[00:17] * Makyo is away: EoD === BradCrittenden is now known as bac [13:01] rick_h_: progress on setting up staging on prodstack [13:01] bac: yes, kind of [13:01] bac: we're #2 on the list and got a priority bump and should be coming up 'very soon' [13:02] bac: I'd expect to see something next week [13:02] rick_h_: did you see liam's note [13:02] bac: no, link or reference? [13:02] rick_h_: email fwded [13:03] bac: thanks, just realized my email sync went belly up. No wonder it was so quiet today :) [13:03] sync'ing down now [13:04] bac: ah cool, can you file a bug on that install issue, or at a maint card for us to update that? [13:04] * rick_h_ is surprised we haven't hit that issue already [13:05] did [13:05] bac: ty [13:05] rick_h_: dn [13:06] * bac internationalizes responses [13:13] hey rick_h_, how do I deploy a multiple bundle bundle [13:13] juju-quickstart: error: multiple bundles found (scalable, single) but no bundle name specified [13:13] jcastro: you don't [13:14] jcastro: you can only deploy one at a time [13:14] yeah but what's the format? [13:14] it's why the gui breaks them up in the results [13:14] oh, sec [13:14] jcastro: -n NAME [13:14] juju-quickstart bundles.yaml/single isn't one [13:14] jcastro: so juju deploy -n single bundles.yaml [13:14] got it, thanks! [13:15] * frankban lunches [13:29] hey benji, can i borrow your eyeballs for a sec? here yaml.dump is behaving oddly and i can't see why. http://paste.ubuntu.com/7090137/ [13:29] * benji starts the eyeball meter running. [13:29] specifically lines 21-23 [13:33] bac: what do you perceve as an oddity on those lines? [13:34] benji: i would've thought at 21 options would have had tuning-level as an entry on a separate line with no braces [13:35] at 22-23, items broken out rather than presented as a dict [13:35] benji: you know, yaml [13:35] rick_h_, ok, a new problem [13:36] jcastro: awesome [13:36] http://paste.ubuntu.com/7090182/ [13:36] yeah, it doesn't seem to be as low-symbol as possible [13:36] this bundle passes proof, it's the same one I made but combined [13:36] but when you deploy the "scalable" one, only haproxy, kibana, and the gui [13:37] also, on a related note, what's the workflow for people drag and dropping combined bundles into the GUI? [13:37] jcastro: right, they were getting an error about subordinates yesterday? Is that not part of this? Was there any error feedback in the gui? [13:37] yeah trying it now [13:37] jcastro: there isn't right now. The idea would be to add a selector when you drop, but right now you can only drag/drop single bundle files [13:38] man [13:38] jcastro: but it's not on the schedule at the moment [13:38] why didn't anyone mention that before I combined the bundles? [13:38] (near term schedule) [13:38] because they're getting promulgated [13:38] and are available in the gui for deploy from the UI [13:38] or via quickstart [13:38] so there's no need to drag/drop them? [13:39] bac: are you wanting a lower-symbol representation or are you worried that the resulting YAML may not be parsed correctly? [13:39] so the gui will support the multi-bundle when they're in the store? [13:40] jcastro: it splits them out [13:40] oh ok, awesome [13:40] jcastro: it's why you have rails:single rails:xxxx [13:40] benji: i don't think it will be parsed as yaml. why are some dicts expanded and these two not? [13:40] jcastro: it's why you get the prettier names [13:40] benji: i don't know what you mean by lower-symbol [13:41] it will be parsed fine; remember YAML is a superset of JSON, so those mappings using curly braces are just fine [13:41] >>> yaml.load(yaml.dump(data)) == data [13:41] True [13:41] (where my data is the same as yours) [13:43] benji: ok, then, i'm bugged by the inconsistency. :) [13:44] yeah, I'm not sure why it goes symbolic on those; if we really want it to look nice it looks like we can create our own serialization class that will pretty-print the output [13:45] yow! whatever you do, don't pass canonical=True [13:45] rick_h_, ok same thing, no errors reported by the GUI though [13:45] http://paste.ubuntu.com/7090229/ [13:45] jcastro: k, and to test it in the gui you removed the single and went only with scalable? [13:46] jcastro: hangout? [13:46] sure, fire it up! [13:46] jcastro: https://plus.google.com/hangouts/_/7acpjonbnmmh5a1hkd5fr81emg?hl=en [13:46] bac: ooh, it looks like default_flow_style=False makes for pretty YAML [13:47] benji: yep, i just discovered that [13:47] I vote for canonical=True just to make it look all technical-like. [13:48] i mean if we can't use canonical=True who can? [13:49] yaml.dump(data, mismash=False) [13:49] s/mismash/mishmash/ [13:50] heh [13:51] benji: i yield your eyeballs. thanks. [13:51] :) [13:51] "I yield my remaining eyeball time to the distinguished gentleman from NY" [13:52] rick_h_: heh, I thought the same thing. [14:01] * Makyo is back (gone 13:44:06) [14:03] that timer is pretty cool :) [14:03] Hah, yeah [14:21] guihelp: I need two reviews + QA for https://codereview.appspot.com/75830045 (quickstart). Anyone available? thanks! [14:21] frankban: in a bit [14:21] rick_h_: could you please check if ^^^ meets Jorge needs? [14:22] bac: thanks [14:22] frankban: looking [14:22] rick_h_: thank you [14:23] frankban: can you verify that bundle:mediawiki/single [14:23] works as well? [14:23] rick_h_: sure [14:24] frankban: and if so let's use that as the bundle: shortest url you can get example [14:24] 159 failing tests... [14:25] rick_h_: that url is already used as example in the help [14:25] hatch__: stop doing that :P [14:27] rick_h_: http://pastebin.ubuntu.com/7090449/ [14:28] frankban: woot thanks. Thought so but brain fried today [14:35] rick_h_, should I file a bug on the num_units:0 thing? [14:35] jcastro: ^^ I'll get it. I think we'll need a couple on the right projects. I've got a card for us to look into it === hatch__ is now known as hatch [14:37] frankban: feedback inbound [14:38] rick_h_: thanks [14:48] rick_h_ nice spreadsheet [14:49] hatch: heh, I still can't get that damn planning poker site to work right [14:49] so this will be the next best thing [14:49] everyone enter a number....NOW [14:49] that's fine, that system was a little too 'hands off' anyways :) [14:49] and then agree on a final one [14:50] speaking of [14:50] jujugui overly long standup in 10 [14:50] kanban and fetch your beverages now please [14:51] rick_h_: did you QA frankban's branch? [14:51] bac: no, just reviewed text as jcastro's proxy [14:52] ok, i'll do it later [14:52] bac: ty [14:53] thanks bac [14:58] jujugui call in 2 [15:00] I love it when tests fail but the code it's testing works [15:01] hatch: call time [15:01] whaaaa [15:13] benji: cw review wygac https://codereview.appspot.com/76100043 [15:57] rick_h_, hey so since bundles don't have a way to feature them in manage, is there a way to manually toggle something to feature them? [16:00] jcastro: Is your bundle charm school thing later going to be recorded? [16:00] jcastro: gave you a url yesterday [16:02] luca_, yes [16:03] BradCrittenden: I'm looking at your review now. [16:03] thanks [16:03] nick change in 3...2...1... === BradCrittenden is now known as ba === ba is now known as bac [16:04] rick_h_, I must have missed that URL [16:04] rick_h_ frankban when we have that juju-core work could we plan to pair on it so I can get more up to speed on how juju-core operates? [16:05] jcastro: sorry was otp, sec [16:06] jcastro: http://manage.jujucharms.com/bundle/mediawiki/single/featured/edit [16:06] jcastro: we've got a bug to add the actual html link, sorry for the url hackery for the moment [16:07] hackery is good enough for a friday! [16:07] :) [16:11] bac: frankban Makyo email on the way with shared docs for monday [16:11] thanks [16:12] hatch: you mean, adding fields to machine info? [16:12] hatch: pairing sounds good [16:12] Thanks [16:12] rick_h_, ok mediawiki was no problem [16:12] any idea why this isn't working? [16:12] http://manage.jujucharms.com/bundle/mongodb/cluster/featured/edit [16:12] I get a 404 [16:12] rick_h_, oh nevermind, we just promoted them, so we need to wait for ingestion [16:12] jcastro: rgr [16:13] rick_h_, how long is ingestion again? 1 hour? [16:13] 15min [16:13] oh, it takes about another 15min [16:13] so normally don't bug me until 30min past :) [16:13] heh [16:16] frankban yeah - awesome thanks! [16:17] mp [16:18] * rick_h_ runs away to lunch outside of these basement walls [16:34] rick_h_: can you send me a debug log to look at? [16:45] * rick_h_ is back from lunch fetching [16:48] rick_h_: how does it look? http://pastebin.ubuntu.com/7091117/ [16:48] frankban: <3 thanks! [16:48] cool [17:15] sweet down to 142 test failures [17:15] * hatch grumbles [17:15] lol [17:19] rick_h_ from installing Ubuntu on your air was there a specific tutorial you found most helpful? [17:19] I think I'm going to take a stab at it this weekend [17:19] hatch: not really, just use refindit or whatever and plug in the usb device and boot from it [17:20] cool ok [17:22] I even have me a nice jelly ubuntu sticker to go over the glowy apple logo :) [17:34] frankban: still around? [17:34] rick_h_: yes [17:34] frankban: the two cards in project 1 those are unblocked now correct? [17:34] frankban: the mega-watcher related ones? [17:36] rick_h_: well, they could be considered blocked by the juju-core one. not tecnically blocked, but in theory it could be better to wait for the server side to be implemented [17:36] frankban: ah, gotca ok [17:37] frankban: I know of all-watcher but not up on mega-watcher :) [17:39] heh [17:40] jujugui I've updated the kanban with the cards for the next cycle. Some stuff is not scheduled. Take a peek and let me know if you disagree on anything left behind. [17:40] WATCH ALL TEH THINGS!!!!!! [17:40] no no, really watch them hatch [17:42] looks good! [17:50] jujugui running away a bit early today. Will be in/out if anyone needs anything. [17:50] enjoy your weekend! [17:50] bac: are you qa'ing frankban's branch and you've got a review on yours? [17:50] rick_h_, rails:simple shows up in the search results [17:50] but not rails:scalable [17:50] * rick_h_ should probably count those today [17:50] rick_h_: i am doing the qa [17:51] nm, I'll file a bug [17:51] jcastro: looking [17:51] have a good weekend! [17:51] rick_h_: yes, benji reviewed mine [17:51] bac ok cool [17:52] rick_h_: meant to mention i've got no approval for travel yet. mramm too busy chasing sloths [17:52] jcastro: yea, please do I guess. Nothing left in the queue and appears to not have gotten in [17:52] jcastro: maybe swap the order and see if that does anything different in the mean time? [17:52] bac try sending an email [17:52] I am not going to touch the bundles until past monday :) [17:52] bac: ok, if you don't hear anything monday let me know. I'll see him in a meeting and can bug him [17:52] jcastro: heh ok [18:08] I have 3200 read emails in my inbox...I think I need to start making folders [18:09] the scrolling is starting to lag lol [18:13] frankban: working through your QA. all good so far. a few more to go. [18:14] bac: thank you! [18:17] jujugui I need two reviews/deep qa's for the final viewlet to view branch https://github.com/juju/juju-gui/pull/180 [18:17] hatch, on it [18:18] thanks [18:20] * hatch cracks knuckles....."glad to be done that" [18:26] hmm not so fast I guess [18:26] CI failure....this is an odd one [18:27] hatch: missing a file in checkin? [18:27] hatch: or something that would cause it to not have the file? [18:27] checkin is good, checking prod now to see if I missed something in a make step [18:28] devel passes no problem [18:28] might be the modules dep issue again heh [18:28] ahh yeah prod is failing [18:29] hatch: cool [18:29] well, cool that it's not CI but a real thingy [18:29] yup, now to just get prod to not crash on me so I can figure it out :) [18:29] phantom keeps crashing....it's running too fast! [18:29] haha [18:30] oh boy now phantom won't even start [18:30] lol [18:35] fixed...I missed a dependency...oops [18:36] jujugui anyone else available to be another reviewer before EOD ? [18:36] https://github.com/juju/juju-gui/pull/180 [18:36] hatch: looking [18:36] cool thanks [18:37] hatch: the lower casing is now that they're all moved you've renamed them? [18:38] rick_h_ well the classes are still upper cased but their keys are now lowercased [18:38] (as they should be) [18:38] k [18:38] I mean, I could change that I suppose, but this I think follows closest to the conventions [18:39] k, yea just odd at first to just start out lower casing things [18:39] reviewer comments helpful :) but I guess I'll see something that auto upper cases the first letter somewhere in a bit [18:41] benji: thanks for the review. i changed "orig_yaml" to "o_y" but then opted for "o" [18:42] lol [18:42] oh boy hah [18:49] yay, ACC tourney [18:49] thanks espn app [18:52] hey benji have you seen any spurious failures of charmworld/views/tests/test_metrics_api.py ? it failed a few times via 'lbox propose' but worked in isolation. [18:53] super annoying [18:53] and it is a pretty dumb test. not sure how it could be failing [18:54] bac: I saw that yesterday; I should have filed a bug or at least mentioned it. A card to look into it might be on order. [18:54] 404 not thrown? [18:54] benji: and was it the same, failing only on lbox propose? [18:57] so I can see the hadoop cluster bundle in the ~charmers branches but I can't find them in the GUI itself, ideas? [18:58] http://manage.jujucharms.com/search?search_text=bundle&op= not ingested yet? [19:01] rick_h_ I replied to all your messages, mind checking my responses plz [19:01] hatch: rgr [19:01] jcastro: just saw ingest fill up with fresh load of charms/bundles [19:01] will watch it go down and see if it apepars [19:02] its in there and I can quickstart deploy it [19:02] I just can't find it in jujucharms.com [19:02] bac: I wonder if that scoring is biting us? [19:02] so I search for say "cluster" and mongodb/cluster shows up, but hadoop/cluster does not [19:02] rick_h_: could be. i'd think a direct search would show it [19:02] bac: if you get a sec can you check staging to see about http://staging.jujucharms.com/search?search_text=hadoop&op= [19:03] bac: they show up in staging but not production :/ [19:03] oh [19:03] tmks [19:03] er, tmns [19:04] hatch: thanks for the udpates [19:05] jcastro: sorry, seems an issue on our end. Shows on staging but not manage. looking into it [19:06] no worries, quickstart works so it's not so bad [19:06] I was worried things were getting unpromoted [19:06] no, they're there. The joys of search. Eternally tweaking results to get enough results, not too many, sorted best, etc. At least that's my first impression. [19:07] http://manage.jujucharms.com/bundle/mongodb/cluster/featured/edit is 403 [19:07] am I doing it right? worked fine for mediawiki [19:08] jcastro: 403 means not authed [19:08] ta [19:08] make sure to go to the main page and logout/login [19:08] got it! [19:11] rick_h_, after I feature unfeature I need to wait for next ingest I assume? [19:11] jcastro: no, shouldn't [19:11] jcastro: when did you push up that cluster bundle? [19:11] bac, earlier today [19:11] jcastro: I see two bundles in featured atm [19:12] around 11ish I wanna say [19:12] jcastro: make sure to remove the charms so the list isn't too long [19:12] yeah I did that [19:12] jcastro: https://manage.jujucharms.com/heartbeat shows no bundles queued up. i'll wait to see what happens when all the charms get ingested but it looks like the bundles aren't getting queued up. [19:13] bac: I saw 20 queued up a little bit ago [19:13] http://staging.jujucharms.com/heartbeat -- shows bundles queued up [19:13] rick_h_: on mjc? [19:13] and they go through [19:13] bac: yes [19:13] bac: I watched mjc for two ingest cycles and saw them go through [19:14] that's weird. i wonder if our cronjob is foobar-ed . will ask webops to grab it [19:14] bac: rgr, thanks [19:15] ok I've set all the featured/unfeatured for everything in manage [19:15] rick_h_, I betcha in 15 minutes they'll show up right [19:15] jcastro: :/ I was pretty sure that was instant [19:18] rick_h_ can we chat about the two 8's in Project A and split them up? [19:19] rick_h_: https://manage.jujucharms.com/heartbeat -- it is ingesting bundles before charms. that is different than i remember [19:19] I know you've EOD'd so that's ok if not :) [19:19] rick_h_: also API3 Interesting fail [19:20] jcastro: feature/unfeature is not related to ingest processing [19:20] hatch: sure link me [19:20] bac: yes, I noticed that as well [19:20] https://plus.google.com/hangouts/_/76cpj8qivva5nfi81adp78a4s4?hl=en [19:21] rick_h_: note staging is working charms then bundles, as expected http://staging.jujucharms.com/heartbeat [19:21] bac: :/ [19:24] rick_h_: good news is the cronjob is correct after deploy. so perhaps that issue is behind us. [19:24] unfortunately not sure what is going on [19:38] rick_h_, benji: we pass the ingest jobs (charm and bundle) to the worker as a dictionary and then loop over the values. so order is not predictable. so that could explain the difference we see right now between staging and prod [19:39] bac: ok, that's cool. jcastro you had other ones promoted today right? I mean the deploy was yesterday so the code's not changed today [19:40] if the insertion order is the same and the machines are the same architechture, then I would expect the order to be the same in the dictionaries [19:40] the API3 failure is odd since it hits https://manage.jujucharms.com/api/3/search/interesting , which works [19:41] the message heartbeat gives indicates an exception (unlogged) was caught [19:41] I just got a 500 on that url [19:41] rick_h_ lol I cannot find that darn doc [19:42] huh, me too now. worked a minute ago [19:42] bac: apache cache? [19:42] hatch: https://docs.google.com/a/canonical.com/document/d/1j2Tyapbhd8Zf3wNeMgdOGBIRYRmKfpzSYe47lfm86lM/edit [19:42] bac: so we should have logs to get now that we're hitting 500s [19:43] starred [19:43] heh thanks [19:43] true [19:43] I have to drop the dog off for boarding for the weekend. biab [20:16] rick_h_ where is the best place to download 14.04 from? === hatch__ is now known as hatch [20:42] rick_h_, benji: i've got the logs and see the reason for the 'interesting' failure -- keyerror. not sure why bundles aren't ingesting, though. [21:01] bac: k, do we have a cowboy for the keyerror? [21:02] bac: or can we unfeature/pull whatever is causing it? [21:03] rick_h_: i'm still investigating. the immediate code has not changed since 130 revs ago [21:03] that's a lot of revs :) [21:03] rick_h_: oh, it may be b/c we're just not featuring bundles! i'll bet that is it [21:03] have jorge turn those off for now [21:03] s/not/now/ [21:04] bac: yep, that's what I was thinking [21:04] k [21:04] jcastro: around? have a list of the bundles you've featured? [21:04] '_source' does not appear as a key for bundles [21:04] bac: k, makes sense [21:08] bac: k, think I've unfeatured them all [21:09] bac: but wondering if we're hitting the request cache we did earlier [21:09] no 500 [21:09] ? [21:09] http://manage.jujucharms.com/api/3/search/interesting is still 500'ing for me atm, but [21:09] I've unfeatured them [21:10] rick_h_: can't you do something with curl to set a no-cache header [21:10] bac: hmm, the settings don't save for me? [21:10] bac: can you login and hit http://manage.jujucharms.com/bundle/mediawiki/single/featured/edit and see if it's still checked? [21:10] I've unchecked it 3 times now [21:11] bac: ah ok, it unchecked in another browser [21:11] Makyo hey need any help with the inspector branch? [21:12] hatch__, I think I got it, I just ran into conflicts with your last few. having to go back and redo some stuff to match style, which is fine. [21:12] ohh ok cool [21:12] rick_h_: good. wouldn't uncheck in safari [21:12] Should get it in late today or Monday. REalizing how little I've worked with the inspector stuff. [21:12] yeah sorry so much diff [21:12] :) [21:12] much diff, very changing [21:13] That's A-OK. it's a good direction to change in! [21:13] now do I download the 14.04 beta 1 or the daily iso....this is the real question [21:13] yeah I Agree this is the right way to do it === hatch__ is now known as hatch [21:14] rick_h_: could you feature a bundle on staging and see if 'search/interesting' 500s? [21:15] bac: http://staging.jujucharms.com/api/3/search/interesting [21:16] bac: must be the bundle that was done? It did work. I had two promoted bundles earlier in the day when jcastro was asking about what was taking his other bundle so long to show up [21:18] bac: http://manage.jujucharms.com/api/3/search/interesting [21:18] bac: it's the hadoop bundle [21:18] bac: that one that fails to ingest on manage, but is on staging [21:18] bac: it was featured manually, but that caused the breakage [21:19] rick_h_: you got an url for the bundle? [21:19] anything obviously wrong? [21:19] bundle:~charmers/hadoop/7/cluster [21:19] bac: not looked yet, re-featuring a couple of bundles to get the content fed [21:22] rick_h_: i don't see anything obviously wrong with http://bazaar.launchpad.net/~charmers/charms/bundles/hadoop/bundle/view/head:/bundles.yaml [21:22] bac: yea, I mean it ingested, I think I ran it through proof when it didn't show up [21:22] at least on staging [21:23] so not sure atm [21:23] rick_h_: it may have not gotten ingested properly on prod [21:23] bac: right [21:25] hey rick_h_ are you running the daily or beta version of Trusty? [21:25] hatch: I installed the daily iso [21:25] and it's been updating nicely? [21:25] yep [21:25] do you know what the update from the daily to release will be like? [21:25] it 'should' be seamless no? [21:26] bac: damn, interesting is now a 4s request with bundles in it :( [21:26] hatch: yep [21:26] bac: but it's not broken now at least [21:26] bac: so we should file 3 cards. One about the hadoop bundle, maybe try to find a way to clear/reingest I don't know. [21:27] one about the 4s load time on interesting [21:27] and one about, something I can't think of now. gah [21:27] rick_h_: we should be smarter when creating interesting and reject bad entries not 503 [21:28] rick_h_: i'm working on that now. not sure if that's what you meant by card 1 [21:28] bac: no more mean that the hadoop bundle ingests on staging but not production. Something is wrong there that needs a reset at the min. [21:28] rick_h_: i also have changes to make the ingest order definitive [21:28] bac: but that's fine, interesting is an api endpoint that cannot fail really. So all for making that more solid [21:29] rick_h_: our bundle deleter would come in handy now [21:29] bac: :/ yea [21:29] bac: I guess we can push to try that first. Just avoid the known bad bundle for now [21:29] wheee launches [21:29] rick_h_: so are we good for now? [21:30] damn feature's been there for 6mo and finding all the bugs in 4 days [21:30] if so i need to take off [21:30] bac: yea, I think so. [21:30] bac: thanks, let's regroup on monday and see what we can do. This will be important to get fixed asap with the announcement [21:30] rick_h_: ok. log files are chinstrap:~bac/files.tar [21:30] bac: rgr thanks [21:30] if you're interested [22:10] oh boy I found a markdown compiler for sublime [22:10] this is awsome [22:11] hatch: have you looked at brackets? just out of curiosity [22:11] lazyPower long time ago - when I did, its JS support was really poor [22:11] I recently made the jump from sublime to brackets - its pretty awesome in terms of a free editor. And i'm impressed that its node powered [22:12] *and an adobe product at that [22:12]