[00:09] davecheney: \o/ [00:10] mwhudson: got local provider *almost* working yesterday [00:10] davecheney: on ppc64el? [00:13] mwhudson: ya === hatch___ is now known as hatch [01:04] what version of juju started using jenv file [01:05] sinzui, abentley there's a separate bug on manual provider trunk, that needs --uploads-tools [01:06] hmm.. although it looks like it found 1.17.0 there [01:07] sinzui, the traceback on that bug is the same as i reported previously in https://bugs.launchpad.net/juju-core/+bug/1280678 [01:07] <_mup_> Bug #1280678: manual provider fails to get tools (1.17.2)

[01:08] the first pastebin has the cli output with --debug [01:16] waigani: did you want to do the stand up now? [01:18] wallyworld__: yep 1min [01:54] axw: it appears the reason local upgrade fails from 1.16 is because the /usr/lib/juju/tools directory is owned by root and so can't be read when running juju upgrade. do you know how we deal with 1.16->1.17 upgrades in other places where 1.16 stuff was written as root? [01:55] umm. why does it matter? the agents run as root? [01:56] i'm not sure, i can check [01:56] ha, 1.16 deploy fails error: symlink /var/lib/lxc/ian-local-machine-1/config /etc/lxc/auto/ian-local-machine-1.conf [01:58] yes, the local jujud for machine 0 does run as root. the required file is there, i'll look further to see why it can't be read [01:58] but gotta get a unit deployed first [01:59] hurm, that looks like something that might've changed to support trusty? [01:59] the auto thing [02:02] yeah, i had to manually make the auto directory [02:02] i thought we fixed that [02:03] sigh [02:03] local on trusty is buggered [02:03] "lxc-start": command get_cgroup failed [02:11] :\ === mwhudson is now known as zz_mwhudson [03:59] ah, that's better. phone point was dodgy. bet the bloody ants have gotten into the wires... last time they made a home in our phone [04:03] axw: better bugs in your phone wires than your code [04:04] heh [04:05] it was pretty horrific. I picked up the phone, and they were swarming around under the screen [04:05] \o/ [04:06] axw: i've given up getting the local provider working in trusty for now - i'll ask tim about it next week and/or raise it in the group meeting. but the root cause issue was that i was using the wrong data dir and so have a fix up for that [04:06] okey dokey [04:06] * axw looks [04:09] wallyworld__: reviewed [04:09] thanks :-) [04:10] just writing a test for mine now... sorry, I was being lazy [04:11] np :-) [04:12] given the nature of the issue, i think a test is crucial [09:28] rogpeppe, https://codereview.appspot.com/66540044 - a relatively easy review? [09:29] dimitern: only 1200 lines, trivial :-) [09:29] dimitern: actually i'm on another review currently, will look after that [09:30] is thumper on vacation? [09:30] hazmat, yep, this week [09:30] rogpeppe, cheers :) it's not that bad really [09:34] wallyworld__: hiya [09:35] wallyworld__: i was just looking at https://codereview.appspot.com/66540044 [09:36] oops, wrong URL! [09:36] https://codereview.appspot.com/65740048, of course [10:01] fwereade: just to let you know, I'm trying to get into the hangout, but the google talk plugin crashed, and Trusty wants to do a dist-upgrade that is about a hundred megs, so I should be there *soonish* but I can't guarantee when [10:01] rogpeppe: wallyworld__ ^^ [10:02] jam, no worries :) [10:02] lol [10:02] jam: ok [10:06] I'm having fun with 2fa/hangouts atm so may take some struggling to get into the meeting [10:14] okay, backup otps generated, lets try [12:04] wallyworld__: here's the kind of thing i'm thinking about for the upgrade-waiter worker: http://paste.ubuntu.com/6965038/ [12:05] thanks, will look [12:05] wallyworld__: the SimpleWorker implementation is in one of nate's in-progress branches [12:05] ok [12:08] natefinch: can you make that branch available, so wallyworld__ can use it? (it doesn't seem to be in https://launchpad.net/~natefinch/juju-core/030-MA-HA any more) [12:20] rogpeppe: launchpad's URL schema has to be the most inscrutable thing I've ever worked with. I don't know what it's doing most of the time. The code should still be there, I certainly didn't do anything to it. This link works for the website at least: https://code.launchpad.net/~natefinch/juju-core/030-MA-HA [12:21] natefinch: oh yeah, *code*.launchpad.net [12:21] natefinch: thanks [12:25] rogpeppe, review poke? [12:25] dimitern: thanks, i needed it :-) [12:39] hi guys. I'm having trouble bootstrapping a precise environment using MAAS provider on 1.17.2 [12:39] http://paste.ubuntu.com/6965180/ [12:40] if I manually run that wget command on the maas provider (same network/firewall etc. as bootstrap node) I get a certificate error [12:40] http://paste.ubuntu.com/6965187/ [12:54] this machine has become unusable [12:54] am rebooting to see if it helps [12:55] rogpeppe: time to upgrade [12:55] natefinch: to what? [12:55] rogpeppe: new laptop [12:56] natefinch: ha ha. it was fine before [12:56] natefinch: i'm not doing anything different [12:56] rogpeppe: just take the excuse ;) [12:57] natefinch: i'm waiting until september... [12:57] rogpeppe: what's in sept? [12:57] three years at canonical = new laptop [12:57] rogpeppe: ahh nice [13:02] natefinch: so, it looks like my laptop has died === wrtp is now known as rogpeppe [13:03] natefinch: when i tried to reboot it, i got a kernel panic (something about a thermal_notify call) [13:04] natefinch: and now it just says "Fan Error" when i try to switch it on [13:04] this is unfortunate [13:04] i'm going to let it cool down for a bit and cross my fingers [13:04] rogpeppe: ouch [13:05] it is very hot underneath [13:06] rogpeppe: that sounds really bad [13:07] rogpeppe: maybe it is time for a new laptop? I think you get a payout at 3 years regardless of whether you actually purchase a laptop... [13:08] natefinch: i've only had this for two years. i can't believe that's the expected lifetime of a laptop [13:09] natefinch: i don't know whether to blame trusty for thrashing it, or the laptop for not coping [13:09] anyway, hopefully in a few minutes it will have cooled down and everything will be back to normal [13:09] rogpeppe: 2 years shouldn't kill any reasonable laptop [13:10] natefinch: indeed - and it was a decent one [13:11] anyway, i guess i can continue with dimitern's review, from this apple laptop that's been rock solid for 4.5 years :-) [13:14] rogpeppe, what kind of laptop? [13:14] hazmat: lenovo x21 [13:14] hazmat: sorry, x220 [13:15] dammit, it's cooled down a reasonable amount, and i still get "Fan Error" [13:15] rogpeppe, odd.. i've got the same model.. its held up pretty well for me.. i had one logic board replace (under warranty), i've tricked it out with an extra msata and 16gb of ram. .. what's the issue? [13:16] rogpeppe, hmm.. i had some thermal issues when i had the logic board replaced.. are you still under warranty? [13:16] hazmat: no, not for some time noe [13:16] now [13:16] hazmat: (it's 2.5 years old) [13:17] rogpeppe, bummer.. not sure you have many options.. the docking station helped me not sure why.. it only do the random shutdown on thermal when unplugged i'd recommend picking up a 2|3 year warranty your next go around.. laptop's get abused [13:18] hazmat: everything was running very jerkily, and the fan was coming on. i decided to reboot. i saw a kernel panic when i did so. did a hard power off. now i can't even get into the bios menu. [13:19] hazmat: seems like it might be bricked :-\ [13:19] rogpeppe, hmmm.. you can still get service from lenovo its just pay as you go on labor parts.. might be cheaper to investigate a new laptop and yank the hd for an external enclosure to save data [13:19] rogpeppe, the xps 15 / m3800 looks pretty nice albeit a bit pricey. [13:19] hazmat: i really wanted to wait until my 3 years :-) [13:20] yeah.. understood and the x220 is the last lenovo with a real (non-chicklet) keyboard [13:20] hazmat: i do have a docking station. maybe i'll try that. [13:20] * rogpeppe googles for chicklet keyboards [13:20] rogpeppe, if works at all with the docking station, its almost 100% a logic/motherboard replace.. [13:20] rogpeppe, its what standard keyboards on new laptops look like.. [13:21] hazmat, rogpeppe: the XPS 15 is pretty slick, btw [13:22] rogpeppe, you can order just about any part from lenovo if your up for diy, although on motherboard replace i'd go for a service tech. [13:45] hazmat: docking station (finally found it!) doesn't seem to make a difference, sadly [13:47] i suppose it might actually be that the fan has gone [14:00] natefinch, mgz, anyone? simple review https://codereview.appspot.com/66550044 to fix bug 1282553 [14:00] <_mup_> Bug #1282553: maas provider can't bootstrap on precise with 1.17.2

[14:01] hi jamespage, i was asked to check in with you regarding getting juju-quickstart into main rather than universe. what's the status? [14:07] dimitern: i was reviewing that when my machine died [14:08] dimitern: i'm not sure about the agent config approach [14:08] dimitern: but i need to think about it some more [14:11] ha, actually my laptop *is* still under warranty [14:11] dimitern: reviewed [14:14] dimitern: I'm pretty sure curl has always been in the base cloud image, did you find you needed to install it? [14:15] rogpeppe, it's the least intrusive approach imo [14:15] thanks for the review mgz [14:16] sinzui, do you know about the cross-team meting? when does it start? [14:16] mgz, looking in #juju on canonical it seemed safer to add it [14:16] dimitern, in 15 mintutes [14:17] sinzui, ok, so because fwereade is not feeling well, i'd like to go in his place as he asked, can I have a g+ link in privmsg? [14:21] rogpeppe, got time for another quick question? [14:22] dimitern, https://plus.google.com/hangouts/_/calendar/bWFyay5yYW1tLWNocmlzdGVuc2VuQGNhbm9uaWNhbC5jb20.bbjh6akgumv8n4suee7li8mrfc?authuser=1 [14:22] sinzui, thanks [14:33] dimitern: if you were basing the need to install curl on my aptitude why trick then I think that's a mistake. http://cloud-images.ubuntu.com/precise/current/precise-server-cloudimg-amd64.manifest shows it's in the base image. I'm +1 on mgz's comment that you don't need to install it [14:35] sparkiegeek, ok, i can remove addpackage then === BradCrittenden is now known as bac [14:40] hey, any juju core members available to join the cross team meeting? [14:40] https://plus.google.com/hangouts/_/calendar/bWFyay5yYW1tLWNocmlzdGVuc2VuQGNhbm9uaWNhbC5jb20.bbjh6akgumv8n4suee7li8mrfc [14:40] mramm, i'm there :P [14:40] ok [14:40] cool [14:52] mattyw: sure [14:53] mattyw: sorry, didn't see your question - i'm on a different machine 'cos my laptop has died :-( [14:54] rogpeppe, no problem - I've just submitted my second part of the juju id stuff for review - there's definately more stuff to do - but I wanted to get it started - rather than finish everything and then have to redo parts... [14:54] one of the things I haven't done yet is the cacheAPIInfo stuff - taking out the user stuff and doing it only on bootstrap [14:55] mattyw: ok [14:56] rogpeppe, there's an example in the current code of how I think the function should look - but I don't think I can call it directly from the bootstrap command without jumping through some hoops (I don't have an apiInfo in the bootstrap command) [14:57] rogpeppe, also it breaks a few tests by not doing it - so I guess the cacheuser function will need to be called somewhere where the tests can invoke it as well [14:57] mattyw: i'm just downloading the code so i can have a look [14:58] rogpeppe, ok thanks - there's no particular hurry - but thanks :) [14:59] rogpeppe, another review poke? :) [14:59] dimitern: sorry, i've been busy trying to get my machine repair scheduled [15:00] rogpeppe, you're the 3rd person I know who's had to send a laptop for repair this week [15:00] mattyw: i wonder if it's anything to do with trusty thrashing things [15:01] mattyw: i know that the fan never *needed* to come on before today [15:01] fans always need to come on - it's their purpose in life [15:12] 6 working days until i get it back, darn it [15:14] * rogpeppe wonders what he should do now. i guess i could get an enclosure for the SSD and try to access the disk drive from my macbook, and develop on that [15:31] rogpeppe, try sticking it in the fridge :) [15:38] dimitern: got a mo to help me out? [15:39] mgz, sure [15:39] I want to test a new (old) api command I've added in state/apiserver/client/status.go [15:40] this isn't used in real code, but I want to exercise the function in a unit test [15:40] where should I add that? [15:40] the existing testing in that package is confusing me [15:40] is it one of these things where a CodeNotImplemented is involved? [15:40] well, that too [15:41] but this particular test isn't for that behaviour, though I should *also* have a test for that [15:41] I just want to call the apiserver function from a sensible test to make sure it does things, on the off chance the old 1.16 code was being relied on [15:44] if you want to test the branch where a 1.16 fallback code should be run, what I usually do is go in api/client and change the name of the called method (in st.Call(...)) to something else, like adding "1" at the end, so the server will return not implemented [15:44] it's otherwise hard (if not impossible currently) to write a unit test for that [15:45] moo, so, second test probably not possible [15:45] am I making sense? [15:45] I'm sure the first is though [15:46] it's just like I've added a new api call that's not used anywhere and want to write a test for the server side [15:46] dimitern: I'll put up an mp so far, may be clearer [15:46] mgz, looking at the code will help me understand better [15:52] mgz: you can use Call directly if you need to [15:53] rogpeppe, that's a bit over the edge I think [16:04] dimitern: https://codereview.appspot.com/66590043/ [16:06] mgz, looking [16:14] mgz, reviewed [16:14] merci [16:19] dimitern: responded to comments will have another look at finding a sane test to emulate [16:19] mgz, cheers [16:30] dimitern, mgz: review please? It's pretty small: https://codereview.appspot.com/66230043/ [16:30] natefinch, looking [16:40] dimitern: i actually think that's ok - it's a public API. we're allowed to use it if others are. === wrtp is now known as rogpeppe [16:41] * rogpeppe has just sent his laptop off for repair [16:41] rogpeppe: boo. That sucks. good luck getting it back quickly [16:42] natefinch, reviewed [16:42] rogpeppe, what's that? [16:45] natefinch: yeah, it's a bummer. i've just got a 2.5" enclosure to see if i can talk to the hard drive from my macbook [16:46] natefinch: it may require me running an ubuntu VM, i guess [16:46] can't remember what fs type i was using [16:50] rogpeppe, ext4 if you have deafults [16:50] hazmat: probably [16:51] * rogpeppe goes to look for an ext4 fs for macos [16:51] rogpeppe, i wouldn't bother.. you'll need a linux machine.. a vm might do depending on enclosure type (usb3) and if you can pass it through to the vm. [16:52] rogpeppe, actually it does look like there are some commercial products for ext4 access from osx [16:53] hazmat: that's actually what i'm doing right now - just upgrading my VMWare installation [16:53] hazmat: perhaps i can boot the vmware directly off my laptop hard drive [16:54] rogpeppe, if you encrypted your home drive.... things get trickier.. [16:54] hazmat: it's a usb2 enclosure, unencrypted [16:54] i'm not paranoid :-) [16:54] i'm not either, but i encrypt ;-) [16:54] dimitern: pushed up an attempt at some tests, are they sane enough? [16:55] hazmat: ok, let's just say i'm hopelessly naive :-) [16:55] mgz, will look in a moment [16:57] dimitern: i'm afraid i'm not going to finish your review today [16:57] dimitern: i've got to stop a little early today, unfortunately [16:58] dimitern: i was referring to your "member:rogpeppe, that's a bit over the edge I think" comment about my "just use Call directly" remark [16:59] rogpeppe, no worries [16:59] rogpeppe, i'll have to stop soon as well [16:59] * dimitern needs to pack for a 5 am flight this morning [16:59] "You have Mac OS X 10.6.8. The application requires Mac OS X 10.7 or later." [16:59] bugger [17:00] mgz, I'm looking at your CL, could you look at mine? https://codereview.appspot.com/66340045 [17:00] dimitern: sure thing :) [17:01] cheers :) [17:04] dimitern: just looking at Andrew's change that introduced it to see if we're going to be borking anything in reverse, change itself looks fine [17:04] ah, eg: [17:04] mgz, I also looked and couldn't find a sane reason for the change - ssh does not support -- neither does -- [17:04] === modified file 'cmd/juju/debughooks.go' [17:05] s/does --/does scp/ [17:05] - args := []string{"--", fmt.Sprintf("sudo /bin/bash -c '%s'", innercmd)} [17:05] + args := []string{fmt.Sprintf("sudo /bin/bash -c '%s'", innercmd)} [17:05] does that then want reverting? [17:05] (looking at `bzr diff -c2116`) [17:05] mgz, I'll try running debug-hooks with the test [17:07] thinking about that bit, it's *probably* fine if we got our shell quoting correct [17:07] one string, should be quoted and interpreted as a command regardless [17:08] === modified file 'cmd/juju/ssh_test.go' [17:08] - sshArgs + "dummyenv-0.dns\n", [17:08] + sshArgs + "ubuntu@dummyenv-0.dns\n", [17:08] er.. not that hunk [17:08] but some later bits, changed to add -- [17:09] mgz: fwereade_ you guys available for a sync on the joyent provider [17:09] === modified file 'environs/sshstorage/storage.go' [17:10] again, that change should be fine if we get our quoting right [17:10] arosales1, fwereade_ is not i'm afraid - a bit ill and resting [17:10] arosales1: alas, I need to go get a bus in a sec [17:10] fwereade_: ? [17:10] arosales1: you may need to rope in dimitern, it's his day of getting roped in :) [17:10] ah dimitern :-) [17:11] have been doing a fair bit with daniele this week, he should be able to report most of it [17:11] arosales1, :) hmm [17:11] arosales1, I have no knowledge on joyent or its progress though [17:12] dimitern: utils/ssh/ssh.go has an interesting bit [17:12] - args = append(args, host, "--", command) [17:12] + args = append(args, host) [17:12] + if len(command) > 0 { [17:12] + args = append(args, "--") [17:12] + args = append(args, command...) [17:12] + } [18:09] anyone know which version of juju started using jenv files? [18:10] * hazmat thinks it was 1.16 [18:13] hazmat: that sounds right, but I don't know for sure [19:00] is there any way in the api to tell which machines are 'dirty'? [19:00] afaicr.. no [19:07] filed a bug [19:07] hazmat: in the API? Possibly not [19:08] hazmat: I know we have that information, but not sure if it's exposed. It should be. [19:08] natefinch, doesn't appear to be any api for machines outside of create/destroy and sidecar via status [19:08] well and inject/provisioning scriptfor manual [19:08] hazmat: I'd love a juju status that gives more details about the machine [19:09] natefinch, yeah.. i think that's the intended direction for extended info === zz_mwhudson is now known as mwhudson === mwhudson is now known as zz_mwhudson [20:46] chris38, btw that go get -u -v launchpad.net/juju-core/.... needs a go install launchpad.net/juju-core/... as well to get the latest binaries. === zz_mwhudson is now known as mwhudson === andrewsmedina__ is now known as andrewsmedina_ [22:07] natefinch, do you think you will merge this by your morning tomorrow? https://code.launchpad.net/~natefinch/juju-core/034-juju-mongo/+merge/207320 === cmars` is now known as cmars === allenap_ is now known as allenap === lazyPower_ is now known as lazyPower === hazmat` is now known as hazmat