btw, any news on getting our own requirements? just wondering where we are
20:28:49 <david-lyle> tqtran_: no, although I had a hair-brained idea about that today
20:29:23 <TravT> hair brained ideas can lead to good things...
20:29:33 <tqtran_> lets hear it lol
20:29:50 <david-lyle> what if instead of 20 xstatic packages, we created 1 vendor package with versioned requirements in it, that contained all our js dependency packages
20:30:10 <david-lyle> then we have one openstack/requirements package to include
20:30:32 <david-lyle> it's messier, but from a packaging standpoint potentially less so, or the same
20:30:35 <TravT> interesting, we'd still have to ask infra for approval each time, right
20:30:43 <mrunge> that wouldn't be able to get packaged
20:30:58 <david-lyle> mrunge: horizon was
20:31:01 <mrunge> because of bundling stuff
20:31:09 <mrunge> as an exception
20:31:26 <david-lyle> yeah, there are a lot of issues with it
20:31:27 <mrunge> but now that we solved that, distros wouldn't allow that back
20:31:55 <david-lyle> well at least one distro just dropped us do to the package proliferation
20:31:58 <TravT> any updates on bower that anybody knows about?
20:31:59 <david-lyle> *due
20:32:22 <david-lyle> there is no pleasing everyone
20:32:32 <david-lyle> but everyone miserable, isn't ideal
20:32:47 <TravT> but i've heard that misery loves company
20:32:58 <TravT> ;-)
20:32:58 <david-lyle> TravT: well we aren't lonely
20:33:36 <mattfarina> there are some ideas on bower and there's https://review.openstack.org/#/c/154297/
20:33:38 <david-lyle> tqtran_: the problem with our own requirements is that it still requires all the same license and vendor checks
20:33:49 <mattfarina> that's openstack_infra speck for bower
20:34:16 <TravT> mattfarina: can you add to that etherpad?  I just added a bullet for this topic
20:34:32 <mattfarina> k
20:35:09 <mattfarina> infra seems to be open to bower if we can get the mirroring solved. going out to the net for bower and GitHub (to download the packages) regularly fails and they won't try it for Horizon
20:35:34 <david-lyle> yup, there needs to be a mirror
20:36:21 <david-lyle> since we have time, I'd also like to talk about Horizon as we move to the big tent
20:36:31 <mattfarina> there needs to be 2 mirrors. One mirror of bower (the registry) and another mirror for the code (which lives separately from bower)
20:36:55 <david-lyle> nothing 4 mirrors can't fix
20:36:58 <david-lyle> :)
20:37:08 <mattfarina> in each AZ of each host
20:37:15 <mattfarina> glad cloud is about scale
20:37:53 <TravT> Every problem can be solved by another level of indirection...
20:38:13 <david-lyle> I think the breadth of horizon is getting beyond our ability to deliver on
20:38:36 <mrunge> yes!
20:38:43 <david-lyle> there are too many projects we support
20:39:01 <mattfarina> and i think vendors want to have their own projects be easily worked into horizon
20:39:25 <david-lyle> in the future, and this will be discussed extensively at the summit, there will be even more projects
20:40:12 <mattfarina> so, can we go to a module approach? where there's a base platform that can be easily extended?
20:40:14 <wchrisj> and "private" projects within a given org
20:40:19 <david-lyle> I propose moving some project plugins out into the openstack namespace/ecosystem
20:40:41 <mrunge> mattfarina, already noted for summit
20:41:01 <mattfarina> so, like there are clients for different projects there are horizon add-ins to support different services?
20:41:29 <mrunge> that might be a way
20:41:39 <TravT> which projects?
20:41:41 <mrunge> I was thinking in the same direction
20:41:44 <david-lyle> yes and they are full members of the eco-system, just not coexistent in the Horizon code
20:41:49 <lhcheng> probably something like tuskar-ui?
20:42:01 <david-lyle> lhcheng: exactly
20:42:18 <david-lyle> but maybe with a slightly different core group
20:42:35 <david-lyle> the three projects that immediately come to mind are Trove, Sahara, and Hear
20:42:39 <david-lyle> *Heat
20:42:47 <wchrisj> yes!
20:42:56 <lhcheng> david-lyle: don't they have their own core group? or does Ana and Radomir need to +1 them?
20:43:05 <lhcheng> *+A
20:43:05 <david-lyle> they are actually very isolated from an API standpoint
20:43:29 <david-lyle> lhcheng: trying to remember
20:43:41 <mrunge> for tripleo?
20:43:55 <mrunge> i think all horizon cores are core in triple-o
20:44:08 <david-lyle> tuskar-ui has a blended core
20:44:14 <mrunge> but they need currently just a single +2 and a+
20:44:32 <mrunge> if I'm not completely wrong here
20:44:39 <david-lyle> yes, because, the horizon core could not take on that load either
20:45:08 <david-lyle> and the attention there has waned
20:45:47 <david-lyle> I think we need to provide style guides and tools for these projects to move forward
20:46:12 <david-lyle> but absorbing a full releases worth of features on rapidly growing/changing projects is very difficult
20:46:28 <david-lyle> it's also very tedious and slow for those projects waiting on us
20:47:25 <mrunge> projects are mostly python projects. we're moving away from that quite fast
20:47:31 <david-lyle> we don't have any support for ironic going into Kilo in Horizon even though it is in the integrated release
20:47:36 <doug-fish> are you envision a plugin model like cinder/nova/neutron?
20:47:47 <david-lyle> doug-fish: we have a plugin model
20:47:53 <david-lyle> we just have to use it
20:48:00 <wchrisj> Agreed
20:48:02 <david-lyle> we do internally
20:48:13 <doug-fish> panels/dashboards right?
20:48:18 <mrunge> we are using it
20:48:20 <david-lyle> projects, admin and identity are plugins
20:48:25 <mattfarina> is the plugin model documented somewhere?
20:48:30 <david-lyle> they are just all in the same code base
20:48:32 <mattfarina> i think it can be expanded on as well
20:48:36 <mrunge> sure, it's all in the wiki
20:48:47 <wchrisj> it's also in the developer docs
20:48:50 <david-lyle> mattfarina: in the horizon docs
20:48:54 <david-lyle> what wchrisj said
20:49:11 <mrunge> but you can not replace single panels or override single tables
20:49:15 <wchrisj> there's even a tutorial now
20:49:31 <david-lyle> mrunge: single panels yes, single tables, no
20:49:35 <mrunge> at least not that easy.
20:50:16 <david-lyle> we should continue to expand the pluggability to finer grained elements
20:50:21 <TravT> david-lyle: i think it would be great to spend more time in strengthening to plugin model, especially for angular usage
20:50:31 <mattfarina> What TravT said
20:50:32 <doug-fish> david-lyle: yeah that's what I was hoping for
20:50:37 <wchrisj> david-lyle: would it help to take the tutorial that's there and expand it to use the plugin model?
20:50:38 <mrunge> yes,
20:50:47 <TravT> david-lyle: and that will be easier to do if we are able to focus on less projects overall like you suggest
20:50:49 <mrunge> wchrisj, sure!
20:50:54 <david-lyle> yes, and I believe we've started some of that work with Launch instance
20:50:56 <wchrisj> I can do that
20:51:06 <wchrisj> as a first effort
20:51:27 <mrunge> wchrisj, just go ahead, that's awesome
20:51:33 <wchrisj> Done.
20:51:34 <david-lyle> sure
20:51:38 <TravT> david-lyle: yep.
20:51:39 <mattfarina> if there are more projects not part of the core it would force the plugin model to work for them and provide multiple public cases testing it
20:51:50 <esp> does the current plugin model support dashboard/panels that live in a separate package?
20:51:58 <david-lyle> esp yes
20:52:05 <esp> david-lyle: nice
20:52:09 <david-lyle> I have a demo package somewher
20:52:10 <david-lyle> e
20:52:19 <esp> I’d prefer moving toward something like that
20:52:19 <david-lyle> plus monasca-ui and tuskar-ui
20:52:37 <esp> but then I guess we have the problem of versioning
20:52:42 <david-lyle> https://github.com/dklyle/mon-ext
20:52:45 <TravT> Along these lines, this is the kind of thing I wanted to be dealing with pre-summit as well.  I'm thinking probably a big meetup will be hard to do pre-liberty, but maybe we can run some virtual sprints on these topic areas.
20:53:46 <david-lyle> esp: that's what I'm suggesting :)
20:53:56 <esp> good stuff
20:54:18 <david-lyle> openstack/requirements should help with dependency consistency
20:54:28 <david-lyle> and we can focus on the core of Horizon
20:54:35 <esp> k
20:54:37 <wchrisj> fwiw, here's the doc on the plugin settings: https://github.com/openstack/horizon/blob/master/doc/source/topics/settings.rst#pluggable-settings
20:54:46 <david-lyle> the workflows that require cross-stack coordination
20:54:58 <david-lyle> and docs and examples
20:55:10 <esp> thx wchrisj
20:55:23 <david-lyle> murano will be here soon and manila and those should be plugins
20:55:25 <david-lyle> as well
20:55:54 <david-lyle> consistent UX will be the biggest issue and we'll need to work with them on that
20:56:01 <mrunge> more to the pain points: are we removing stuff?
20:56:09 <mrunge> like trove or sahara?
20:56:16 <mrunge> move to plugins?
20:56:22 <david-lyle> and heat, that's what I'm suggesting
20:56:36 <mrunge> router dashboard?
20:56:41 <david-lyle> ceilometer will be valuable cross-stack when ready
20:56:45 <david-lyle> mrunge: +1000
20:56:56 <robcresswell> haha
20:57:00 <mrunge> even neutron?
20:57:06 <robcresswell> thats a whole lot of approval.
20:57:27 <mrunge> given, that it's only covered by a very few people?
20:57:28 <david-lyle> additionally, there could be extensions that just focus on nova if desired
20:57:48 <david-lyle> mrunge: I think neutron is central to what we need to support
20:57:58 <mrunge> agreed david-lyle
20:58:16 <mrunge> but in reality, only amotoki really understands, what's going on there
20:58:19 <david-lyle> with less context switching, maybe more can focus there
20:58:36 <TravT> david-lyle: i really like the spirit of what you are saying
20:58:37 <david-lyle> we need to work more proactively with the neutron team
20:58:42 <david-lyle> but..
20:58:49 <mrunge> :D
20:59:19 <mrunge> agreed. removing that would be too painful
20:59:59 <david-lyle> I thought that was the set up punch TravT, I waiting for the haymaker
21:00:11 <TravT> david-lyle: no, that's pretty much it...
21:00:28 <TravT> maybe just, the devil is in deciding what is in and out.
21:00:40 <david-lyle> first round is easy I think
21:00:47 <mrunge> let's toss a coin
21:00:53 <mrunge> not to blame a devil
21:00:55 <david-lyle> the PaaS layer essentially
21:01:09 <TravT> we've also run into some things on launch instance within projects
21:01:44 <TravT> for example, network profile support which appears to come all the way back too if something = 'cisco'
21:02:14 <david-lyle> That shouldn't be there, it needs to be a general solution
21:02:22 <mrunge> yes. I could assume other features added by vendors to inflict with launch instance
21:02:26 <doug-fish> and an appropriate extension point for our cisco friends
21:02:43 <mrunge> like more scheduling
21:02:55 <mrunge> different scheduling,
21:02:56 <david-lyle> yes, launch should probably allow extensions for more steps
21:03:00 <TravT> yep
21:03:01 <mrunge> nova extensions
21:03:06 <david-lyle> well it definitely should
21:03:33 <david-lyle> we're over time, and I'm not sure if someones waiting for the room
21:03:34 <TravT> that was the idea from the begining... but with the idea that we solve the concrete of what is there now and work on better extensibility
21:03:35 <mrunge> can we move that discussion to #horizon?
21:03:38 <david-lyle> Someone ?
21:03:55 <david-lyle> probably the better choice mrunge
21:04:07 <david-lyle> moving to horizon room
21:04:12 <mrunge> thanks!
21:04:13 <david-lyle> Thanks everyone!
21:04:17 <david-lyle> #endmeeting