[00:44] any bash experts in? [01:03] hatch: not an expert but I might be able to help [01:03] bcsaller: oh np I figured it out already [01:03] thanks though [01:03] heh, ok [01:03] in fact I think I found a bug in the Hadoop bash script [01:04] copied it wholesale into a script to test and it failed...so I was like wtf hah [01:04] but since you're here mind taking a look at a diff? [01:08] https://gist.github.com/hatched/fc70d29198df22cf32ad [01:08] I'm assuming that the size of the diff is because there were trailing whitespaces? [01:20] hatch: I'm guessing line 174 was where your question was :) [01:21] nope, see 126 vs 132 [01:21] that being on the next line caused it to not work in my test script [01:21] so I wasn't sure how to debug it [01:21] bash -x filename [01:21] the -x was what I needed [01:21] now I have to figure out how the heck to deploy a charm from my launchpad account [01:22] :/ [01:22] ahh, that I could have told you, you put set -xe at the top of scripts to get the echo/exit on error behavior [01:22] oh cool [01:22] what does the -e do? [01:22] exit on error [01:22] ahh cool [01:22] x is the echo [01:22] learn something new every day [01:22] :) [01:23] looks like it should work, you verified that was all the usage of config-get/set? its not in the relation hooks or anything? [01:23] or are they all symlinks? [01:23] all symlinks [01:23] interesting approach but smart [01:24] now I'm trying to test it but I can't figure out how to deploy it lol [01:25] looks like it's gota be local [01:32] hmm I want to deploy a local charm via the GUI [01:32] I wonder if this is a rare occurance [01:33] yeah, it doesn't work by default [01:34] I'm not actually sure how you would do it withou a local charmstore [01:34] but it would be cool if we could select a dir to deploy [01:45] hatch: we've talked about a number of longer term plans for how to do this, but currently I think your instinct was correct a local charm store and maybe changing the config settings to point the API endpoint there [01:49] yeah I'm definitely not going to attempt setting up a local charmstore [01:49] :D [01:49] I gota go cook supper soon haha [02:13] me too [03:06] hatch: it's a charm, you can push your branch up and get it in the official store [03:06] hatch: just 'juju publish' it. It'll be under your username, etc. [03:07] https://jujucharms.com/fullscreen/search/?text=apache look at the 'more charms' section :) [04:14] rick_h_: oh cool thanks good to know [06:38] hatch, thanks [06:39] anyone know how to get personal charms to show up in the gui search? [06:42] hmm.. looks like its a manage.jujucharms.com issue.. [12:10] benji: is charmworld still sans lbox? [12:11] rick_h_: some people have claimed that it works for them, but I haven't been able to get it to work. The problem may be the way I have lbox installed on my LXC container. [12:12] benji: hmm, that's what I'd be doing so if you're willing to do LP review I'll skip it then. [12:12] benji: writing things up and will ping once done if you've got time to peek at it. [12:13] rick_h_: sure [12:18] benji: https://code.launchpad.net/~rharding/charmworld/bundle-metadata/+merge/188905 (one line lint fix just pushed and will sync in a min) [12:18] thanks! [12:20] benji: and so the idea is that with this we get support for the trailing bits of a url after the id, so going to add a bundle/some/id/charmdetails to provide the different payload format withthe charm details per our previous discussion? [12:20] does that jive with what you recollection of our talk on Tues? [12:22] rick_h_: makes sense [12:23] rick_h_ (and gary_poster): I need to run an unscheduled child to daycare, I'll be back in about 15 minutes [12:23] ok benji [12:23] thanks benji [12:53] luca__: hi, could you point me to any css/html resources for building a webpage similar to juju.ubuntu.com? [12:54] benji: when you get back want to see if you've got a sec for a chat. A question came up from my call with gary [12:54] frankban: there's the base resources. /me goes to look where they shared that. [12:55] rick_h_: thanks [12:55] rick_h_: I'm available now. [12:55] benji: want to use your chat thing? [12:56] rick_h_: sure: benjiyork.com/chat [12:56] frankban: so there's a lp project of shared base css/etc resources and I'll have to dig for it. If luca__ comes by he might remember the link. The web team in HQ runs it [12:56] it's like ubuntu bootstrap [12:57] frankban: rick_h_ heya [12:58] frankban: rick_h_ this is the link: https://dl.dropboxusercontent.com/u/46840621/Web%20style%20guide/v3/get-started.html [12:58] frankban: rick_h_ it's work in progress but should be what your looking for [12:58] benji: http://bazaar.launchpad.net/~bac/charms/bundles/wiki/bundle/view/head:/bundles.yaml [12:58] hi luca__, thanks! [12:59] frankban: rick_h_ if you need any help or information, ping Anthony Dillon (ant on IRC) and he should be able to fix anything [13:02] http://charmworld:2464/~bac/bundle/wiki/wiki [13:03] https://manage.jujucharms.com/~bac/bundle/wiki/wiki [13:32] frankban: luca__ http://bazaar.launchpad.net/~ya-bo-ng/ubuntu-brand-guidelines/trunk/files is what I was thinking of I think [13:33] hey jujugui, I feel not good. Going to go lie down for a few. back when I can; will periodically check email [13:34] gary_poster: good luck [13:34] ty [13:44] when we remove stuff in the GUI [13:44] does that destroy-service or destroy-unit? [13:46] jcastro: there's the option for each, not following you. [13:47] http://askubuntu.com/questions/353114/how-can-i-know-which-machine-juju-is-actually-using [13:47] I am wondering if people think they're removing units [13:47] but are not [13:49] I know lxc has a thing where destroy service doesn't remove the machine, maybe azure has an issue there? [13:49] * rick_h_ checks some code quick [13:50] yeah, since this is an azure one I was wondering that too [13:50] nice to know people are trying it on azure! [13:52] well, so destroy service calls destroy service. I'd be curious to know if the gui instance was recent enough to have the inspector. In the older UI I could see where destroy unit might be what he clicked vs service destroy. [13:52] when you destroy a unit does the underlying machine get removed? [13:53] guess, the only real way to chase it would be to try it out and see if we could dupe it. [14:01] I'm going to ask him to mail me, I have a bunch of other questions to ask him [14:01] more about "hey so did our docs suck?" and so on [14:03] jcastro: ok cool. Let us know if you can help when you get more details. Knowing more of what version of the Gui he was using, what he actually did, etc [14:04] I don't know if any of us have tried things on azure yet so cool to know it at least got going and working :) [14:09] jcastro, rick_h_ : from the GUI we send a DestroyService API call. ServiceDestroy does not destroy the machine (and I believe this behavior is shared by all providers). AFAICT there is no API for destroying a machine. DestroyServiceUnits also seems to leave the machine up. [14:13] yeah the thing is this guy thought he was running X VMs, but was really running X+whatever. [14:14] that's going to cost people money [14:14] then they're going to say "wait, but I told the GUI I didn't want those services anymore, why are there blank machines running up my bill?" [14:17] jcastro, rick_h_: That's right. I believe we will handle that as part of our machine/containerization story. Right now the GUI does not manage/expose machines at all, and as I mentioned my understanding is that there is no API to do that (we will need an API for all the provisioning CLI commands). [14:19] * jcastro nods [14:20] jcastro: you might want to ping gary_poster when he's back, he surely have a better knowledge of the problem [14:20] I think in the meantime ... would it be a bad thing to note this in the docs or the charm readme? Like in huge letters? [14:20] I am just envisioning some guy trying out juju at work and running up this huge bill [14:20] and his boss getting stuck with the check with a mental image of juju... that thing that cost me money. [14:22] jcastro: yes, that's a good idea, I'll make a card [14:22] frankban: oh, I thought it was just an lxc thing which is how I end up doing things. Knowing is half the battle I guess [14:23] rick_h_: yeah, that's the same on ec2. It's the way juju-core works when you just destroy a service [14:25] rick_h_: pyJuju did that differently: destroy-service == bye bye machine [14:25] yeah, I think we should bring in design at the sprint too. We need to ensure that the GUI always is upfront on how many instances are running for real [14:25] we can't let that be hidden or put behind somewhere where it won't be discoverable [14:25] frankban: ahhhh, maybe that's where I'm confused then. I only moved to juju-core once lxc came out [14:26] jcastro: yea, there was work on ideas for a 'machine view' during the last sprint, but that's work that's a little bit out atm [14:28] abentley: I have a demo bundle I want to get into charmworld, will it be picked up automatically, or is there something I have to do? [14:31] rick_h_: created a card in maintenance high: GUI Charm: document that machines are not removed when you destroy services from the GUI [14:32] rick_h_: yeah I get that, but maybe there's some low hanging fruit somewhere [14:32] rick_h_: if you don't know how juju works it looks like the gui is lying to you [14:32] jcastro: yea, just meant the idea is on the radar. [14:32] benji: It will be picked up automatically, as long as the naming conventions are right. [14:33] abentley: great, thanks [14:33] benji: file must be named "bundles.yaml", series must be "bundles", branch must be named "bundle" [14:33] benji: This just got picked up automatically yesterday: http://manage.jujucharms.com/~abentley/bundle/charmworld/staging [14:33] abentley: I think I have it right. I branched from bac's bundle, which appears to work [14:45] 40 minutes -> 10 minutes. I kind of hate this battery. [14:45] Makyo: replacement battery? [14:45] rick_h_, I guess so. It's been less than a year, though, so I'm mostly just disappointed. [14:46] Makyo: gotcha, yea mine is just over a year and getting to that time so battery on the brain [14:57] Makyo: http://paste.mitechie.com/show/1027/ ? [15:02] .182222 [15:02] wow! [15:02] so when you charge it up now, it's only going to 18% of the original capacity [15:02] and it's less than a year old?! [15:03] Yeah, I got it just before Copenhagen. [15:03] mine's at 75% and on the replacement block. [15:03] hit 1yr around july I think [15:03] crazy [15:09] Oh well. Sent in a support ticket. If the solution is to buy a new battery, that's fine, will just have to wait. [15:10] yea, if less than a year that seems a warranty issue for sure [15:11] I checked, it's just about a year, got it about this time last year, from the looks of it. I didn't do the "periodically drain the battery" thing, though I'm not sure if you need to do that anymore. [15:11] yea, it should destroy it that bad :/ [15:12] Damn. [15:12] heh, warranty die yesterday? [15:12] Last week. [15:12] Oh well. [15:13] Makyo: well time to get the Air setup for dev :) [15:14] Haha, I guess! [15:14] I'll have it set up by Oakland, for sure. [15:14] It kinda works now, just not ideal. [15:14] it has 8GB or 4GB? [15:14] 4 [15:15] So lboxing will take a while. [15:15] ahh - well you might want to run Ubuntu on metal then if you want some better perf outa it :) [15:15] It works fine for actual dev work, tbh. [15:16] oh that's cool - I was a little worried the vm would use too much ram [15:17] I have it somewhat constrained, but it's also a very light-weight installation of ubuntu server, rather than desktop. [15:20] abentley: how long should I wait for my bundle to be picked up before becoming concerned? It has been well past the 15 minute enqueue/ingest cycle. [15:21] benji: It should only take 15 minutes, so I'd be concerned now. [15:21] darn [15:21] benji: What's the branch? [15:22] abentley: https://code.launchpad.net/~benji/charms/bundles/wiki/bundle [15:24] benji: Nothing obviously wrong to me. [15:24] abentley: thanks for looking [15:26] well I have now spilt two cups of coffee on my desk today....w t h [15:27] benji: So, it's up on staging, kinda. http://staging.jujucharms.com/~benji/bundle/wiki/wiki [15:28] abentley: the kinda part being the "Internal Server Error"? :) [15:28] benji: Yes. I think it's because your bundle doesn't have a series. [15:29] ah [15:30] I added a series, we'll see how that works. [15:30] I'm confused though because it worked on my local charmworld [15:31] benji: /home/webops_deploy/charmworld/charmworld/templates/bundle.pt refers to ${bundle.data.series}. I don't know how it worked on yours. [15:32] abentley: well, I didn't view the page, I was hitting the API with the GUI, but that isn't working on production either so I'm still confused [15:32] benji: Your bundle isn't showing on production, so that's a separate issue. [15:33] mmm, good point [15:33] I'll try pointing my GUI at staging. [15:34] bac: benji, jcsackett : I jot hate mail from charmworld staging: https://bugs.launchpad.net/charmworld/+bug/1234780 [15:34] sinzui: Yes, I triggered it by visiting benji's bundle: http://staging.jujucharms.com/~benji/bundle/wiki/wiki [15:35] yep, we'll have to fix... something so bad bundles won't go down in flames like that [15:35] sinzui: But there's a separate issue: that bundle isn't appearing on production, even though it's had plenty of time to ingest. [15:35] abentley, from when? [15:36] benji: ^^ [15:37] benji, abentley we killed a stale proc from before the bad release to get charms ingesting again. [15:37] sinzui: 09:11:01-0500 [15:38] okay, that is just minutes after we completed. I think the cron or run-forever queuing is broken in production [15:39] abentley, I think we just need supervisord running to queue. and cron will ingest about every 2 to 4 times per hour [15:40] I'm pretty sure cron does the queueing and supervisord manages the ingest. [15:40] sinzui: ^^ [15:42] abentley, yep. This is what webops created to make cron happy (i think) https://pastebin.canonical.com/98319/ [15:43] sinzui: Looks sane to me. [15:44] I will enquirer about it [15:50] jujugui call in 10 [15:51] * hatch is pretty pumped that he beat Makyo to it today [15:52] Whoa whoa whoa. [15:52] Coffee is WAY more important than this. [15:57] rick_h_: so I've noticed that `bzr grep` doesn't give line numbers or syntax highlighting - am I missing a flag or something? [15:57] abentley, benji looks like mthaddon found the issue, crontab was manually put in the wrong place in production. He is fixing it [15:58] hatch: no, that's ack stuff. I just open the file and use vim /xxx [15:58] oh ok, then back to ack it is :) [15:59] jujugui call in 2 [15:59] damnnn [15:59] Hahaha [16:01] benji: my new desktop env pulled your charm in [16:02] rick_h_: good; I think production is on its way to being fixed [16:02] cool [16:02] luca__, ping [16:03] abentley, benji : I misspoke. I still have no idea what is not always running to ensure we ingest [16:03] mmm [16:03] Makyo: heya [16:03] Makyo: I need to check the upgrade stuff [16:04] sinzui: You can try running bin/ingest-queued manually to see if the queue's getting stuff. Remember to set INI and run as ~charmworld. [16:06] luca__, yeah, have a sec after daily meeting? [16:06] Makyo: sure, just ping when your free [16:13] abentley, I don't want to manually run. that is what got the last ingest sorted. I want to know what is not runnung [16:14] sinzui: I was suggesting manually running as a way to determine what's not running. If the queue is full, then you know that cron is happy. If it's empty, then you know that it's not happy. [16:15] okay, I will ask again if the queue is full [16:19] luca__, https://plus.google.com/hangouts/_/9ad52af076c648af1c3c61fcda614bb034cfc336?hl=en [16:19] Makyo: give me 2 mins :) [16:19] luca__, okay [16:22] Makyo: coming now [16:24] well I finally had to turn the heat on... 16C is too cold :) [16:34] hey rick_h_ & benji, am I right that this is something we need to address asap? https://bugs.launchpad.net/bugs/1234780 [16:34] <_mup_> Bug #1234780: Dict has no attribute series [16:41] abentley, I think we will ingest https://pastebin.canonical.com/98468/ [16:41] abentley, I think queuing via cron is bad [16:42] I could ask for a manual queue, wait 15 minutes and verify all it good. then return to the question of why cron is not [16:42] sinzui: I'm pretty sure I changed the default delay to 1 minute, not 15. [16:43] abentley, not on staging and that is where we stole the crontab from [16:43] sinzui: I meant the default delay for ingest-queued. [16:44] oh [16:44] gary_poster I'm not sure. Do we have the idea of a proof for bundles? this seems like if it's a required attribute then proof/ingest would catch/ignore it [16:45] gary_poster: so I guess yes we should adress but unsure how atm [16:48] rick_h_, doesn't that look like a problem in the charmworld code, not the bundle? [16:49] AFAICT the traceback is saying that the template is trying to look at the data as an object, not a dict [16:49] gary_poster: well, series is expected to exist. The question is, can a bundle not provide it? (reasonable to expect it to always be there) or is it ok to not have and thus the template/charmworld needs to update to deal with the lack of it [16:50] rick_h_, the bundle has a series: http://bazaar.launchpad.net/~benji/charms/bundles/wiki/bundle/view/head:/bundles.yaml [16:50] gary_poster: ah ok then. Yea. [16:50] rick_h_, but data is a dict, so the code would have to say /series [16:50] strange if it's an issue there it dodn't show befoer [16:50] or ['series'] [16:51] cool rick_h_ , thanks.I'll make a card for bug [16:51] gary_poster: cool, will coodinate with benji when he's back around then and peek at the code after I fix up these review changes [16:52] cool [16:52] thank you [17:09] gary_poster: benji just added the series. [17:10] abentley, oh! so it was what rick_h_ said: bundle had the problem, not code? [17:11] other than proof equivalent needed in charmworld [17:11] gary_poster: I think series is supposed to be mandatory. But the error handling could stand improvement. [17:11] gary_poster: yea, I think we had a chat with marcoceppi once about the idea of updating the proof tools to be able to detect or be told to proof using rules for charm or bundle in one command [17:11] and then run proof against both types on ingest [17:11] got it, thanks abentley and rick_h_ [17:12] gary_poster: (Now that it has a series, it's not broken: http://staging.jujucharms.com/~benji/bundle/wiki/wiki) [17:12] gotcha [17:13] sinzui: successful test run, from creating the container, through running the test suite, to destroying the container. [17:13] abentley, there is a fix for that in 1.15.1 [17:14] sinzui: A fix for succcessful test runs? [17:14] abentley, sorry. I am in too many places [17:15] any I just put my mouth the the production firehose. prod will email me [17:17] abentley: benji is charmworld landing still just mark approved and tarmac comes along and grabs it? I've not been following it lately. [17:18] rick_h_: It's still just mark approved, but it's all-Jenkins, no Tarmac. [17:18] abentley: ah, very cool. Thanks. [17:19] abentley: hmm, I can't see the "approved" in the drop down for status here? https://code.launchpad.net/~rharding/charmworld/bundle-metadata/+merge/188905 [17:20] ah, pending approval to the charmworl developers team. [17:21] can anyone approve me in please or add the Juju Gui Hackers team? sinzui abentley ^ [17:21] rick_h_: You should see it if you're a member of the branch owner team, Juju-Jitsu Hackers. [17:21] abentley: looking [17:21] abentley: ah, I am not. Asking back in now. [17:22] will need an ok on that one as well. [17:22] * rick_h_ is feeling the repercussions of losing super orange powers [17:22] * rick_h_ wonders how benji got to submit [17:24] rick_h_: that (just mark it approved) is my understanding [17:24] benji: right, but I'm not member enough to have permission to makr it approved [17:24] benji: I only see the options to mark it as WIP, needs review, merged [17:25] rick_h_: hmm, want me to do it? [17:25] rick_h_: He's a member of yellow squad. [17:25] benji: I'm not seeing you in the group abentley mentioned either though. Chasing the permission chain [17:25] benji: if you don't mind until I get admins to approve my requests to join their teams [17:25] rick_h_, I have approved you [17:26] sinzui: thanks [17:26] I'll ask gary_poster to update yellow squad when he's got time. [17:26] thanks sinzui got it marked approved now. [17:30] rick_h_, everybody is in yellow. probably should retire yellow instead but enh [17:30] gary_poster: thanks for the update [17:30] rick_h_, I mean, I added you and everyone else in gui [17:30] welcome [17:32] rick_h_: is 'editorial' the token list? [17:33] it feels like it should be the details view? [17:33] hatch: editorial is the default content in the side panel which is the three containers (featured, popular, new) [17:33] they're 'editorial' selected charms to show on page load by default [17:34] uhh wut? [17:34] hatch: so in the sidebar View you can load two bits of content: editorial content or search results content [17:34] lol [17:34] which are each a View [17:34] so editorial is not editorial [17:34] hatch: call? [17:34] it's the token list [17:34] sure [17:34] hatch: sure it is, it's three containers (and used to have the listed categories) and such === matsubara is now known as matsubara-lunch [17:55] hey bcsaller I'm playing around with the visualization via tests and would like to make faster progress. you have a sec for a hangout? [18:01] rick_h_: how close are bundles to being ready? [18:01] I'm planning on pressing a release before cloud sprint, if you'd like that in there I need to know sooner than later [18:02] also, can someone fill in this guy with a roadmap to being able to use "--to" in the gui? http://askubuntu.com/q/353262/41 [18:02] marcoceppi: so we can ingest/show them in manage.jujucharms.com now. hatch will have a details view in the gui next week, and so parts of it will be giong on next week [18:03] marcoceppi: so yea, bundles will be pre-sprint and so bundle support in proof would be greatly appreciated :) [18:05] marcoceppi: for the ask ubuntu gary_poster would have to provide the best answer, but I think it's part of a big chunk of work for 14.04? [18:07] rick_h_: if you haven't already, I need deets! https://bugs.launchpad.net/charm-tools/+bug/1222833 [18:07] <_mup_> Bug #1222833: charm tools should proof bundles as well [18:07] sample no longer works [18:09] bcsaller: are you a deployer expert to know which all fields are required/optional in a file format? Can you comment in ^^ bug? [18:09] or maybe frankban since I know he did a lot of that in the server bits. [18:10] marcoceppi, rick_h_ I answered the askubuntu thing. [18:10] gary_poster: thank you. [18:10] rick_h_, benji is your best bet. frankban has EoD'd [18:11] gary_poster: right, no hurry. I figured I'd shoot for someone that's used the deployer/messed with that code itself wouldhave the best knowledge [18:11] cool rick_h_. thx [18:14] rick_h_: I would have to read the code to see what is required, but I would assume "services" is the only required key ("services" should be made optional by charmworld) [18:14] oops, that second "services" should be "series" === matsubara-lunch is now known as matsubara [18:22] benji, I'd guess proof could go farther if desired--validating the deployer format generally, yeah? [18:22] gary_poster: yea, one items was to check it's valid yaml/json [18:22] yeah, reasonable [18:23] gary_poster: yep; for one thing, it could check for a bug I have seen, which is relation names not being correct in the bundle (it will obviously need net access to load charm details to do that, though) [18:24] stuff like that I think charmworld can do easier. I'm not sure proof should be online? [18:24] proof doesn't check everything ingest does, but it's a basic linter [18:25] curious what marcoceppi thinks on that part though. [18:26] marcoceppi, I am making a card for us to give you some details. how soon do you need them? and thank you, btw. :-) [18:31] benji: isn't there an issue with setting kwarg defaults to mutable types? Is dict() considered immutable? [18:33] rick_h_: dict() just returns an empty dict, so it is the same (mostly) as using {}, so yeah, you wouldn't want to use that as a default (unless you really do, but still you probably shouldn't because it will confuse people) [18:34] benji: ok, just noticed it in get_bundle_data and hasn't seen dict() used and wondered if it somehow got around the badness of using {}/[] for kwarg defaults [18:35] rick_h_: the "issue" isn't so much a bug or anything like that, it's that people often don't understand default keyword arguments are evaluated when the function is read, not every time it is called, so a mutable default could be changed by the function from one call to another [18:35] benji: rgr, which from a factory method creating different versions seems like a particularly troubled spot to reply on those defaults [18:36] benji: but yea, all good. Thanks for the sanity check. I'll tweak in a drive by in this branch [18:37] rick_h_: yeah, for the code in question it will technically work, but it is an attractive nuisance; I agree that it should be tweaked [18:41] rick_h_: looks like you have that resolved? [18:41] bcsaller: rgr, thanks [18:42] gary_poster: did you still want to chat about the vis stuff? [18:42] bcsaller, yeah, thanks. 1 sec... [18:42] bcsaller, calling on https://plus.google.com/hangouts/_/58de968d2a9ae12f362336b9c9a13fb586d4c581?hl=en [18:44] very odd [18:50] gary_poster: sooner the better. By Wednesday if you want me to have time to get it in the release [18:50] ok thanks marcoceppi [18:50] rick_h_: proof probably shouuld, orange has opened a bunch of bugs about this IIRC [19:08] hey hatch, is your branch working enough for me to use as a basis for playing around with integrating Ben's visualization? If so, could you push it somewhere and let me know the address? [19:08] gary_poster: I have 12 failing tests I'm just finishing up now, then it'll be ready to propose [19:08] so can you give me a few minutes to eval these failures then I'll get back to you [19:09] hatch, cool. if it is working in qa then I can use it, test failures or no. so, please push it soon-ish however that turns out [19:09] oh ok, yeah as long as you aren't going to base any real code off of it, just incase the tests expose changes that need to be done [19:11] gary_poster: lp:~hatch/juju-gui/bundle-view-url http://localhost:8888/bundle/~benji/wiki/wiki/:flags:/charmworldv3/ to view [19:11] ok thanks hatch! [19:13] no problemo [19:14] I -really- wish I knew why phantom crashes when being run in the foreground [19:14] it's gota be these 'get' logs [19:33] abentley: please let me know when you have time for a short call about the ingest process and backfilling [19:33] benji: Okay. Probably in 30. [19:34] abentley: thanks [19:45] jujugui could I get a quick review and qa on https://codereview.appspot.com/14355043 thanks [19:46] hatch: pulling it down now [19:46] thanks [19:58] benji: I'm free now. [19:58] abentley: ok, one second; I left my camera outside [19:59] bcsaller: thanks for the review - I'm not really clear on the direction you want me to go with your comments [20:00] abentley: try benjiyork.com/chat [20:00] hatch: sounds like you want to make changes in later branches then and go as is? [20:01] bcsaller: yeah I want to try and do atomic changes and work towards the final goal - I think the reviews/qa's will be easier this way [20:01] then I'll +1 it [20:01] benji: I get a black screen that says "Initializing..." with Firefox [20:01] abentley: ok; I'll create a hangout then [20:02] benji: Chrome works. [20:02] benji: Or at least it says Connecting... and shows my face. [20:03] bcsaller: cool thanks [20:03] gary_poster: I'm going to be submitting that branch now just FYI [20:03] cool hatch thanks! [20:03] I'm playing with integrating. I'll put my face on the card. [20:04] Makyo: Loving OneTab :) thanks for the tip [20:05] Yeah! [20:05] It's not perfect, but it's a pretty good grouping solution to cut down on memory footprint from multiple windows. [20:05] yeah I have 54 windows in it right now haha [20:05] tabs* [20:07] Yeah. It's also kind of neat that you can restore one, or a whole group, or publish the list. [20:19] Oh, so these aren't battery crashes. 23 LXCs crashing the machine. [20:20] BEcause I'm not destroying machines. [20:20] ouch [20:20] I have everything tuned way down to try and increase battery, though, so I guess it's both. [20:21] lol [20:25] sinzui: Complete test run in lxc: http://162.213.35.27:8080/job/charmworld-autoland-lxc/lastBuild/ [20:25] bcsaller, good news is that I have it working. (we do need fakebackend, it turns out, to load bundle into db). Not as good news is that the one good bundle we have doesn't render very well. I suspect that the root cause is that the bundle does not have x-y annotations. The symptoms are that I see a lot of errors trying to draw relation lines ('Invalid value for attribute y2="NaN"') and I only see a single [20:25] charm (because they are all on top of one another. So... [20:26] gary_poster: wp-deployer? I thought that did have annotations, we even test the positions [20:26] oh... from the store [20:26] yeah... [20:26] there isn't really any auto layout [20:27] the assumption being these things will be placed, but thats an area to work on for sure [20:27] so a few options/things to consider... [20:27] (1) we can ask marco to enforce that bundles have x-y coords [20:27] benji, rick_h_, sinzui, jcsackett: I'm switching over to doing the landings with testing "make sysdeps". (by using an lxc). Sorry in advance for any bumpiness. [20:28] that will probably be the cheapest option and a reasonable short term solution [20:28] (2) auto layout [20:28] abentley: so it will be a "fresh" system each time the tests are run? [20:28] maybe there are only two ;-) [20:28]