15:03:16 <piet> #startmeeting openstack_ux
15:03:17 <openstack> Meeting started Fri Feb 12 15:03:16 2016 UTC and is due to finish in 60 minutes.  The chair is piet. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:03:18 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:03:20 <openstack> The meeting name has been set to 'openstack_ux'
15:05:22 <piet> Huh...I managed to break the UX Meetings wiki
15:05:35 <robcresswell> uh oh
15:05:42 <piet> Let me wing it
15:06:08 <piet> #topic London/Ironic
15:06:12 * ducttape_ suggest some ux work on the wiki edit ui
15:06:43 <piet> Oddly, I made a slight change...
15:07:23 <piet> So I had to change the London mid-cycle slightly which will impact robcresswell
15:07:59 <robcresswell> :o
15:08:08 <piet> The last day (Friday) will be focused exclusively on completing the Ironic standalone
15:09:10 <piet> Rather, the design proposals for the Ironic standalone.    Unfortunately, it has fallen way behind and we need to deliver something soon.
15:09:42 <piet> The plan is to try to review progress with operators remotely for an hour halfway through the day
15:10:15 <piet> The deliverably at the end of the day will be agreement on the design direction including panels and workflows
15:10:34 <piet> And have the design uploaded to Invision by Tuesday
15:10:42 <piet> Hell or high water
15:11:26 <piet> robresswell - need your help reviewing work, proposing designs and keeping the team moving forward
15:11:43 <piet> robcresswell
15:11:50 <robcresswell> Cool. You want me to attend the Friday too then?
15:11:53 <robcresswell> Or would I be in the way
15:12:00 <piet> Yes, please
15:12:08 <robcresswell> Cool, thats fine by me
15:12:24 <piet> betherly will also be there and has expereince with Ironic
15:13:12 <piet> Unfortunately, OpenStack UX didn't deliver and we need to get back into good standing
15:13:33 <robcresswell> Gotcha
15:13:56 <robcresswell> Y'know, a couple of my colleagues are working on ironic in the UK. Should I asked them along?
15:13:59 <piet> It may be a bit uncomfortable
15:14:21 <piet> Are we good?
15:14:30 <robcresswell> Yeah sure
15:16:19 <piet> ducttape_ Depending on time difference I may try to pull you in for an hour review. However, it also feels like you've had to do too much of the review work....
15:16:36 <piet> ducttape_ totally OK if you don't have time
15:17:14 <piet> #topic End user dashboard interviews
15:17:16 <ducttape_> ok, for ironic ?
15:17:33 <piet> ducttape_ Ironic standalone
15:17:51 <ducttape_> ok, I'm less useful there, but I'll try
15:18:09 <piet> k - I can also pull in a few others
15:18:34 <piet> Has anyone attended any of the end user interviews?
15:19:07 <hurgleburgler> A few of us have
15:19:12 <matt-borland> not in the last week
15:19:17 <matt-borland> but yeah, I was in a few
15:19:22 <robcresswell> I did pre-skiing, but been away and now kinda swamped with Horizon M-3 work
15:19:29 <piet> I think robcresswell , mattborland and hurgleburgler were there
15:19:34 <piet> What did you think?
15:20:12 <robcresswell> I think the idea is great. I view UX as research primarily and forming opinions from that.
15:20:27 <robcresswell> From the early interviews, ithink they need to be shorter to get more attention
15:20:51 <piet> Yeah, maybe down to 45 minutes
15:21:04 <piet> Any take-aways at this point?
15:21:11 <robcresswell> I'd honestly drop it even lower, and go for very direct questions, but thats just me.
15:21:30 <piet> Thirty minutes?
15:21:43 <robcresswell> Yeah I'd go for 20 - 30.
15:22:13 <piet> 20 may be a bit short because it keeps you from going off on tangeants
15:22:34 <hurgleburgler> +1 robcresswell
15:22:36 <hurgleburgler> I like that idea
15:22:56 <piet> I'm kind of thinking that Overview panel does not equal Dahboard
15:23:30 <piet> Not sure that end users are in Horizon enough to effectively use it as a dashboard
15:23:59 <piet> Discuss among yourselves ;^)
15:24:01 <robcresswell> So, from what I get from general feedback is:
15:24:34 <robcresswell> Project overview should be something like quick tasks for the average pleb who wants to grab a couple of instances for his dev team etc
15:24:59 <robcresswell> Admin overview should be like... your control panel. Whats the status of my cloud, is anything on fire
15:25:32 <ducttape_> robcresswell - admin overview is not that, right now
15:25:40 <robcresswell> At the moment Horizon is a collection of tabled resources
15:25:45 <piet> ducttape_ shoudl it be?
15:25:51 <ducttape_> it is a way to have a page fall over and not tell you very many useful things
15:25:58 <robcresswell> And really lacks a overall view
15:26:03 <robcresswell> ducttape_: Yeah atm its a dead page.
15:26:32 <ducttape_> I think it's fine to just remove overview, and not try to consolidate one central overview page at this time
15:26:37 <piet> ducttape_ Do you use a different dashboard ?
15:27:08 <ducttape_> icinga is general health for operators, monasca we use too.  there are multiple dashboards we have on monitors
15:27:43 <piet> ducttape_ so dashboard is used elsewhere?
15:28:12 <piet> ducttape_ Do you ever look at the overview panel?
15:28:13 <ducttape_> the admin overview is used (nearly) never in our setup.  it always fails
15:28:27 <piet> k
15:28:48 <ducttape_> and an admin needs undercloud health, like status of machines, rabbit health, db health, etc
15:29:12 <piet> It was kind of stunning how many end users were made aware of an issue AFTER a customer complains
15:29:20 <ducttape_> horizon is not interested in doing that undercloud stuff, b/c it will vary from deployment to deployment so much
15:29:50 <hurgleburgler> is that just from large deployments to small deployments?
15:29:56 <hurgleburgler> or is it even more granular than that?
15:30:00 <ducttape_> piet you should try to write a health monitor for every single place that can fail.   it's never ending the amount of fail that happens
15:30:42 <piet> ducttape_ Right after I walk Waterbug
15:31:05 <ducttape_> I think it varies, but most every operator will get notified of an issue by a customer.  hopefully this is the exception to the rule - but this will almost always happen
15:32:29 <piet> hurgleburgler what was your take-away?
15:32:47 <hurgleburgler> Its difficult to make horizon a proper dashboard
15:33:07 <hurgleburgler> but, we already knew that was a problem
15:33:10 <ducttape_> I would assume that operators are going to get notified of an issue through some monitoring system.   at the point they might log into horizon, they have an idea that a problem is in area X or Y
15:33:50 <ducttape_> what should be the focus is getting into the admin dashboard to resolve well known problems
15:33:55 <piet> hurglerburgler Because of technical challenges or confusion around the use case?
15:33:59 <ducttape_> like a compute host dies
15:34:30 <hurgleburgler> mostly technical challenges
15:34:50 <ducttape_> when a compute host dies, you need the people and projects impacted, move the vms to new hosts, etc.   doing this within horizon is several pages
15:34:54 <piet> Feel like we're talking across both overview panels
15:34:56 <robcresswell> ducttape_: Yesss, this is what I'd like to see
15:35:17 <robcresswell> Overviews should have the ability to catch 3/4 of users immediately
15:35:28 <piet> ducttape_ should the operator overview panel be a priority for UX?
15:35:47 <ducttape_> I don't think so, not compared to other issues
15:35:54 <piet> k
15:36:17 <ducttape_> an effective operator panel would be great, but there are other issues that horizon does not really handle well at this point  - which are bigger pains
15:36:22 <piet> ducttape_ +1
15:36:47 <ducttape_> operator "overview"
15:36:51 <ducttape_> missed that word
15:37:05 <piet> Kind of stitching together panels in the operator overview
15:38:52 <piet> matt-borland what did you take-away from the interviews?
15:39:08 <matt-borland> my main takeaway
15:39:20 <matt-borland> frankly was that there are many workflows that are too slow for people to use
15:39:23 <matt-borland> which I kinda knew
15:39:28 <matt-borland> but now I have evidence
15:39:37 <matt-borland> :)
15:39:39 <piet> how do you mean?
15:39:48 <ducttape_> like a workflow to do end user or operator tasks ?
15:39:54 <matt-borland> yes.
15:40:18 <matt-borland> there are many activities that don't scale
15:40:25 <matt-borland> doesn't matter what domain or what user
15:40:32 <matt-borland> "domain" as in "subject"
15:40:55 <piet> matt-borland There is a need combine workflows into a single use case?
15:41:04 <matt-borland> it's scale problems
15:41:10 <matt-borland> more than other things.
15:41:16 <piet> Ahhh...back end issue?
15:41:19 <matt-borland> yep
15:41:28 <piet> Gotcha
15:41:36 <matt-borland> but also it doesn't hurt to make workflows more accessible :)
15:41:46 <robcresswell> Horizon and scale are not two words that go well together, its something we've been stalling on for too long.
15:42:15 <piet> Yeah, combining workflows into a single panel may have wheels
15:42:51 <matt-borland> yep
15:43:03 <robcresswell> It gets difficult when you have a lot of other resources to pull in.
15:43:14 <robcresswell> Look at the instances code, makes requests all over the place
15:43:19 <ducttape_> I'm not sure having lots of big workflows is what I would shoot for, this seems like a complex solution where we might have simpler options
15:43:22 <robcresswell> when you;'re an admin, it just falls over.
15:43:23 <matt-borland> yeah, hence my interest in getting Searchlight really going
15:44:06 <ducttape_> if all admin pages would not try to list everything, that is a simple change, does not require a workflow, and makes things better in those cases
15:44:18 <robcresswell> Searchlight has a long way to go before it becomes Horizons panacea though.
15:44:33 <robcresswell> ducttape_: this guy, he knows :)
15:44:53 <robcresswell> Requiring a filter parameter before listing anything would be a relatively straightforward change
15:45:11 <robcresswell> Would probably make the panels much more usable?
15:45:28 <ducttape_> well, taking advice from someone named ducttape_ will produce simplest solutions possible
15:45:44 <matt-borland> robcresswell, short answer is probably "YES" for many operators
15:45:49 <matt-borland> but it will also frustrate others
15:46:03 <hurgleburgler> yeah
15:46:08 <matt-borland> so, settings :)
15:46:20 <ducttape_> there could be an option where you visit the page and say, to hell with this, go ahead and show me everything and bring the pain
15:46:26 <matt-borland> yep
15:46:33 <matt-borland> "I'm feeling lucky"
15:46:43 <robcresswell> Yeah
15:46:49 <robcresswell> and just have that default to on/off in settings
15:47:01 <robcresswell> little show all checkbox or similar.
15:47:07 <ducttape_> if you are an admin and in horizon, you no longer "feel lucky"  ;)
15:47:09 <piet> ducttape_ Can I quote you "bring the pain!!!!"  ;^)
15:47:28 <ducttape_> yes, a "bring the pain" button should happen
15:47:55 <piet> What does it do?
15:48:10 <TravT> it opens horizon
15:48:20 <ducttape_> it reverts an admin table listing to go ahead and work in the current mode, where we try to show everything and stuff is slow
15:48:32 <piet> Hi travist !
15:48:36 <ducttape_> touche TravT - well played
15:48:39 <piet> TravT
15:48:48 <hurgleburgler> lol
15:48:55 <TravT> hi, for some reason i still have your old meeting invite in my calendar piet
15:48:58 <TravT> just killed it though
15:49:11 <robcresswell> Yeah, my calendar also said meeting is in 10 mins time
15:49:27 <robcresswell> luckily I saw piets mention in horizon chat
15:50:23 <piet> Better to download the invitation from the OpenStack wiki
15:51:04 <piet> http://eavesdrop.openstack.org/
15:51:37 <ducttape_> so, we have some consensus, matches horizon plans too.....   for admin stuff it would be good to make the listing pages not list all the things, and instead prompt for a query filter
15:51:58 <ducttape_> and the admin overview page perhaps should be configurable for off / on
15:52:11 <ducttape_> as it will likely need to go away / be re-written
15:52:12 <piet> #startvote
15:52:13 <openstack> Unable to parse vote topic and options.
15:52:44 <piet> Let's do a quick vote
15:52:58 <TravT> syntax: #startvote (question) ? choice1, choice2
15:53:16 <piet> Yeah...maybe not this morning...
15:53:41 <robcresswell> TravT: You're more useful than the meetbot
15:53:49 <piet> Pasting to my IRC notes
15:53:50 <hurgleburgler> :-D
15:54:02 <ducttape_> I don't think the discussion above differs from horizon plans already discussed / in place
15:54:13 <piet> ducttape_ k
15:54:16 <TravT> so, it seems that eavesdrop doesn't have any meetings logs since 2015
15:54:17 <robcresswell> Yeah, its an existing Horizon plan really
15:54:17 <TravT> http://eavesdrop.openstack.org/meetings/ux/
15:54:20 <TravT> is that the right one?
15:54:39 <robcresswell> http://eavesdrop.openstack.org/meetings/openstack_ux/
15:55:07 <TravT> ok, link from main page is wrong then
15:55:16 <piet> I go to http://eavesdrop.openstack.org/ , search for UX and download the invitation
15:56:09 <piet> #topic Nominees for OpenStack UX
15:56:35 <piet> I'm putting forward Dan Kingshott to be added as a core
15:57:26 <robcresswell> Mailer it
15:57:32 <robcresswell> So its more open :)
15:57:36 <piet> We need more operators and he has been the most vocal
15:57:43 <piet> Sure, just putting it out there
15:57:50 <robcresswell> Cool sounds good
15:58:03 <piet> robcresswell Which email list?
15:58:30 <robcresswell> the openstack-dev mailer
15:58:51 <piet> Deep knowledge as an operator and very opinionated
15:59:00 <robcresswell> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
15:59:03 <piet> robcresswell Thanks!
15:59:12 <TravT> the traditional process is to send out a message to mailer and give a 5 day period for naysayers
15:59:22 <robcresswell> Yup :)
15:59:27 <piet> Kk
16:00:08 <matt-borland> gotta run...thanks!
16:00:15 <piet> Alright - thanks for you continued participation!  A little worried about the OPenStack UX project these days....
16:00:35 <robcresswell> Nah, keep it up Piet
16:00:42 <piet> Losing a bit of wind...
16:00:50 <piet> Looking forward to London
16:00:54 <piet> #endmeeting openstack_ux