[02:58] huwshimi for reals?? [02:58] maybe Makyo can reproduce [02:58] :) [03:00] What. [03:00] hah you fell into my trap! [03:00] fired. [03:00] Just got back from Chicago :P [03:00] lol [03:00] have fun? [03:00] Ish? [03:00] well....thats....good? [03:00] James got laryngitis part way through so some of it was spent at urgent care. [03:01] oh damn, all good now? [03:01] Also, the hotel was an atrium (another Hyatt Regency, like October), so there was a lot of heights to deal with. [03:01] I know know now that I can ask for a low floor and they'll give it to me :) [03:02] lol [03:02] Yeah, just some antibiotics to clear it i[. [03:02] up [03:02] well that's good [03:02] Didn't look at a computer screen all weekend. I'm so proud of myself :) [03:02] ahh it's awesome when that happens [03:02] I brought the MBA and didn't even take it out of its bag. [03:03] haha nice [03:04] You should try and reproduce this bug for me https://bugs.launchpad.net/juju-gui/+bug/1254903 :) [03:04] <_mup_> Bug #1254903: Sidebar does not open if GUI is loaded in the background [03:04] Sure [03:04] yussss [03:05] I need it to happen to at least one other person so it doesn't get closed haha [03:06] Pulling. [03:06] oh you can do it on comingsoon [03:07] sorry I'll update the bug [03:14] hatch: I wasn't being serious :) [03:15] these are lies [03:15] you're all fired up with energy after your holiday [03:15] :P [03:16] hatch: Hey, I'm just happy I figured out a way to do animations that we haven't been able to do for over six months :) [03:17] no-no-no we were ABLE to do them, we just didn't want to do all the extra work to get them to work [03:17] lol [03:19] hatch, reproduced, chrome 31 [03:19] hatch: heh, yeah, we can also build a monkey. It'll just take us a little while to get all the different sciences worked out. [03:19] Makyo YUS VICTORY!!!!! [03:20] huwshimi those monkeys are a tough breed [03:20] * Makyo sneaks for the night. [03:20] cya tomorrow Makyo [03:21] 2 vs 2 then. I guess we need to sort this out then... [03:21] clearly you're doing it wrong [03:25] huwshimi so are you going to switch us to sass soon? :) [03:25] hatch: As soon as we upgrade Node :) [03:26] crap that's my job too isn't it? [03:26] hatch: Well, that's the blocker at this point [03:26] hatch: I think so! [03:26] well then [03:28] hatch: Have you been following Gary's authtoken branch at all? [03:28] hatch: Just trying to figure out what this means: "One of the things it does is that now the sandbox actually says that its name is "demonstration". This shows us that the notification boxes don't line up properly if they move. Do you have any brilliant ideas on fixes for this, by chance?" [03:29] huwshimi nope [03:29] :( [03:29] reading.. [03:29] hatch: Can the environment name change dynamically? [03:30] ohh I think I know [03:30] one sec, testing [03:30] Changing the name doesn't seem to break anything anyway [03:31] ok I was wrong... [03:31] I thought that he meant the notification when 'something' happens like deploying a bundle [03:31] but changing the environment name to 'super long environment name' and then making a notification still has the popup thing in the right place [03:31] and the dropdown 'list' also appears in the right spot... [03:32] at least this is on comingsoon [03:32] assuming that his branch landed? [03:33] hatch: It appears to be the same for his branch too. [03:33] hatch: Not to worry I'll ask for clarification [03:33] yeah...then I have no idea [03:33] hatch: Do you think there's any timeline for upgrading Node? [03:33] hatch: I know it's a "test and see if we can" kind of thing... [03:34] well...I can probably start on it after my branch tomorrow [03:35] It's probably important that we get it upgraded [03:35] there have been lots of changes with the charm so I'll have to coordinate with someone else too probably [03:37] hatch: Is there a specific reason you were thinking about SASS again? [03:37] source maps [03:38] ah right [03:38] beyond that, I think it offers some nicer syntax stuff [03:38] but that's just personal preference [03:38] I am not sure we will use the more advanced sass stuff [03:39] but it would be nice to use for things like your animations and whatnot [12:39] morning bac: could you please take a look at https://codereview.appspot.com/32760043 ? [12:42] sure frankban [12:43] bac: thanks === gary_poster|away is now known as gary_poster [13:16] frankban: the code looks fine, working on qa [13:16] bac: great thanks [13:58] frankban: following your (excellent) instructions i encountered this: http://paste.ubuntu.com/6479076/ [13:58] not sure why [13:58] bac: looking [13:59] bac: what does `which juju` return in the container? [13:59] doh [13:59] my custom go version [14:00] bac: I see, indeed juju-core was not installed [14:00] nope [14:00] so it uses the existence of 'juju' on the path to decide whether to install the package? [14:00] maybe it should do a 'which juju' and ensure it is /usr/bin/juju [14:01] bac: yes, the code uses "juju version" or similar. maybe we should make that check with the full path [14:01] frankban: yay, for my non-standard system to break qa [14:02] :-) updating the branch, so that you can continue the QA after pulling it [14:02] after removing my custom juju and re-running, it did install juju-core and re-used the ec2 env [14:07] bac: changes pushed, now juju is always used with the full path [14:08] cool [14:10] bac: delta here: https://codereview.appspot.com/32760043/diff2/1:20001/quickstart/app.py [14:11] frankban: looks good. i [14:11] cool [14:11] am just waiting for the qa to finish. not going to restart with new patch. [14:12] bac: sounds good [14:12] guihelp: I need another review (no QA) for https://codereview.appspot.com/32760043 ? Anyone available? [14:13] * gary_poster on calls [14:13] frankban: sure thing, sec [14:13] rick_h_: thank you [14:13] frankban: it seems it is trying to re-use an existing ec2, even though juju status does not show it started [14:14] frankban: i wonder if that is an effect of it starting with my custom juju? [14:15] frankban: aws console shows no instances running [14:15] frankban: but i get http://paste.ubuntu.com/6479151/ [14:17] frankban: destroying the (non-existent) environment caused quickstart to bootstrap a new one the next time [14:17] bac: could you please 1) juju destroy-environment -e ec2 2) pull the branch, 3) re-run quickstart. [14:17] heh\ [14:18] bac: for some reason juju got confused [14:19] bac: maybe a bug in your compiled version of juju [14:25] frankban: ok, qa done successfully. [14:26] bac: thanks a lot [14:26] frankban: np. [14:27] frankban: thanks for the good lxc primer. i may clip that and squirrel it away [14:28] cool :-) [14:31] bac: my last discovery was "sudo lxc-ls --fancy", which also returns all the ip addresses in the container's network namespace [14:32] um, --fancy is fancy [14:34] rick_h_: i tried to reproduce 1251420 on friday but had no luck. do you actually see that gigantoid error message in the notifications? [14:35] bac: yes, the value for "Error" is stuck into a notifcation with the text (something like) Deploy id: X failed. There was an error: bac: using that, it is possible to ssh into an lxc, for instance, with a function like http://pastebin.ubuntu.com/6479241/ [14:35] frankban: you duped that correct? Wasn't it from having a constraint set to '' ? [14:37] rick_h_: uhm... not sure if that generated the empty error... [14:37] frankban: that wasn't the empty error, that was the "casting a EnvError to string is fugly" error [14:38] rick_h_: to dupe that I suppose we want a bundle with invalid data, e.g. an invalid value for an option, or invalid constraints yeah [14:43] rick_h_: i modified one of my bundles to have an invalid constraint and dragged it locally onto an ec2 canvas. that should've aggravated it, no? [14:43] bac: in a live env? [14:44] bac: and did you get anything back? The websocket should show the traffic and that dump is from the websocket connection [14:44] rick_h_: yeah [14:44] rick_h_: i added '''constraints: "cpu-cores=" ''' to mediawiki [14:44] bac: I mean if you didn't get that, did you get something else? Did it work? [14:44] bac: that might read as undefined, try "cpu-cores=''" [14:44] rick_h_: istr that it did not work and i did not get the error. will spin up again [14:45] rick_h_: sorry, my eyesight is not good enough to parse that. [14:45] is that = [14:45] or = [14:45] bac: rgr, set the constraint to an empty string [14:45] single-single [14:46] sec, looking for the old rev of jcastro's bundle that did it [14:46] rt [14:46] http://bazaar.launchpad.net/~jorge/charms/bundles/jenkins/bundle/revision/4 [14:46] that was erroring, so he removed them entirely [14:47] bac: so maybe you can try to get his bundle at older revisions and use it [14:51] rick_h_: are you saying r4 was bad in the right way? [14:51] bac: I believe so [14:51] rick_h_: ok [14:51] bac: sorry for the vagueness of memory here [14:51] bac: but I know he fixed them by removing the constraints and it caused the env to blow up with an invalid value [14:51] rick_h_: your two suggestions are very good, thank you!. Is it ok if I create cards? one refers to pre-existing code (lxc) and the other seems out of scope for this branch (settings.COMMANDS) [14:51] frankban: rgr [14:51] rick_h_: cool [14:51] rick_h_: i'll try my changes to local muletrain and see what happens first [14:52] bac: yea, sorry I bet it's the right constraing [14:52] I think the cpu-cores is cast to an int before going to the client [14:52] so maybe that's why it's not failing for you [14:52] while the branch from jorge has all of them set to empty and not all were valid/cast to a number [14:58] sweet, so I get to add constraints back to all my bundles? [14:58] jcastro: yea, but they can't be those empty ones or it'll error. [14:58] they have to be legit [14:59] jcastro the ghost charm has been pushed to lp and a bug created to get it into charmers.....question though...do I need to re-submit it to charmers for every update? [14:59] I like the idea of having empty ones, makes it easy to figure out "oh neat, I can tune this" [14:59] 2 legit to quit [14:59] hatch, from then on you'll do a merge proposal and we'll autoget it in the queue [14:59] ahh gotcha [14:59] after X amount of times we can hand it over to you as a maintainer [15:00] but at first we like to peer review [15:00] I noticed that the 'owner' of the charm isn't anywhere in the jujucharms.com listing [15:00] should we not have a field to contact the owner? [15:00] huh, I thought we had the maintainer field showing up? [15:01] https://jujucharms.com/precise/mongodb-18/ [15:01] once they get recommended that goes away [15:02] ah, maybe that's on purpose then? [15:02] not sure - I think we should have it though to encourage discussion between the user and developer [15:03] I don't remember why we took it off [15:03] maybe to encourage people to discuss things like that on the general list instead of email? [15:04] hmm maybe [15:05] * hatch adds maintainer to the UI, doesn't tell anyone [15:12] hatch: stop that, it's a long standing argument [15:13] the maintainer and the cake is a lie [15:13] lol - well I can't help it if I'm right [15:34] gary_poster: ping [15:45] TheMue, hi [15:49] gary_poster: i'm currently working on the migration of the debug-log command from using ssh to the api [15:50] TheMue, ah, cool! [15:50] jujugui call in 10 [15:50] ooo nice! [15:50] gary_poster: here i wanted to ask you what you/the gui expects from the command [15:52] TheMue, great! Thank you. May I ping you after our team call and we can try to have a call? So, maybe in 30 minutes or so? [15:52] gary_poster: sure, will be here [15:52] hatch did you ever write your reply about the local charm integration? [15:52] thank you TheMue [15:52] gary_poster: yw [15:53] sorry I'll finish it now [15:53] cool thanks hatch, np. [15:58] jujugui call in 2 [16:40] TheMue, you available now? [16:40] TheMue, https://plus.google.com/hangouts/_/calendar/Z2FyeS5wb3N0ZXJAY2Fub25pY2FsLmNvbQ.j0rk5d371ph8331ijtf48t2uj0 [16:40] frankban, lmk when you have a moment for pre-imp [16:49] gary_poster would you mind seeing if you can reproduce this bug https://bugs.launchpad.net/juju-gui/+bug/1254903 it's very odd how different people aren't able to [16:49] <_mup_> Bug #1254903: Sidebar does not open if GUI is loaded in the background [16:50] hatch, k on it [16:54] Makyo: I am available now [16:54] hatch, can't dupe. also, we know from tests that some aspects of the code require that the browser be visible. That seems fairly reasonable to me. Even if we decide we can sufficiently dupe it, I'd be tempted to triage it low at the most. [16:54] frankban, https://plus.google.com/hangouts/_/7ecpi5n825fv8nflmovv6gqa4c?hl=en [16:55] gary_poster: the focus issues are tests around FF and the valueChange YUI event since it only polls when the input has focus. Focus shouldn't be an issue here unless specific browser versions were trying to cut corners by not rendering invisible bits while not looking a the page. [16:56] gary_poster yeah I am not too concerned with it as far as it being a bug goes - it's just a very curious bug with how not-repeatable it is for some [16:56] but we can't dupe in a few diff verisons/machines so seems hit/miss/too browser specific atm imo [16:56] rick_h_, hatch, ack [17:09] hey frankban [17:10] jcastro: hi [17:10] do we have a bug for allowing passing constraints from quickstart? [17:10] like, say I want an all in one ghost bundle [17:10] and I want everything on node 0 [17:10] but I'm going to want more resources than the standard default [17:11] jcastro: so those are bootstrap node constraints [17:11] right [17:13] oo that would be very cool [17:13] juju quickstart ghost all-on-one-superpowermode [17:13] oh sorry [17:13] juju quickstart ghost --all-on-one-superpowermode [17:14] :D [17:14] jcastro: quickstart does not support that. AFAICT provider constraints are specified in the envs.yaml file. We could add that functionality as part of our environment creation story [17:14] yeah because for small bundles, it still sucks to have a node just for orchestration [17:14] frankban, ok I'll file a wishlist bug then [17:15] I'm really happy we are now getting features for the little guy [17:15] indeed [17:15] the little guy drives adoption [17:15] that! [17:15] jcastro: hum, they can also be passed as bootstrap args, so a quick fix could be adding a --constraints arg to quickstart [17:15] "you think we did a good job with your ghost blog? wait until you see what we can do with your stuff at your job!" [17:16] haha [17:16] frankban, yeah so maybe just allowing quickstart to pass arguments to deployer underneath? [17:16] that also sounds like less work! [17:17] jcastro: well, to pass argument to juju, not the deployer. I have a question, what is the behavior of a bundle which deploys a service to machine 0 if the local provider is used? [17:17] I was not thinking of local [17:18] that is a good question [17:53] frankban, hatch, others, I'm planning to add these mock function and method test helpers to the GUI tests. feedback welcome. http://pastebin.ubuntu.com/6480086/ [17:55] cool I'll take a look in a bit [17:57] gary_poster I took a look real quick - it looks like you're replicating some of the functionality of http://sinonjs.org/ [17:59] maybe we want to implement that instead? [18:00] hatch, maybe so. ugh. I just want to get my branch finished. Adding the dependency is more annoying, unfortunately. Didn't you or someone else look into this once before? [18:01] we did - at the time we didn't have any real need for it :) [18:01] I'm not saying we shouldn't land your code [18:01] I just have a need in one test [18:01] just that we should probably look into this type of lib more seriously [18:01] because it could simplify things I bet [18:01] agreed [18:03] gary_poster, hatch: +1 those two functions look good, perhaps we could have another card to evaluate introducing a mock library [18:04] good idea +1 [18:04] thanks both [18:05] * hatch looks at the bug list, board, backlog..... [18:05] * hatch bows head [18:05] lol [18:05] :-) [18:12] * Makyo walk dogs. Maybe they'll stop being such brats :T [18:20] the relations need to move under the service model like the units did [18:20] not sure I agree [18:20] units are definitely children of services [18:20] relations have two masters [18:21] hmm [18:21] good point [18:22] well, that's a new one: "no reachable servers" [18:23] benji sounds like you are having a bad week [18:23] it might sound that way, but I'm enjoying myself at the moment [18:23] a service should have reference to the relation object.....that's how it should be done [18:23] right now they are too separate imho [18:34] sounds like a potentially nice change. [18:47] jujugui looking for review and qa https://codereview.appspot.com/33130043/ plz and thanks [18:49] and performance card gets pushed down again....poor performance card [19:09] forget that, I seem to have found a bug in the real env qa [19:12] if I'm working on the source in the charm how do I get it to re-build? [19:12] sorry I'm totally blanking [19:19] hatch: elaborate please? 'working on the source in the charm' you working on the gui source js from a live env? [19:19] yeah [19:19] hatch: you shouold be able to turn on debug mode to get it to use the debug files [19:20] that's what I thought too [19:20] then you can edit those and it'll be fresh on each page load [19:21] well the files in juju-gui-source need to be built somehow [19:21] and all of my my make commands in there just hang [19:22] well the charm takes care of the build and getting the uncompressed files into place. They're just in the build-debug vs build-prod [19:22] and no re-making should be necessary [19:23] yeah it's definitely not working because I put an alert at the top of app.js and nothing happens on refresh [19:23] the gui is also not hitting any of my breakpoints [19:23] which sources is it pull? [19:23] compressed all.js? [19:23] or each individual file? [19:24] the individual files are here under juju-ui [19:24] hatch: right, but in the network tab of your browser, which files are loading [19:25] all of them [19:25] each single file? [19:25] you're not being clear :P [19:25] when I try to put a breakpoint in, say the service-relations.js it puts it at the bottom of the file as if it hasn't been parsed [19:25] yes, every single file [19:26] ok, so then do a sudo updatedb and locate a filename and make sure you're editing the right copy [19:26] there's going to be a couple of hits in there [19:27] /var/lib/juju/agents/unit-juju-gui-0/charm/juju-gui-source/app/app.js [19:27] hatch: that's the only copy of app.js on the system? [19:27] there should be a build dir based one [19:27] well the other one is the release version [19:28] the build dir based one are just links to that [19:28] so /var/lib/juju-gui/release/juju-gui-0.13.0/build-debug/juju-ui/app.js is the file ot edit [19:28] but that's not the branch I'm using [19:28] oh [19:28] thats the issue [19:28] well it's the path in my deploy I've got here that is [19:29] adjust to your charm/deploy/setup [19:29] I used a different source directory and the charm serves up the release one on debug [19:29] well that's an odd issue [19:29] just keep editing app.js files until you find which one it's using :P [19:29] that would make sense as to why it's not being parsed [19:29] then file a bug on the charm after that [19:30] yeah... this is definitely a bug [19:30] if it's not acting like you're expecting, but there are only so many files to test which it's serving [19:31] yeah - it's serving the wrong files [19:32] well this will be interesting to debug [19:36] going to grab some lunch [19:36] bbl [19:40] hatch, not serving wrong files. Supposed to be elsewhere. [19:40] can talk later [19:47] oh reread. may be wrong, yes [20:08] hey gary_poster can we whack the juju-gui trunk series (https://launchpad.net/juju-gui/trunk) or change the charm to not recognize it? [20:08] juju set juju-gui "juju-gui-source=trunk" [20:08] does not do what you might think it does [20:08] and it is very disturbing [20:09] bac, I'm OK with it but would prefer if it were renamed, not removed. [20:09] but I've used it only rarely [20:09] old-stinky-trunk [20:09] so maybe ripping it out until we need it in anger again is reasonable [20:09] lol [20:22] jujugui: I have a charm branch up that adds randomized service names and tweaks the retry decorator at https://codereview.appspot.com/33230043/ [20:22] there are even some nice pre-review comments for whomever looks at it, going fast! [20:22] benji: i'll look at it [20:22] thanks [20:26] ahoy [20:27] ahah fixed [20:30] woot! https://github.com/CanonicalJS/juju-gui/pull/2 and http://ec2-54-196-48-122.compute-1.amazonaws.com:8080/job/juju-gui-merge/15/console [20:31] lookin goooood [20:31] sweet, rick_h_ ! I like the Juju GUI icon too ;-) [20:31] oh that's what that is? [20:31] hah, that's just a default github thing [20:31] lol [20:31] oh ok [20:31] but yea, I totally did that. I rock boss :) [20:32] heh [20:32] rick_h_ are the symbol's necessary? [20:32] $$merge$$ [20:32] hatch: the $$? [20:32] yeah [20:32] yea, I did that for now since the check is just a 'is $$merge$$ in the comment' [20:32] ahh [20:34] but yea, make a pull request, add a $$merge$$, I reload the page on the small web service, and it triggers a merge build, test, and that then does an api merge once it passes [20:36] benji: done [20:36] rick_h_ so if two branches are in the queue, A and B, and A fails, but B passes, will B get merged in ahead of A? [20:36] bac: cool, thanks [20:36] hatch: rgr [20:36] hatch: each runs on it's own, it merges develop (trunk) first and then runs tests [20:36] coolio [20:36] hatch: so basically A + trunk fails, but B + trunk works, then B + trunk gets merged and A needs more love [20:37] awesome [20:37] and jenkins makes sure only one runs at a time since it had to go through the single juju-gui-merge job, and that doesn't watch for changes, it had to be manually triggered via the web service which is the gatekeeper to develop (trunk) [20:38] ahh cool, so we require jenkins and a webservice? so two machines? [20:38] no, right now the goal is to run the webserver on the jenkins machine [20:38] it can run on port 80/443 or whatever since jenkins runs on 8080 by default [20:39] ahh ok - so can we run this on prodstack? [20:39] heh, not any time soon. It needs the latest jenkins and such. It's not charmed, lots of tweaking in the jenkins install to get it to work [20:39] right now the goal is to run it in ec2 land [20:40] ahh, well we should be able to run it on a free node [20:41] :/ slow tests? and eventually we'll have more things in here. Not sure on that [20:43] hmm that is a good point [20:44] ec2 is just pretty expensive for an 'always on' server [20:44] well, not for reserved instances and it'll be a lot cheaper than trying to get everything on prodstack [20:45] testing automated code pulls from github and such breaks a BUNCH of egress filewall, executing random code, etc rules [20:45] jujugui, I beg your indulgence of two reviews of this monster, with my apologies: https://codereview.appspot.com/33290043 [20:45] gary_poster, on it [20:45] it adds authtoken suport [20:45] thank you [20:45] I need one QA as usual [20:46] rick_h_ yeah I suppose you're right, a reserved medium is $50/mo a lot cheaper than the sysadmin/dev time to put on prodstack :) [20:46] maybe? :-) [20:46] ok biab [20:46] gary_poster: I'd like to review it but don't have enough time left in the day to digest that thing :) [20:46] haha I suppose that $50 never ends [20:47] benji, understood and thanks for looking [20:48] oh and I guess that's +bandwidth [20:59] ok jujugui https://codereview.appspot.com/33130043/ all fixed ready for review/qa [21:00] hatch, ok. trade you reviews? Please? :-) [21:00] wasn't Makyo doing yours? [21:00] hatch I most definitely need 2 :-P [21:01] well in that case you got-a-deal [21:01] cool thank you :-) [21:01]