[00:33] heh, busy day when now is the first time I've loaded up the dell order status page fo the first time today [00:33] and still building :) [09:08] bac: thank you for doing the SRU justifications for me. I'm working on the Trusty SRUs now. [11:13] morning [11:13] rbasak: awesome thanks [11:30] morning rick_h_ [11:31] bac: morning [11:31] bac: talked with wallyworld last night. They're building a CI setup for several of the small libraries around juju [11:31] bac: we talked about having the charmstore CI be located there and in the same place as the other libraries used [11:31] rick_h_: progress this morning! axw discovered i was encountering two bugs with juju. one just committed to trunk and the other on the critical list. [11:31] the latter has a work-around [11:31] bac: and I think everyone is on board with that. [11:31] rick_h_: cool [11:31] bac: oh, very cool [11:32] rick_h_: the other bug i think only affected us east. everyone says to use us west [11:32] bac: so I'm tempted to say our goal is to work with them to get things up for now. [11:32] bac: ugh, sounds like nutty bugs [11:33] rick_h_: to complete my day of frustation i found my car battery dead. and when i used my neighbor's car to jump start it, his car died and wouldn't restart. should've stayed in bed. [11:33] ouch, taking down the neighbors with you? That's not good. [11:33] rick_h_: is wallyworld working with anyone who overlaps with us? [11:34] bac: martin I believe is running the show from his team [11:34] which martin? [11:34] nicks, man. proper names due me no good. [11:34] bac: packman? /me goes to check spelling [11:34] mgz. cool. [11:34] I'm not sure on his nick :P [11:34] and best, he's in a proper time zone [11:35] there you go, mgz :) [11:35] so he's doing the actual CI setup and overlaps with us. [11:35] ok, i'll ping him when i come back. [11:35] * bac bbiab [11:35] have fun [11:58] rick_h_: so the goal is to just move charmstore over there, not the others? [11:59] hi rbasak, i've put in SRU info on the three bugs for quickstart. can you have a look and tell me if i need to make changes? [12:03] bac: yea, the gui and private tools I think we'll keep with the current setup and be aware that we should rebuild CI at some point due to juju issues [12:03] bac: but core folks will also work on the charmstore and it makes sense to have that together with the other juju tools [12:04] bac: and rbasak is looking at the SRU, he commented before you joined irc this morning so rocking on woo! [12:04] rick_h_: ever see this on azure? [12:04] juju-quickstart: error: bad API response: cannot assign unit "juju-gui/0" to machine 0: unit placement is not supported with availability-sets-enabled === rogpeppe3 is now known as rogpeppe [12:12] bac: oh hmm no. But we got CI up before the availability set support landed in juju [12:12] that's going to be interesting [12:12] rick_h_: yeah, the old instances don't have a.s. my new one does and it is pissy [12:13] bac: yea, I wonder how that effects machine view [12:13] rick_h_: i'm going to tear it down and file a bug against quickstart. we can reassign later if needed [12:13] that sounds a LOT like we need to support doing something there in order for MV to work in the GUI [12:13] i have done some mighty fine juju QA this week. :( [12:15] bac: https://github.com/juju/docs/issues/85 says there's some docs on the feature in juju-core/doc/azure-provider.txt [12:16] cool [12:17] bac: yea, availability-sets-enabled is a new attribute at bootstrap time and kills manual placement [12:17] wtf [12:27] hey, i found a work-around to a juju annoyance: [12:27] juju destroy-environment `juju switch` [12:28] no more juju nanny state [12:33] * rick_h_ relocates to the coffee shop to get over the shock of the news [12:36] rick_h_, bac: I made a couple of very minor changes to meet SRU requirements for the "Support trusty environments" backport: http://paste.ubuntu.com/7710956/ [12:36] Just FYI. Tests still pass, and I'm happy that this should be fine. [12:37] I think you've improved processes quite a bit since that commit, so this should be easier next time - just one issue per commit, please :) [12:38] rbasak: thank you [12:39] rbasak: so it'll go in as version 1.3.1? [12:40] bac: yes - because I'm cherry-picking from trunk, and I think I'm missing at least one commit that went into 1.3.2. So I can't really call it 1.3.2. [12:40] rt [12:41] This is pretty common in SRUs - no process issue here. I just shouldn't bump the version unless I'm taking the whole new version wholesale, and I'm not doing that because it's easier for me to SRU-justify fewer changes. [12:41] Well, for *you* to SRU-justify fewer changes. Thanks for doing those bugs :) [12:43] np rbasak. i felt my 'regression rationals' were a bit weak and transparentl so. :) [12:44] bac: we'll worry about that if the SRU team object :) [12:45] I've got the juju-quickstart SRU ready for upload now, but we can't get it landed or tested without the juju-core MRE SRU in place. I'll work on that now. [13:00] rbasak: ok. just let me know if you need anything from me. [13:02] rick_h_: it is a little bit of a non-issue now, but i did just get a quickstart launch on azure with gui running. [13:02] bac: yay? :) [13:03] rick_h_: while i wait on mgz i'll see if i can get a jenkins slave to work [13:11] bac: do we have bugs/links/etc for the stuff you hit that caused us grief? [13:12] bac: just want to make sure we're good citizens and when I say we've had issues they're logged [13:13] of the two juju ones, one is fixed in trunk, the other was on the critical list for 1.20. i have not filed one for the availability-sets impact on us [13:14] bac: yea, that'll be one of my goals to look at what we need to do for that across our tools [13:14] rick_h_: i'll open a bug limited to quickstart just for tracking [13:14] bac: ok thanks [13:15] bac: I think the bug needs to be two-fold. One to make sure we don't colocate by default on azure (like lxc) and second that we allow that flag to override the bootstrap is part of quickstart as well. [13:15] bac: which kind of falls into an existing quickstart bug of not being able to pass extra bootstrap config already [13:16] rick_h_: qs could look at the yaml and error if azure without that setting [13:17] bac: don't follow? [13:17] which yaml? [13:17] environments.yaml [13:17] bac: well we want them to use availability set [13:17] sets [13:17] azure without availability-set-enabled: false is an error [13:17] it's enabled by default because it's the only way to make sure your env doesn't fall over on yo [13:17] you [13:17] ok [13:18] what they did makes sense, it just puts us in a bind as far as user interactions and how our tools currently work [13:18] i shall write a descriptive bug and the proscriptive part can come later [13:18] so I can't really even argue against it, just have to get it fixed for users [13:23] rick_h_: bug 1335121 filed, maint card created [13:23] <_mup_> Bug #1335121: Azure availability set support causes failure [13:23] bac: ty [14:14] rick_h_: fwiw, i finally got jenkins and jenkins-slave up in the new azure env and the slave does show up in the config. recall it did not on our current environment. but the new azure uses work-arounds that are not appropriate for production. [14:15] bac: yea [14:15] bac: ok, I think it's good that we leared a lot, we know that we need to update CI at some point, but can't today due to azure issues. Maybe we can when 1.20 hits [14:15] rick_h_: so our current ci is a little brain dead but we dare not tear it down! :) [14:15] bac: but should hold pattern for now and coordinate with martin/ian on CI for charmstore [14:15] :) [14:16] rick_h_: mgz does not appear to be around today [14:16] bac: ok, let's add a tracking card to the charmstore lane and bring it up next week [14:16] maybe by email cc'ing me/ian as we had the conversation [14:17] bac: did you get rogpeppe's branch to pass tests and build properly? [14:17] ok. mgz works with ian and axw, so no one around until monday [14:17] bac: if so can you LGTM it and rogpeppe can land it and move forward. [14:17] rick_h_: i'll double check [14:17] bac: sounds good (on the mgz/monday work with the net week) [14:18] bac: make sure you pull juju-core and godeps to update its dependencies first, please [14:18] bac: because it needs to work with juju-core's deps [14:19] jujugui I'm still in need of a review https://github.com/juju/juju-gui/pull/407 [14:19] Looking [14:19] jcastro: or Makyo or kadams54 ^ ? [14:19] sorry, jcsackett ^ [14:19] thanks kadams54 [14:20] thanks [14:33] hmm. there's an awkward ambiguity in charm urls with respect to the store URL name space [14:34] i was thinking that it would be logical to allow https://charmstore.com/precise/wordpress/any/file/path/in/the/zip/file [14:34] for retrieving contents of charms [14:35] since https://charmstore.com/precise/wordpress currently retrieves a zip file of the whole thing [14:35] but... [14:36] you can't tell if charmstore.com/trusty/precise/foo is a file called precise/foo in the charm called trusty, or a file called foo in the trusty version of the charm called precise [14:37] so perhaps we're best just serving individual files from a different root [14:37] say charmstore.com/contents/trusty/wordpress-23 [14:38] and we could accept only fully qualified URLs with a specified series and revision [14:38] rick_h_, hatch, bac: what thinkest ye? [14:39] rogpeppe: juju-core dependencies.tsv reference errgo but it isn't in my tree. why doesn't 'go get -u' fetch it for me? or is it not really required? [14:39] bac: you probably need go get -t [14:40] damn that -t [14:40] bac: but godeps -f -u *.tsv [14:40] should do the job [14:40] bac: if you do that, you shouldn't need to use go get at all [14:40] i didn't know of -f [14:40] bac: it's a new flag [14:40] bac: you might need to go get -u launchpad.net/godeps [14:42] * rogpeppe is minded to get rid of -f and make it the default. possibly with a -F flag to supress automatic get [14:42] it's almost always what you want [14:42] rogpeppe: I think that we can't allow a charm to be named a series. [14:43] rogpeppe: and that will avoid any confusion/issue. [14:43] rick_h_: is that actually enforced now? [14:43] rogpeppe: I believe it's an error on charmworld ingestion [14:43] rogpeppe: I don't know about the charmstore [14:43] rick_h_: that does mean we can never add a series that happens to mirror an existing charm name [14:43] rogpeppe: but I think it's sane to enforce and helps us keep things clean [14:43] rogpeppe: we'll end up having a nice conversation with the charm author :) [14:44] ok, rogpeppe, i've done all of that now. so i'm in my GOPATH for juju/charm. how do i fetch your version for testing? [14:44] bac: git add remote rogpeppe git@github.com:rogpeppe/charm [14:44] bac: git fetch rogpeppe [14:44] bac: git checkout rogpeppe/bundledata [14:45] bac: wait! [14:45] bac: it's no longer in juju/charm [14:45] rogpeppe: well i want the exact version from your pull request [14:45] bac: it's in gopkg.in/juju/charm.v2 [14:45] rogpeppe: we're not talking about https://github.com/juju/charm/pull/9 [14:45] bac: so, first do: go get gopkg.in/juju/charm.v2 [14:45] bac: yeah, we are [14:45] rogpeppe: but I think this is why I want to do the api space spec and planning. To hammer all the rules/etc out in the merging of charmstore vs charmworld apis. [14:45] * bac is totally confused [14:46] bac: gopkg.in goes to github.com to get the code [14:46] rogpeppe: because precise/wordpress doesn't return a zip in charmworld, but a json of the charmdata [14:47] bac: the actual branch is "v2" in juju/charm [14:47] rick_h_: i see [14:47] rogpeppe: so we need to plan out a v4 api that makes sense of both ends and update clients/etc [14:47] rick_h_: planning the url name space is what i've been doing [14:47] rogpeppe: and I'd rather do it once vs ad-hoc [14:47] rick_h_: yup [14:48] rogpeppe: so do i keep $GOPATH/src/github.com/juju/charm ? [14:48] bac: you do need to keep that for a while, because juju-core still depends on it [14:48] rogpeppe: ok, can you share out a doc then with the current store api endpoints and the charmworld ones and we can go through them in a call and resolve conflicts/plan updates that make sense? [14:49] bac: but if you go into $GOPATH/gopkg.in/juju/charm.v2 and follow above instructions, you should get a copy of my branch for testing in the right place [14:49] rick_h_: yup, sounds good. [14:49] rogpeppe: ok, will do. pulling from a rogpeppe remote seems dangerous, though, as i dont' know it is really the one under review. [14:50] rogpeppe: http://charmworld.readthedocs.org/en/latest/api.html has the info for charmworld [14:50] rogpeppe: what if you had two branches in flight? [14:50] bac: i suggested "git fetch", not git pull [14:50] bac: that fetches the branch data but doesn't actually update anything [14:51] bac: then you can check out the branch you need [14:52] rogpeppe: ok, i didn't see that bit [14:52] bac: of course, it's git remote add, not git add remote [14:52] rogpeppe: yeah, i figured that out [14:56] jujugui call in 5 [14:56] kanban please [14:56] rogpeppe: http://paste.ubuntu.com/77115 [14:57] bac: The Paste you are looking for does not currently exist. [14:57] rogpeppe: oops: http://paste.ubuntu.com/7711522/ [14:59] jujugui call in 2 [14:59] and that's after you've done a "godeps -u dependencies.tsv" on the latest tip of juju-core? [15:00] jujugui call now jcsackett Makyo (ant is out) [15:01] * bac joins [15:06] rogpeppe: es [15:06] s/es/yes/ [15:07] rogpeppe: yes, i'd done the updates as you requested [15:28] jcastro marcoceppi you can now hide the charmbrowser with a shortcut key :) [15:39] \o/ [15:42] marcoceppi ctrl+alt+h will hide it, you can see all the shortcuts with shift+/ [15:43] hatch: or ? [15:43] I thought ? worked for shortcuts [15:43] well to get a ? you hit shift+/ :D [15:44] oh [15:44] well ? seems a lot easier than shift+/ :P [15:44] haha I always say shift+/ because I just assume people will hit / thinking of ? [15:44] maybe I'm the edge case :D [15:46] kadams54.....the test is still running...... [15:46] lol [16:12] oops I ordered something from amazon from a third party.... [16:12] darn, it's going to take a couple weeks to get here :( [16:15] whoops [16:16] next time I guess I should pay more attention [16:16] heh [16:17] it's just so easy to buy things sometime [16:23] * rick_h_ goes to get lunch === robbiew is now known as robbiew-afk [16:39] Makyo so I'm working on getting the ghost inspector to show on drop again and running into an issue where if you open an inspector [16:40] then close it, when you deploy it'll re-open that inspector [16:40] I remember you fixed this once before do you have any pointers? [16:41] Doesn't sound familiar to me, hmm... May be in the deploy callback? [16:45] well you worked on the 'on deploy switch to a service inspector' bit [16:45] maybe it's in the inspector code [16:46] ahh I think that's where it is [16:46] go sounding board go! [17:04] these ads....hilarious... http://www.autoblog.com/2014/06/27/colorado-washington-pot-dui-psa-ads-video/ [17:17] jujugui https://github.com/juju/juju-gui/pull/406 QA steps are a little more extended now. Works in sandbox, but requires a bit of a wait - feel free to do something else in the meantime. [17:21] Makyo: I'll take a look. [17:24] * rogpeppe is done for the day. [17:24] happy weekends all [17:25] have fun rogpeppe [17:26] cya rogpeppe have a good weekend [17:56] jujugui lf a quick review/qa on https://github.com/juju/juju-gui/pull/408 [17:56] rick_h_ any preference for another card for me? [17:57] hatch: sec looking [17:57] thanks [18:03] hatch: no preferences. take your pick. the bundle delay thing might be a short/sweet one in maint? [18:04] sure that sounds like a good small one [18:07] Makyo: branch land failed. npm bit the farm [18:07] Booo [18:11] jujugui just tried to walk a person in #juju through machine view to do some work. |-| far from getting it working like he wanted [18:16] rick_h_ just read the backlog.... his comment about wanting to drag to the container column was something I was thinking about too (remember the hover over machine to show containers?) I think the issue we are solving is one of massive scale, if you have 10-20 containers total we can easily show them all at once [18:16] hatch: yep [18:16] this is a UX discussion of course [18:16] without UX [18:16] :D [18:17] hatch: definitely, I think the main thing is that there's not much unknown/new. The lxc/container syntax threw him off [18:17] hatch: and the fact that machine 0 doens't show any services because it's node 0 and has gui hulk-smashed on it I guess? [18:17] The 'root level' threw him off. So we need to correct that [18:17] I never really thought of this, but if you were a noob and were presented with machine view, how would you know to click on each machine to see the containers [18:17] ? [18:17] and some things that are in trunk that he didn't have in his gui deploy [18:17] definitely a UX problem there [18:18] hatch: right, but that's why we've talked about having the column just say "select a machine to view container details" or something by default [18:18] right right [18:18] but maybe showing all containers is also a possibility [18:18] and then when you select a machine it filters the list [18:18] overall felt good but was interesting to grab someone off the irc street and have htem walk through it :) [18:19] I think I'll send that idea to peeps [18:19] hatch: do me a favor, grab that chat maybe and just put it in a pastebin and put it up? [18:19] hatch: and we can break it down from there at some point [18:19] sure [18:21] hmm my client does not copy/paste well heh, investigating [18:23] hmm and no logs [18:23] hatch: ok, np. I can pull it. [18:23] delegation fail! :) [18:24] if you could that would be awesome - the logs werent' enabled for that channel for some reason [18:24] and copy/paste fails hard because of the layout I guess [18:24] it puts each thing on a new line and skips names sometimes so it's very hard to read [18:30] hatch: https://gist.github.com/9f0f521c60ca7a15fd13.git [18:30] thanks [18:31] rick_h_ do you remember if we brought up to them already about having a onboarding message in the third column? [18:32] hatch: no, but I've got a UX meeting monday and will bring it up [18:32] ok so I won't send this email then? Or should I send it so you have something to reference? [18:32] hatch: you can send it, we can chat about it on monday [18:32] ok [18:33] start a conversation and all that [18:53] rick_h_ email sent, sorry it's lengthy [18:53] :) [18:54] jujugui lf a quick review/qa on https://github.com/juju/juju-gui/pull/408 [18:55] rick_h_: throwing the poor new guys at amulet in the first week. brutal. [18:55] bac lol [18:55] oo rah! [19:00] hatch: heh all good [19:00] bac: :) well I was hoping to find papercuts in charms that had amulet tests since they audit I thought there'd be a bunch [19:01] rick_h_: it'll just be funny when the ask the rest of us about amulet and we all shrug and say "dunno, haven't really ever used it." :) [19:02] bac: look ok overall? [19:02] rick_h_: yeah, nice [19:02] i think i'd be happy to get such an intro doc on my first day. [19:03] bac: cool, yea we'll see how it goes [19:03] it took me a year to get invited to all the right mailing lists! lol [19:03] that's half the fun! [19:03] haha [19:09] hatch: still need someone to look at 408? [19:10] I do! [19:10] thanks [19:10] What? [19:10] Oh… [19:10] No. [19:10] I didn't mean I was going to look at it. [19:10] ;-) [19:11] * rick_h_ steps away for a few [19:12] kadams54 hah damn uuuuu [19:15] hatch: shipit [19:15] dun! thx [19:37] jujugui looking for a quick review/qa https://github.com/juju/juju-gui/pull/409 [19:43] is lp or git the recommended way to file bugs against juju-gui? [19:44] * arosales doesn't see "issues" on https://github.com/juju/juju-gui [19:44] arosales, lp, I believe. [19:45] Makyo: ok, thanks. Are you still doing proj mang. and releases out of lp? [19:45] yep lp plz [19:46] we are [19:46] er [19:46] well the charm [19:46] hatch: you ok? [19:46] bac lol what? [19:46] arosales, hatch the charm is through lp, but can pull gui trunk from github now [19:46] hatch: well, that last flurry of comments made no sense. [19:47] * arosales will submit bug in lp -- thanks! [19:47] bac haha sorry [19:48] arosales: what's the issue? [19:49] rick_h_: I had a user report http://paste.ubuntu.com/7712762/ but I was uable to reproduce myself [19:49] so I was going to suggest he open a bug if it continues to plague him. [19:49] arosales: right, he filed a bug and I triaged it [19:50] arosales: it's FF only atm and filed but low atm [19:50] what the.... [19:50] heh [19:50] https://bugs.launchpad.net/juju-gui/+bug/1335192 [19:50] <_mup_> Bug #1335192: Drag and drop sidebar text to itself navigates in firefox [19:50] ah ok, good to hear it was filled [19:50] yep, actually someone I know...used to work for him ages ago [19:51] rick_h_ ohh ok - there is a e.stopImmediatePropogation() missing somewhere there [19:51] hatch: something with drop target/etc something [19:51] there was an issue with that in FF before with dropping on the canvas [19:51] hatch: yea [20:12] stepping out for lunch bbl [20:13] hatch: looking at PR 409 [20:15] jujugui: i'm toast so i'm calling it a day. have a nice weekend [20:15] bac: have a good weekend too! [20:15] bac: have a good weekend [20:18] jujugui quick review, slightly longer QA: https://github.com/juju/juju-gui/pull/410 [20:19] Makyo: I'll take a look in just a moment. You and hatch are killing it today. [20:26] Feels good to be off that stupid upgrade task for a day or two [20:47] I'm also going to run away. Have a good weekend folks! [21:13] yo [21:18] kadams54 isn't confirmation the religious thing [21:18] vs conformation [21:18] or did I mix those up? hah [21:18] yeah I'm pretty sure it's conformation :) [21:20] maybe it's confirmation [21:20] oh, confirmation is the religious thing and the act of confirming [21:20] hah [21:20] long week I guess [22:48] hatch: yeah, "confirm" is the root word, so "confirmation". Which is also a religious ceremony. Yay weekend! [22:48] lol