[00:00] hatch, indeed [00:09] hatch: good to know, thanks === jcsackett_ is now known as jcsackett [06:52] mornin' all [07:06] rogpeppe: Morning [07:15] huwshimi_: hiya [10:00] morning rogpeppe, welcome back! [10:00] frankban: hiya! [10:02] rogpeppe: hope you had fun! I guess you have tons of emails to read, but please let me know when you have a minute to chat [10:04] frankban: yeah, had a great time thanks [10:05] frankban: am slowly getting through the emails [10:05] frankban: but happy to have a chat any time [10:20] rogpeppe: not sure I was connected: https://plus.google.com/hangouts/_/canonical.com/daily-standup?authuser=1 no rush [11:02] morning party people [11:27] rick_h_: morning. I started later this morning (economic consultant fun), will postpone my EOD today. I had a call with rogpeppe about current work. It seems we have found a plan to split filetesting, he will double check with core devs. I also created a card for Friday retrospective: it's just to share knowledge so we are can all better defend the decision to split store out of juju-core, rather than just importing [11:27] from juju-core [11:28] frankban: awesome sounds good [11:56] hi redir. [11:56] morning rick_h_ [11:56] morning [12:15] hmmm [12:16] morning. Looks like my computer was awake despite being closed... [12:21] you cannot stop working, go computer go [12:28] morning kadams54, when you get settled can you peek at hatch's branch in review please? [12:28] Sure. [12:28] kadams54: I'm looking at your between calls this morning, will try to get it through [12:29] rick_h_: I found a bug yesterday when QAing mine… you can see it in action at comingsoon.jujucharms.com [12:30] rick_h_: Switch to machine view and then deploy a bundle while in machine view. [12:30] kadams54: ok, please file and I'll triage and add a card to the board [12:31] rick_h_: Will do - it seems to be a state thing. When the bundle details panel slides open, it creates a new MachineViewPanelView instance without destroying the old one. When the details panel closes, a third instance is created. Then all three of these instances render machine tokens to the machine column. [12:32] gotcha [12:32] yea, that won't do :) [12:38] rick_h_: https://bugs.launchpad.net/juju-gui/+bug/1322158 [12:38] <_mup_> Bug #1322158: Multiple instances of MachineViewPanelView created [12:38] thanks kadams54 [12:55] rick_h_: we update the autocomplete short-circuit to do an ngrams search. So you shouldn't have to change anything for it to work [12:55] rick_h_: also we added it unboosted to the standard search api, so it should "help" with results there. [12:55] rick_h_: http://bazaar.launchpad.net/~reedobrien/charmworld/ngrams/revision/512 [12:56] note the different results http://staging.manage.jujucharms.com/api/3/search?text=crm&autocomplete=true vs. http://staging.manage.jujucharms.com/api/3/search?text=crm [12:57] redir: ok, I was just trying out some qa and sugarcrm was the original issue [13:00] OK rick_h_ so I think you shouldn't need to change anything in gui [13:00] redir: ok cool, will play with it some more when it's landed on prod [13:00] redir: thanks for the update [13:00] rick_h_, redir: so we are ready to move that version to prod? [13:01] bac: I think so, how's the restarting issue? [13:02] rick_h_: it continues to spin up new workers at a pace more frequently than expected. [13:02] rick_h_: new one two minutes ago [13:03] bac: ok, well I don't think the code is an issue so we can deploy and then maybe we should look at upping the worker limit in the run command? [13:03] IS told me they were throttiling the gsa but i have not gotten logs yet to see whether we're still being clobbered [13:03] rick_h_: yeah, perhaps more workers and a longer max-request [13:03] bac: and then we can keep an eye after that to see if we're seeing the same charms disappearing that makyo/hatch reported [13:03] 250 seems pessimistic [13:03] rick_h_: bac any particular reason we arent' using supervisord to manage workers? [13:04] bac: yea, well it was so low volume it seemed a nice safe bet :) [13:04] redir: gunicorn does it for us and manages things like this. Restart based on request counts/etc [13:04] rick_h_: i'd characterize that as another issue, like 'charmworld produces odd results under load' [13:04] bac: +1 [13:05] ok, first things first, i'm going to request the deploy of r514 to production [13:05] bac: sounds good, thank you [13:10] no vanguard in webops until the top of the hour [13:14] hi frankban, i've looked at urwid on OS X and it looks to work well, *except* it claims to not support unicode. urwid.supports_unicode() returns False. [13:14] get_encoding_mode() is 'narrow' [13:14] bac: uhm, that's weird [13:15] bac: perhaps we need to force unicode with set_locale or similar? [13:15] frankban: i see there is a set_encoding that may just work [13:17] jujugui one note about ngrams and autocomlete. [13:17] bac: we already call set_encoding in cli.base._check_encoding [13:17] jujugui the minimum ngram size is 3 [13:17] so sending 1 or 2 characters will never get a result [13:17] redir: cool that's peachy [13:17] frankban: actually, adding set_encoding('utf-8') in their graph.py produces ugly results [13:18] if you aren't setting min chars before sending query you might want to. [13:18] frankban: so it is an issue for us to be aware of [13:18] won't hurt if not just extra traffic and noise [13:18] brb workers in apt:| vacating to cafe for an hour or so. [13:18] bac: ack [13:22] rick_h_: i don't understand the issue for RT 71801 [13:22] in tracking wrt staging [13:22] bac: this is jujucharms.com having a prodstack staging [13:22] bac: the gui UI [13:22] rick_h_: ok, i see now [13:22] too many similarly named things [13:23] now that staging.jujucharms.com is free from its charmworld oppressor it's free to be a gui thing [13:45] bac is there a juju charm for CW? [13:46] redir: https://jujucharms.com/sidebar/search/~juju-jitsu/precise/charmworld-58/?text=charmworld [13:46] rick_h_: has there been any talk of quickstart on centos/other rpm-based systems? just thinking about linux/os x/windows may better be divided around packaging methods of apt | rpm | brew/pip | [13:46] redir: note that the ES charm used is not the normal one in the store. Check the doc and bundle for it's location [13:47] bac: not really, the talk has been osx and windows. Is juju packaged on other distros? [13:47] bac: I think it makes sense for sure. [13:47] there was some announcement lately about centos [13:48] rick_h_: we needn't do it now, but keep in mind. just saying 'linux' doesn't cut it when we mean an apt-based linux [13:48] bac: +1 [13:58] jujugui reminder to get time off in the system before the tomorrow since there's no system next week while they transition over [14:00] rick_h_: thanks, I should have a national holiday on June 2nd, will file it [14:00] frankban: cool thanks [14:01] redir: can you update kanban? I think your card in landing is landed and you're working on a new card in maint for the demo stuff? [14:01] Makyo: or hatch will ask one of you to run the standup. I've got the IS meeting that conflicts [14:01] sure I'll do it [14:06] bac: is this usable? bundle:~bac/charmworld-local/5/charmworld-local [14:07] rick_h_: sure [14:09] redir: otp. will check later. short answer: probably. [14:11] rick_h_: doesn't look like you created a card fro the create demo thing so I will. [14:14] Commuting to a coworking spot before standup [14:17] hey jujugui I still need two reviews on https://github.com/juju/juju-gui/pull/338 - kadams did the qa === hatch__ is now known as hatch [14:20] marcoceppi hey I put the ghost-charm up for review again [14:20] hatch: sweet [14:21] yeah this version is pretty awesome - I'll be putting together a blog post so that it can go out as soon as it's promoted [14:27] rogpeppe: did you get the invites and such to the daily standup in 30? and I think you should have access and such to kanban. [14:28] rogpeppe: let me know if any of that is missing to help get you setup [14:28] rick_h_: i haven't seen the invite, but i might have skimmed over it accidentally when on holiday [14:28] rick_h_: i'll have another look [14:28] rogpeppe: ok cool, hopefully it shows on your calendar [14:31] morning rogpeppe [14:31] hatch: hiya [14:42] rick_h_: replied to your feedback on the PR. Let me know how you want to handle switching over to the pluralize helper - new card or in this branch. [14:42] kadams54 hey are you also doing a review on my branch? [14:42] kadams54: if it's small this branch is fine as a drive by === hatch__ is now known as hatch [14:43] kadams54: if it goes a bit wide in the changes then a follow up is fine [14:43] rick_h_: it's on the large-ish side of small :-) [14:43] kadams54: hah, ok follow up cards then. [14:43] safer that way [14:43] hatch: yeah, I QA'd and reviewing now. [14:43] cool thanks [14:54] jujugui call in 6 [14:59] jujugui is anyone currently in the call? Getting a "This party is over" message. [14:59] I am [14:59] Using the tinyurl link… [14:59] kadams54: check the "authuser" when you get that [14:59] kadams54: I always just click on the calendar link [14:59] jujugui call in 1 [15:00] Yeah, right now I get a message that the calendar is in an infinite redirect loop [15:00] rogpeppe you'll want to set up your irc client to ding on 'jujugui' and 'guihelp' [15:00] :-( [15:18] rick_h_: whats a .io domain cost? [15:18] damn [15:21] name.com $100 and namecheap.com was $50 [15:21] I bought mine from name.com [15:21] heh [15:21] I found out after about namecheap [15:25] gandi.net has .io for $39 + tax [15:27] stupid bank of america snatches up all of the good 'bac' domains [15:30] redir: do you need a hand with the demo stuff? [15:33] bac I would have thought that bank of america would be boa, not bac [15:34] yep, bac is their ticker. lots of fun for me on twitter [15:34] haha [15:34] Makyo the notification error has a bug filed for it [15:35] the inspector under the headers....that will need a new card [15:35] at least I don't think there is a bug for it yet [15:35] hatch, okay, cool. was just about to ping. The sticky headers thing is new, but only because it's broken in a different way on comingsoon. I'm fine with it being a new card and moving forward with this. [15:35] awesome thx [15:38] bac sure [15:39] it creates an instance but then never connects to it. [15:40] redir: i need more context? g+? [15:43] hatch, bug created - https://bugs.launchpad.net/juju-gui/+bug/1322255 :+1: [15:43] <_mup_> Bug #1322255: Sticky headers render over inspector [15:44] thanks! [15:46] * rick_h_ goes to get food [15:47] TIL cd - [15:47] my life will never be the same [16:24] hatch: wait until I get you to zsh and autopushd popd :) [16:25] I've tried zsh a few times and ever time i run into conflicts with other apps, tmux, iterm, ect [16:25] one of these days.... [16:27] brb relocating [16:28] relocated [16:28] * rick_h_ hits the boot button [16:30] bac: can you file an RT for IS to look into the google search appliance traffic please? Per the call we had, they want to look at the config and check up on that. === hatch__ is now known as hatch === nottrobin is now known as cHilDPROdigY1337 === cHilDPROdigY1337 is now known as nottrobin [16:53] where's the bundle docs/ [16:53] ? [16:54] marcoceppi: ^ is there a spec for bundle yaml? [16:55] redir: kind of [16:55] heh [16:55] anything would help [16:55] marcoceppi: I am looking to add more than one charm to and environment, I think [16:56] redir: http://pythonhosted.org/juju-deployer/index.html https://juju.ubuntu.com/docs/charms-bundles.html [16:56] redir: it's best to just mock your deployment in the GUI then export it [16:57] marcoceppi: lemme rephrase, I want a bundle that deploys several services to one machine, can I do that in the gui? or in a bundle? [16:58] redir: you can do that in a bundle, but not in the GUI at the moment [16:58] marcoceppi: OK [16:58] redir: see the Juju Deployer pythonhosted link "placement" [16:58] * redir looks at the doc page that marcoceppi sent [17:03] its a chicken and an egg [17:04] must use gui to get nice canvas placement, must use cli to get machine placement atm. [17:04] * redir is a dinosaur [17:04] and bundles have to be in bzr atm? git not supported? [17:04] hazmat or marcoceppi or jujugui [17:05] correct [17:05] redir: they just have to be a yaml file [17:05] redir: I'd not put it in any repo atm [17:05] redir, git works [17:05] redir: this is not something to publish/share I think, but an internal use thing [17:05] redir, not for gui sharing.. but bundles support git charms [17:06] rick_h_: will juju quickstart bundle:.... support local FS or only repo? [17:06] redir: only repo, there's still the issue with local charms there. [17:06] oh, redir you can just juju quickstart filename [17:06] you don't need the bundle: bit [17:06] * redir blinks [17:07] juju-quickstart bundle.yaml [17:07] so juju quickstart ~/demo/bundles.yaml [17:07] yep [17:07] OK [17:07] great [17:07] tx all [17:07] you can have charms in git now? [17:08] aww man, I spent a bunch of time submitting the ghost charm to lp last night [17:08] hatch: no, that's the store work we need to do after we split it out [17:08] hatch: well it's part of it at least [17:08] hatch: he's talking about a format in the bundle that deployer can use to fetch the charm in a bundle [17:08] ohhhhhhhhh [17:08] * hatch shuts up [17:09] hatch: http://blog.chromium.org/2014/05/chrome-36-beta-elementanimate-html.html :) [17:10] rick_h_ yeah I saw something about that last week - it's going to be pretty awesome when we can use Object.observe [17:10] yep [17:10] * hazmat hungers for chrome 36 and object.observe [17:10] heh, we're all drooling over the same candy [17:10] everyone is making a big fuss about how awesome arrow functions in js are gona be.....I really could care less - I wish they would spend that time on doing things like finishing generators [17:11] no kidding, I hate arrow functions [17:11] like woopiedoo I get to save a few keystrokes [17:11] and make the code harder to read [17:11] there's enough noise in there [17:11] :) [17:11] haha [17:11] crap, hatch and I agreeing...runnnnnn! [17:11] we agree on something....shit [17:11] rofl! [17:12] * redir backs away slowly [17:12] redir it's just history haha [17:17] * rogpeppe is done for the day. [17:17] g'night all [17:18] cya rogpeppe [17:18] later rogpeppe [17:20] ok, redir i'm back if you have questions about that bundle, doc, etc [17:29] rogpeppe: have a good night [17:38] bac sure [17:38] holler if you want to hangout [17:39] I have lots of questions. [17:42] redir: ok. just a sec [17:50] can I deploy a bundle without quickstart? [17:51] redir: you can use juju-deployer [17:51] redir: yes you can use the deployer package or drag/drop on the gui (but no colo support) [17:52] redir: hangout? [17:52] ringing [17:52] redir: can you paste url [17:53] https://plus.google.com/hangouts/_/g2kxutye67iyynydgwuu5flxwia?authuser=3&hl=en [17:53] bac ^ [18:00] bac http://pythonhosted.org/juju-deployer/config.html#placement [18:00] farmers must be spreading the manure today.....peeeeeeeuuuuuuuu [18:07] jujugui fyi http://www.html5rocks.com/en/tutorials/developertools/async-call-stack/ [18:14] I've been having mixed results with that [18:14] sometimes the stack trace gets lost in yui land [18:14] sometimes it shows where the event was attached/triggered which is nice [18:16] yea, I keep meaning to try it out but haven't seen how it works [18:19] rick_h_ that page was using 20% of 4 cores of cpu to run in the background :O [18:58] Makyo: how's that branch in review? Did that get duped or dropped? [18:58] hatch: moved your card to done, are you on a follow up now? [18:58] rick_h_, I'm working down a different path, running into double-dispatch stuff. [18:58] Makyo: :( with relation lines? [19:00] rick_h_ yep, just getting back on it after lunch [19:00] rick_h_, I think so. This is such a mess. It's like the relation code doesn't understand that the subordinate services are subordinate until they receive some data from a delta (i.e. receiving annotations from moving a service). currently tracking it down to app.js and the dbChangedtimer, but I'm not even sure that's it. [19:00] hatch: cool, can you update the board when you get settled? /me is catching up from wtf happened today a bit :) [19:00] just did [19:01] hatch: ty [19:01] Makyo: ok, let me know if you want another set of eyes. I've got a break in 20ish I can help look if you need [19:01] rick_h_, alright, thanks. Will keep poking [19:02] heh or maybe sooner if kadams isn't around [19:03] heh [19:05] howdy kadams54 https://plus.google.com/hangouts/_/canonical.com/kyle-rick?authuser=1 [19:05] I think I'm already there? [19:22] oh man so hot outside 25C [19:41] Yesssss [19:41] I think I got it. [19:41] wooooooooooot! :) [19:41] ok, then I'm going to take a break before the AU call. biab all [19:42] 10 calls down, 3 to go [19:42] yay [19:46] At least it wasn't actually double dispatch, but what I thought before: a problem with bundle files not including subordinate information. [20:07] jujugui QA/review: https://github.com/juju/juju-gui/pull/340 [20:09] awesome, not a double dispatch bug :) [20:09] I'm knee deep into this refactoring so hopefully someone else can get on it [20:10] it's small, promise! [20:11] Er, others, not hatch :) [20:11] I'm trying to figure out why we have 19 recommended 'apache' charms and 1 more charm [20:14] oh I broke it [20:14] that's why [20:14] Root cause analysis: hatch. [20:15] ding ding ding ding [20:25] what would we call our 2 week 'sprints' ? [20:26] calling them sprints is a little confusing...but that's sort of what they are in the agile world [20:32] * Makyo steps out for a few. [20:35] hatch: you have a window box up? [20:36] negative [20:36] I COULD if you needed [20:36] we may have one in the house... [20:37] haha, I just have to tear down my vagrant instance to spin up my win 8 one [20:37] so lemme know if you need [20:37] eod [20:37] mongo not starting and I haven't figured out why [20:38] so manana maybe bac or someone can make sure I am not the error [20:45] redir: ok, does 'juju status' say anything interesting [20:55] bac no but here's stuff if you're bored [20:55] http://paste.ubuntu.com/7503032/ [20:55] bac, it jus tdiesnt' start (seems like perm error), no log nothing [20:56] but I can manually start it as mongodb with the installed config and it starts fine [20:56] walking out now [21:06] will try to look later redir [21:10] bac I am going to reboot now, will you need me to look at anything in windows? [21:51] so...many.....cases [21:59] hatch: cases of wine? [22:05] hah, edge cases in this sidebar [22:24] Morning [22:28] morning! [22:30] is the aus call right now? [22:30] late 1 sec [22:31] ok I'll just hang out here [22:31] messing with the effects [22:33] joinin [22:59] oh that was a weird bug [22:59] heh [23:02] the home button works again as a side effect [23:02] boo yea [23:15] hatch: That may be, but what did you break in the process? [23:27] huwshimi haha, I guess we'll see! [23:45] :P [23:57] bac: howdy