12:01:01 <david-lyle> #startmeeting Horizon
12:01:02 <openstack> Meeting started Wed Feb  3 12:01:01 2016 UTC and is due to finish in 60 minutes.  The chair is david-lyle. Information about MeetBot at http://wiki.debian.org/MeetBot.
12:01:04 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
12:01:06 <openstack> The meeting name has been set to 'horizon'
12:01:12 <r1chardj0n3s> o/
12:01:15 <tsufiev> o/
12:02:11 <david-lyle> packed house
12:02:58 <r1chardj0n3s> so, I guess you're wondering why I called you all here
12:03:13 <david-lyle> :)
12:03:22 <david-lyle> general items
12:03:37 <david-lyle> reminder March 3 is the date for Mitaka 3
12:04:46 <david-lyle> Second is midcycle https://wiki.openstack.org/wiki/Sprints/HorizonMitakaSprint
12:05:07 <david-lyle> the wiki page contains a link to the topics etherpad
12:05:46 <r1chardj0n3s> yeah, I need to get onto that
12:06:22 * tsufiev started thinking if we should consume latest versions of external dependencies
12:07:47 <r1chardj0n3s> tsufiev: we should for more restrictive pinning of dependencies, especially in toolchain like selenium, etc
12:08:12 <r1chardj0n3s> s/should for/should have far
12:08:22 <r1chardj0n3s> it's late
12:08:41 <tsufiev> r1chardj0n3s, it's never too late to retreat :)
12:09:08 <david-lyle> there is a reason upper dependency limits on master are open and set on stable
12:10:14 <tsufiev> david-lyle, well, it's just a moment of weakness
12:11:42 <njirap> o/
12:12:29 <njirap> though the meeting time on the wiki is set at 2000 UTC today
12:12:57 <njirap> hi everyone
12:13:14 <r1chardj0n3s> hi :-)
12:13:42 <r1chardj0n3s> someone has been doing a poor job of copy/pasting in the meeting wiki ;-)
12:14:04 <njirap> i am an intern at openstack
12:14:26 <r1chardj0n3s> ohai njirap, I think we've exchanged email
12:14:33 <david-lyle> njirap, r1chardj0n3s fixed the wiki
12:14:43 <njirap> oh hi richard its you
12:14:50 <r1chardj0n3s> yep
12:15:29 <david-lyle> we have a couple items on the agenda today
12:15:33 <tsufiev> camouflage nick :)
12:15:43 <david-lyle> #link https://wiki.openstack.org/wiki/Meetings/Horizon#Agenda_for_2016-02-03_1200_UTC
12:15:57 <njirap> so just to intro myself i have hitherto been working on the openstack-manuals(no particular project) but i am now specializing in horizon
12:16:36 <david-lyle> welcome njirap
12:16:46 <njirap> thanks david
12:16:54 <david-lyle> #topic Are we ready to move to keystone v3 for horizon in default devstack config?
12:16:57 <tsufiev> njirap, welcome!
12:17:04 <mrunge> o/
12:17:06 <njirap> thanks tsufiev
12:17:16 <tsufiev> david-lyle, I felt it would good to remind everybody about this thing
12:17:18 <david-lyle> I think the simple answer is yes
12:17:27 <david-lyle> I test all the time with v3
12:17:55 <david-lyle> I also talked to lhcheng and he was also of that opinion
12:18:19 <r1chardj0n3s> I run against v3 with OSA
12:18:22 <tsufiev> david-lyle, is it no longer tied to enablement of non-cookies SESSION_ENGINE in devstack?
12:18:34 <mrunge> there is https://bugs.launchpad.net/python-novaclient/+bug/1522402
12:18:35 <openstack> Launchpad bug 1522402 in python-novaclient "Novaclient does not support v3 Keystone API" [High,Confirmed]
12:18:36 <r1chardj0n3s> but I don't use all of keystone
12:18:43 <tsufiev> the concern was that v3 means bigger token
12:18:49 <david-lyle> tsufiev: ah yes
12:18:51 <david-lyle> good point
12:19:04 <david-lyle> I have a different SESSION_ENGINE enabled all the time
12:19:20 <mrunge> yeah, same here
12:19:28 <tsufiev> AFAIR, SESSION_ENGINE in devstack was a blocker for this enhancement for a long time
12:19:34 <david-lyle> I may just move horizon in devstack to use locmemcache
12:19:40 <tsufiev> what I really wanted to understand if it's still a blocker
12:19:45 <david-lyle> tsufiev: yeah my patch withered a bit
12:19:52 <david-lyle> I can try pushing that again
12:19:54 <r1chardj0n3s> david-lyle: that's a sensible move
12:20:03 <david-lyle> r1chardj0n3s: which?
12:20:06 <tsufiev> david-lyle, \o/
12:20:11 <mrunge> can't we just use like memcached, and use memcache even for session engine?
12:20:31 <mrunge> I mean, memcached is easy to setup
12:20:46 <david-lyle> mrunge: https://review.openstack.org/#/c/193611/
12:20:48 <r1chardj0n3s> david-lyle: swicthing devstack to locmemcache
12:20:49 <mrunge> and we should have it anyways, no?
12:21:01 <r1chardj0n3s> since people shouldn't be deploying devstack anyway
12:21:04 <r1chardj0n3s> (lol)
12:21:18 <mrunge> good point r1chardj0n3s
12:21:24 <mrunge> make it useless
12:21:57 <david-lyle> so we either do locmemcache or I rebase my devstack patch
12:22:05 <david-lyle> other devstack patch
12:22:14 <r1chardj0n3s> (can't wait for the "my login doesn't work" bugs filed against horizon by locmemcache-devstack-deployers)
12:22:39 <r1chardj0n3s> so memcache is already part of devstack?
12:22:47 <david-lyle> r1chardj0n3s: not really
12:22:56 <mrunge> we could even use redis
12:22:59 <david-lyle> they don't want it in by default
12:23:03 <r1chardj0n3s> right
12:23:11 <mrunge> that is required by ceilo
12:23:36 <tsufiev> mrunge: aren't they using mongodb anymore?
12:23:47 <r1chardj0n3s> yeah, I thought they used mongo
12:23:55 <david-lyle> I'll rebase my memcache adding patch and if that doesn't move in the next week or so, put up a competing LocMemcache patch and see who wins
12:24:15 <mrunge> wait, I thought redis was pulled in anyways
12:24:54 <david-lyle> db mgmt increases the complexity of the devstack change
12:27:22 <mrunge> ceilometer seems to use mongo for data store
12:27:31 <mrunge> and redis for backend coordination
12:27:55 <david-lyle> why can't we have two different dbs ?
12:28:16 * david-lyle is really kidding
12:28:18 <mrunge> sure....
12:28:32 <mrunge> seriously, we should start using postgres
12:28:40 <r1chardj0n3s> I was about to say that mrunge
12:29:03 <r1chardj0n3s> or rather, that we have a db around because of the other services (which IIRC is actually mysql)
12:29:36 * r1chardj0n3s is painfully ignorant of how database access is structured in openstack
12:29:52 <mrunge> r1chardj0n3s, that's somehow constantly changing
12:30:02 <david-lyle> I prefer to stick with a cache that will clean up after itself
12:30:10 <r1chardj0n3s> of course, it's OpenStack(tm)
12:30:15 * mrunge still looking for redis reference in ceilometer
12:30:33 <mrunge> anyways, any of them should be fine
12:30:45 <david-lyle> what people use in the privacy of their own deployment is up to them
12:30:46 <mrunge> if we already have redis somewhere, we could use it
12:30:50 <mrunge> yupp
12:31:10 <r1chardj0n3s> I think david-lyle's proposal for rebasing memcache and seeing if it gets traction is worthwhile. if that gets in, then we're done.
12:31:30 <mrunge> yep
12:31:41 <tsufiev> +1
12:31:55 <david-lyle> properly testing devstack patches is time consuming to I have let that one rot a bit
12:32:02 <david-lyle> will find the time today
12:32:27 <david-lyle> ok moving on
12:32:30 <david-lyle> #topic What is actually going on with JS linting anyway? Anyone know?
12:32:30 <mrunge> for reference: https://github.com/openstack/ceilometer/blob/master/devstack/plugin.sh#L87
12:32:59 <david-lyle> ok, to step back topics for a sec
12:33:23 <david-lyle> the other thing we could do is make horizon a devstack plugin rather than integrated content
12:33:44 <david-lyle> then changing things like the SESSION_ENGINE become a lot more trivial
12:33:54 <r1chardj0n3s> david-lyle: I'm not sure what that means but it sounds intriguing
12:34:05 <david-lyle> but I'm not sure the extra effort for tempest
12:34:18 <david-lyle> although horizon should only plugin into tempest too
12:34:20 <mrunge> david-lyle, as I understand it from here: we would be fine with any cache
12:34:35 <david-lyle> I'll look into the plugin path
12:35:04 <david-lyle> r1chardj0n3s: projects can own how they are set up in devstack by putting a devstack dir in their repo
12:35:12 <r1chardj0n3s> ahh
12:35:16 <david-lyle> and setting it up as a devstack plugin
12:35:28 <david-lyle> same for tempest
12:36:23 <r1chardj0n3s> ok, shall we talk linting?
12:36:25 <david-lyle> so to use horizon with devstack, you'd have to add an plugin line to your local.conf
12:36:32 <david-lyle> yes
12:36:36 <david-lyle> lint away
12:36:39 <r1chardj0n3s> (thanks for the explanation)
12:36:42 <r1chardj0n3s> So I'm working on this Swift UI rework, and it's a bunch of JS and I've just run the "npm run lint" and hoo boy there's a lot of garbage spewing out of that
12:37:03 * david-lyle is shocked
12:37:08 <r1chardj0n3s> so I'm wondring is there anyone who knows how this stuff works, or is it just krotscheck?
12:37:25 <david-lyle> matt-borland maybe
12:37:33 <r1chardj0n3s> a lot of the spew is warnings, apparently a misconfiguration, which also results in actual errors
12:37:33 <mrunge> heh, I had the feeling, krotscheck just proposes rules
12:38:03 <r1chardj0n3s> but if I'm reading this config right (and I might not be) we're overriding some of the "openstack" lint rules with our own
12:38:30 <david-lyle> r1chardj0n3s: I believe there were many we weren't ready for
12:38:38 <david-lyle> and some that were angular specific
12:38:56 <david-lyle> but I may be making that up
12:39:03 <r1chardj0n3s> there's a lot of these "lint" rules that are all about code *style* which I'm not sure I agree with :/
12:39:28 <r1chardj0n3s> ok, maybe I should email this stuff to openstack-dev and see if I can get some info there
12:39:30 <tsufiev> r1chardj0n3s, then I think we should more actively participate in openstack-eslint-rules reviews
12:39:39 <mrunge> that is probably the issue with lint rules defined outside of horizon
12:39:46 <r1chardj0n3s> tsufiev: I kinda keep an eye on it, but the volume, along with other things...
12:40:21 <r1chardj0n3s> mrunge: as I understand it, krotscheck is trying to create an "openstack js lint standard" ... which is funny given the number of non-horizon JS projects...
12:40:36 <tsufiev> Horizon community needs at least 3 people per review to counterbalance krotchek's enthusiasm :)
12:40:45 <david-lyle> we there is at least one non horizon
12:40:50 <david-lyle> *well
12:40:56 <david-lyle> js project
12:41:08 <r1chardj0n3s> regardless, at the moment the lint configuration is *broken* and I need someone to 'fess up to knowing what it's supposed to be :/
12:41:28 <mrunge> yes, I agree with having just one definition around all openstack projects
12:41:46 <mrunge> just to have sane defaults for all
12:41:58 <mrunge> depends on %sane% here
12:42:15 <tsufiev> I'd say this initiative is just premature
12:43:14 <r1chardj0n3s> so at the moment we've got lint errors because we're not assigning the "this" variable to another variable called "vm" in our angular controllers
12:43:36 <r1chardj0n3s> except Horizon's configuration explictly says that variable should be called "ctrl" which is what we use everywhere
12:43:45 <r1chardj0n3s> but the linter complains anyway
12:43:52 <tsufiev> r1chardj0n3s, heretics, you deny the holy John Papa's Guide! :)
12:44:01 <r1chardj0n3s> tsufiev: don't make me come over there
12:44:16 <tsufiev> lol
12:44:35 <mrunge> that would be handy for meetings, though ;-)
12:44:48 <r1chardj0n3s> ok, I shall write a TODO for tomorrow to email openstack-dev to try to resolve this
12:44:52 <r1chardj0n3s> david-lyle: please move on
12:45:39 <david-lyle> #topic Install Guide and Quickstart revamp?
12:45:48 <david-lyle> njirap: o/
12:45:56 <njirap> yes thats me
12:46:15 <njirap> so i installed horizon outside devstack
12:46:33 <njirap> i followed the two guides available, the quickstart and the install guide
12:46:55 <njirap> it is my opinion that the two have the potential to be better
12:47:09 <david-lyle> njirap: dev docs or docs docs
12:47:33 <njirap> david the quickstart http://docs.openstack.org/developer/horizon/quickstart.html
12:47:42 * mrunge would think, install docs should be moved to docs.openstack...
12:47:54 <njirap> and the other install guide docs.openstack.org/developer/horizon/topics/install.html
12:47:56 <david-lyle> there is an install guide on docs.open
12:48:22 <mrunge> quickstart is meant for folks using git checkouts
12:48:34 <r1chardj0n3s> the quickstart could use some attention though, it's a little ... oddly structured :-)
12:48:50 <mrunge> yeah
12:48:54 <njirap> issue1 -> the install guide does not document some dependencies
12:49:10 <david-lyle> njirap: http://docs.openstack.org/kilo/install-guide/install/apt/content/install_dashboard.html
12:49:21 <david-lyle> is that latest version I know of
12:49:50 <mrunge> njirap, are you installing for a deployment?
12:49:58 <mrunge> or for a dev setup?
12:49:58 <david-lyle> sorry http://docs.openstack.org/kilo/install-guide/install/apt/content/ch_horizon.html is the top level
12:50:14 <njirap> hi mrunge
12:50:28 <njirap> i am personally installing for development
12:51:10 <mrunge> what dependency is missing?
12:51:11 <njirap> david am lookin at that now
12:51:27 <david-lyle> njirap: ok then the docs you referenced are more applicable
12:51:44 <david-lyle> I thought you were trying to deploy horizon
12:51:51 <mrunge> njirap, I would make a distinction between developer setup or deployment for a larger base
12:52:03 <mrunge> horizon docs just cover developer setup
12:52:23 <mrunge> the real stuff should go to docs.openstack.org, what david mentioned
12:52:50 <mrunge> but if it's not clear from our docs, that's room for improvement
12:53:01 <njirap> hi mrunge.. that information is not clear in the two guides,,, a production setup'pper is not explicitly directed to the right doc neither the development setup'per
12:53:22 <david-lyle> njirap: short answer is if you feel you find errors in either set of docs, feel free to propose changes
12:53:33 <njirap> aah ok thanks david
12:53:51 <njirap> issue2-> both guides ends at the login page
12:54:05 <david-lyle> I'm not sure a wholesale revamp is necessary, but there is always room for improvement
12:54:35 <mrunge> njirap, is there something to set up after login??
12:54:36 <njirap> i think i can setup launchpad blueprint for individual changes
12:54:54 <david-lyle> njirap: http://docs.openstack.org/user-guide/dashboard.html
12:55:05 <njirap> mrunge.. it is not documented where one can get or set the login credentials
12:55:09 <david-lyle> is past the login page
12:55:31 <david-lyle> njirap: mostly because that is owned by another service
12:55:42 <mrunge> feel free to add that info. (keystone)
12:56:12 <david-lyle> but yes, feel free to fill in any of the blanks you find
12:56:41 <njirap> hi david. it is my opinion that that should be documented as well,,the keystone bit,,,other than leaving(assuming a beginner) at the login page...
12:57:15 <mrunge> njirap, we should not duplicate that bit, hinting to keystone is fine
12:57:26 <njirap> and there are so many install guides
12:57:48 <mrunge> feel free to drop the ubuntu one...
12:57:50 <mrunge> :P
12:57:54 <njirap> haha
12:58:23 <mrunge> seriously, that is generated from one source
12:58:37 <mrunge> there is actually no duplication
12:58:52 <david-lyle> we have two minutes left
12:58:56 <mrunge> would it make sense to move that topic over to #openstack-horizon?
12:59:19 <david-lyle> sure
12:59:23 <mrunge> I would like to ask for the way forward with https://review.openstack.org/#/c/255854/
12:59:26 <njirap> yes it would mrunge but i can also attempt bug reports on launchpad
12:59:29 <mrunge> if there is any way forward
12:59:43 <njirap> i am good he meeting can move on
12:59:49 <njirap> *the
13:00:03 <mrunge> that patch adds info about a the user launching an instance
13:00:08 <david-lyle> mrunge: let's move to #openstack-horizon
13:00:09 <mrunge> rather than just showing the id
13:00:16 <mrunge> ack
13:00:16 <david-lyle> Thanks everyone.
13:00:21 <david-lyle> #endmeeting