=== defunctzombie_zz is now known as defunctzombie === defunctzombie is now known as defunctzombie_zz [07:33] <_mup_> Bug #1182774 was filed: JuJu not found after having installed juju packages. [09:32] could one of the charmers team +1 https://code.launchpad.net/~james-page/charms/precise/ceph/fix-notify-clients/+merge/161835 [09:33] its quite a trivial fix and resolves a common race condition [09:33] jamespage: k, will take a look [09:39] jamespage: k, done [09:39] bbcmicrocomputer, ta muchly [11:01] bbcmicrocomputer, if you are feeling brave - https://code.launchpad.net/~openstack-charmers/charms/precise/mysql/ha-support/+merge/165059 :-) [11:03] jamespage: hmm, I think this would take me a day (at least) === dimitern is now known as dimitern|afk === hazmat` is now known as hazmat === BradCrittenden is now known as bac === rogpeppe1 is now known as rogpeppe [13:49] marcoceppi: so tldr [13:49] jono started with juju [13:49] and was on pyju, installed charm-tools [13:49] * jcastro makes explosion sounds [13:49] will file a bug to provide context [13:49] Okay, Well I woke up this morning, upgraded my computer, and now pyjuju is disappeared from update-alternatives for juju [13:50] But that has nothing to do with charm-tools [13:50] * marcoceppi awaits bug [13:50] hey so from your email, it seems you changed the region after you had deployed something? [13:51] * jcastro will just wait for the detail in the bug === wedgwood_away is now known as wedgwood [13:51] jcastro: OH HE INSTALLED CHARM-TOOLS and that depends on pyjuju and everything borkd? [13:52] yeah, I believe that's the case [13:52] he had installed goju but I don't think he did the update-alternatives [13:52] and from looking at the log he was using pyju the whole time [13:52] unless we added timestamps to goju recently? [13:52] the console output I mean [13:52] no, only pyjuju does console timestamps [13:53] right [13:53] that awkward phase, where we still tell people about pyju in the docs. [13:53] So if you install juju-core, then juju-0.7, update alternatives re-configures to use juju-0.7 as it has higher weight in update-alternatives [13:53] marcoceppi: this is why I pray for the local provider every day [13:53] yeah [13:53] what is unexplained is how he was working fine until he installed charm-tools [13:53] jcastro: we really just need to change charm-tools to suggest juju, it's not a dependency by any means to run [13:54] * marcoceppi double checks packaging [13:54] https://bugs.launchpad.net/ubuntu/+source/charm-tools/+bug/1182905 [13:54] <_mup_> Bug #1182905: charm-tools requires extra environments.yaml config

[13:55] jcastro, why why doesn't juju-core just default to using the go version? [13:55] brb [13:55] because the go version doesn't have the local provider yet [13:56] and it should, I suspect you had juju installed already and installing core installed goju, but by default we don't switch you from pyju to goju unless you are explicit about it [13:56] the ideal is of course, we have a local provider ready for goju and that removes like all the complications you just had, other than the -tools thing [13:58] jcastro, dude, as I have said twice, I didn't have it installed :-) [13:58] this is a new machine running saucy, I haven't installed juju on it before [13:59] jcastro: based on the bug, it looks like the flow as "Install juju-core, deploy stuff, install charm-tools, charm-tools magically installs juju (0.7), try to deply somemore, juju 0.7 envs not compat with juju-core, weep" [14:00] yeah, if I try to install charm-tools here, I get [14:00] The following NEW packages will be installed: [14:00] charm-tools juju juju-0.7 mr python-cheetah [14:00] I don't even have a juju-core package installed, I use the trunk built version in GOPATH [14:00] ahasenack: yeah, charm-tools recommends Juju [14:01] marcoceppi is right, and then it defaults to pyjuju, and env file and actual environment is not compatible [14:01] evilnickveitch, started using your docs by the way [14:01] does charm tools work with gojuju? [14:01] I can move juju from recommends to Suggests, but it really *does* recommend juju, it just needs to either recommend juju-core instead [14:01] evilnickveitch, I am testing juju by using them, is there a place you want me to file docs bugs/ [14:02] ? [14:02] I don't even remember what it is [14:02] ahasenack: charm-tools doesn't rely on juju at all, so yes :) [14:02] jono, cool - for the moment just email me [14:03] I'm going to just add juju-core | juju to the recommends, should resolve this for future users [14:03] jono - and don't listen to anything the constraints section says, it's all wrong [14:03] evilnickveitch, ok [14:03] thanks [14:04] jono - btw are you checking out the branch or using the online version? [14:05] marcoceppi: that sounds like a plan [14:10] evilnickveitch, online version [14:11] jono, ok, well, it is currently updating quite often, so refresh before you flame me :) [14:11] evilnickveitch, np :-) [14:13] marcoceppi: hey, moving forward maybe we should test the charm-tools/juju installation stuff in a container or something [14:13] jcastro: we can make it part of charm-testing since we have the jenkins infrastructure already [14:13] err, charmtester* [14:13] yeah [14:13] Could also push it in to qa.ubuntu.com [14:14] Though we have plenty of good testers already like jono :P [14:14] not sure I am a good tester [14:14] but I am a tester :-) [14:15] you find things that advanced users won't [14:15] like I suspect your findings will be different from what say, kapil would find. [14:17] evilnickveitch: I send mattyw your way wrt. doc inline videos [14:17] jcastro: I've patched it in the ppa, waiting for it to be merged for saucy [14:17] jcastro, yeah, I saw that thanks :) [14:18] another sucker... er, I mean helper... [14:19] evilnickveitch, jcastro I plan to do a test video at some point over the next few days to see if it's good enough [14:19] mattyw - cool! I'll mail you the specs we are using and a suggestion for something to record :) [14:20] yeah, make the test video count for something. :) [14:31] is there a way for a charm to set config values when running a hook, say config-changed? [14:33] nextrevision: no, charms can't set configuration, only "humans" can [14:34] marcoceppi: so then best practice around providing, say a 'port' and 'ssl' config option, would be to trust the user to set the port to '443' and 'ssl' to true [14:35] nextrevision: You can just have them set "ssl" to true then have the charm make an opinion on what the port should be, so if ssl is true, port will just be 443 (then remove port from configuration) [14:37] marcoceppi: ok, have you found that it is better to include a port configuration parameter or to leave that logic up to the charm itself? has there been much demand for non default port configuration settings? [14:38] nextrevision: depends on the service, I think it's expected most web service will run on 80/443 but it really depends on what people's expereince is when setting up a service. Some services default to things like 8080, etc [14:38] marcoceppi: ok sounds good, thanks! [14:45] ahasenack: did you install charm-tools yet? [14:45] marcoceppi: no [14:45] Cool [15:14] jcastro, what happens in the weekly charmers meeting? [15:16] status updates, syncing with charm work [15:17] juju packaging seems to have gone wonky. pyjuju no longer gets installed to /usr/bin/juju (which might just be update-alternatives problems), and gojuju now just dumps itself into /usr/bin/juju so the docs on https://juju.ubuntu.com/get-started/ are now out of date [15:18] stub: yeah, I just noticed that this morning. Not sure what broke it or where to report it. [15:19] Not to be confused with the other getting started docs at https://juju.ubuntu.com/docs/getting-started.html ;) === wedgwood is now known as wedgwood_away === defunctzombie_zz is now known as defunctzombie [15:56] Weekly Charm meeting hangout will be here: https://plus.google.com/hangouts/_/d0ea0bcfb82ba886057c147953ed10eaabff6831?authuser=0&hl=en [15:56] arosales: marcoceppi m_3 mattyw and anyone else ^^ [15:56] jcastro, thanks === wedgwood_away is now known as wedgwood === medberry is now known as med_ [16:14] jcastro, arosales where can I find the notes? / agenda [16:14] mattyw http://pad.ubuntu.com/7mf2jvKXNa [16:14] arosales, thanks [16:15] mattyw, sure, np. We'll note that next time to paste with the hangout URL. [16:34] marcoceppi, https://code.launchpad.net/~mattyw/charms/precise/mongodb/auth_experiment/+merge/162887 [17:11] hazmat, I think juju 0.7+bzr628+bzr630~raring1 is on crack. I got the update and now have no commands. Well I do, they are in /usr/lib/juju-0.7+bzr628+bzr629~raring1/bin. How do you propose I solve this? downgrade? [17:12] that's not good... [17:12] * sinzui could change PATH [17:12] what heppends if you run update-alternatives? [17:12] I'd expect the install to have borked on the version, rather than happening then not linking [17:13] I did keep that stuff out of the ppa semi-deliberately, but it was asked for... [17:13] probably need to look at the scripts again [17:13] looking at the apt log might be enlightening [17:15] mgz, its been crack for two days.. it looks like the packaging changed? [17:15] mgz, mostly atm i see intermittent test failure on the builders atm, but not able to reproduce locally [17:16] hazmat: I'm in the process of writing a bug report for it, since my install is "broken" as well [17:16] I just downgraded to raring's universe package. [17:16] * sinzui hopes lxc-ls still works with this version [17:21] Actually, there's already a bug: https://bugs.launchpad.net/juju/+bug/1182774 [17:21] <_mup_> Bug #1182774: JuJu not found after having installed juju packages. [18:11] marcoceppi, sinzui origin back to distro ? [18:13] I did switch back to distro to get a working juju [18:14] hazmat: I've just been using absolute path [18:18] m_3 aksed why the ppa wasn't using update alternatives, which is because I didn't want the ppa actually tracking a branch daily any more... but I switched it, and apparently it also breaks things [18:19] was trying to fix the breakage from hazmat applying my distropatch on the branch [18:21] mgz, where it should have been.. [18:22] I don't really agree, but it was too late at that point anyway [18:29] marcoceppi, sinzui other option is specify the branch directly [18:29] for origin, till the ppa is working again [18:31] thanks hazmat, I locked to old version. I'll unlock when I see a fixed version [18:51] jcastro, grrrrrrrrr. u never replied back to me [18:51] :( [18:52] hmm? [18:52] oh, wrt. that openstack guy? [18:52] so I responded to him with like 3 emails of options for him [18:52] but he never responded. :-/ [18:57] jcastro, haha. ok [18:57] Daviey, poke poke [18:57] did you see my email [18:59] koolhead17: hey [18:59] koolhead17: i did indeed. [18:59] And i fully agree [19:00] Daviey, well you want me to write a big RANT if you think that will work [19:00] am telling you clock is ticking [19:00] tick tock [19:00] i know :) [19:00] if that will put some noise in ear of our big bosses in canonical [19:01] grrrrrrrr [19:01] i will be most unhappy to see all our hard work going down drain that`s it [19:01] :P === marlinc is now known as Marlinc === sarnold_ is now known as sarnold [21:08] jcastro, filing more bugs: [21:08] https://bugs.launchpad.net/ubuntu/+source/juju-core/+bug/1183109 [21:08] <_mup_> Bug #1183109: 'juju destroy-environment' missing from 'juju help'

[21:08] https://bugs.launchpad.net/ubuntu/+source/juju-core/+bug/1183110 [21:09] <_mup_> Bug #1183110: Handshake error when first bootstrapping with AWS

=== defunctzombie is now known as defunctzombie_zz [21:13] yeah! [21:13] hi jcastro [21:13] thumper: heya [21:14] * thumper looks for jono too [21:14] jono's done like 6 bugs today dude, you will be busy! [21:14] jcastro, jono: did you two want to have a quick hangout? [21:14] most of it small fry stuff I think :-) [21:14] I am literally out the door in 5 [21:14] jcastro: np [21:14] I just wanted you to be aware of the incoming papercuts [21:14] :) [21:14] jono: your second one might be related to https://bugs.launchpad.net/juju-core/+bug/1042106 [21:14] <_mup_> Bug #1042106: environs/ec2: certain operations should be retried if possible [21:14] I've not actually looked at the bug list for some time [21:14] probably should [21:15] very possibly [21:15] that sounds like a good idea thumper [21:15] marcoceppi, [21:15] :) [21:15] I think in general [21:15] when running a status [21:15] it is very easy to get stuck into new stuff [21:15] we suck at telling the user what is up [21:15] and forget what is there [21:15] it's like ... some weird error [21:15] that I agree with [21:15] when we can just say "Deploying the instance and the juju agent, hang tight yo." [21:16] heh, is there a locale for that? [21:16] man that would be funny to implement [21:16] "Deploying with Juju ... reticulating splines." [21:16] EN_st for "street english" [21:16] so, running juju status after I have detroyed my env gives me: [21:16] jono@forge:~$ juju status [21:16] error: The specified bucket does not exist [21:16] is that normal? [21:16] yes [21:17] lol [21:17] * jono files bug [21:17] heh [21:17] it should say "You've destroyed the environment, there's nothing here." and so on [21:17] that makes no sense [21:17] you guys can wishlist it :-) [21:17] jono: what it means is "oi, dumbass, there is nothing there" [21:17] aka "Dat bukkit's gone, da env probs destroyed yo" [21:17] you should run all errors through gizoogle's engine :-) [21:17] hunh, 'apt-cache search jive' returns nothing.. [21:17] marcoceppi: fyi, plugins should land today [21:18] well, we're too far along for these to be wishlist, it's getting to the time where we should really start polishing these up [21:18] thumper: Awesome, can't wait for the next release [21:18] jcastro, totally [21:18] jono: can you tag bugs "cli feedback" or something? [21:18] thumper, sure [21:18] hey [21:18] ta [21:18] papercuts [21:18] that's the perfect tag [21:18] that too [21:19] * thumper realises that he should integrate the new logging shizzle [21:19] then when local lands, and he sees juju switch [21:19] our lives will be complete [21:19] jcastro: local is third on my list [21:20] * thumper double checks [21:20] yeah, third [21:20] need to get general containers first, and fix machine addressability [21:20] then comes local provider [21:20] right [21:20] can't have local without containers! [21:21] thumper: so, we can expect local provider in a week then ;) [21:21] so general containers should give us a better co-location story [21:21] pfft [21:21] containers will be awesome, I'm really looking forward to that as well [21:21] https://bugs.launchpad.net/ubuntu/+source/juju-core/+bug/1183116 [21:21] <_mup_> Bug #1183116: 'juju status' output when no env is non-intuitive

[21:21] tagged with cli-ui and papercut [21:22] jono: awesome [21:22] marcoceppi: i'm hanging out for user namespace support in the kernel [21:22] man, a status bug just reminds me of how badly I want juju-top [21:23] jcastro: you mean juju watch? [21:23] juju-top sounds neat :) [21:23] marcoceppi: sure [21:25] jcastro: http://i.imgur.com/OhvuaEg.png we have the power now! [21:25] * thumper sighs [21:25] hah [21:25] I want to implement top (aka watch, observe) correctly [21:25] and plugins, while hacky, and may work... [21:26] is icky [21:26] thumper: you could, it'd just just superced the plugin :P [21:26] aye [21:26] robbie was pretty clear to me not to distract the core team with my crazy juju top idea. === mbarnett` is now known as mbarnett [21:26] jono: while messing around, I'd love it if you could think about what you'd like to see as 'help topics' [21:26] thumper: but if we find someone _new_, mwahahaha [21:27] jono: like yesterday, I went 'juju help constraints' and was frustrated that there wasn't anything there [21:27] jcastro thumper actually, it'd probably be better to write it as juju top and leave juju watch as something for core to land [21:27] jcastro: roger tells me that most of what we need is implemented already... [21:27] just the plumbing that needs to be sorted out [21:28] thumper: I was told to contact you as well, any chance we can get a quick look at: #1182224 -- from the outside looking in it seems easy and it blocks charms migrating from pyjuju. [21:28] <_mup_> Bug #1182224: relation-list returns null with json-output [21:28] thumper: yeah, I'd rather have containers first, heh [21:29] dpb1: interesting... [21:31] thumper, sure [21:31] dpb1: triaged, but not sure how soon I can get someone to it [21:31] one thing that is bugging me is 'juju status' [21:31] * thumper isn't really in charge [21:31] what an incomprehensible pile of nonsense :-) [21:31] haha [21:31] I am working on an alternative way of presenting that information :-) [21:31] jono: file a bug with what you want to see :) [21:31] thumper, will do :-) [21:32] jono: trust me, status is only going to show you more shit [21:32] all I care about with status is: [21:32] thumper: I was told you *were* in charge. believe in it! [21:32] * are my machines there [21:32] * is my shit connected [21:32] dpb1: hmm, by whom? [21:32] * where is it exposed [21:32] jono: 'juju status --summary' [21:32] jono: it isn't there yet, but might be an idea [21:33] thumper: oh, just kidding with you. Just was trying to get a hold of mark and was told you could fill in. [21:33] +1 to a --summary flag, but I truly love the verbosity of the output from status [21:33] thumper: thanks for looking at it. I'll be patient. :) [21:33] dpb1: heh [21:35] thumper, I think juju status should be the summary and juju status -v gives you the detail [21:35] :-) [21:35] filing bug now [21:35] I don't even like running juju status [21:35] what I want is to run juju watch or whatever [21:35] * thumper agrees with jcastro [21:35] and have a realtime representation of status [21:36] just think of the charm schools [21:36] instead of "watch juju status" [21:36] right [21:36] so basically, a CLI version of the GUI [21:36] anyway... [21:36] * thumper goes to land code [21:36] but for charm schools it doesn't matter, we show the real gui [21:37] so many tools and swamp code would break if it was required to be juju status -v [21:42] thumper, jcastro, marcoceppi: https://bugs.launchpad.net/ubuntu/+source/juju-core/+bug/1183129 [21:42] <_mup_> Bug #1183129: 'juju status' difficult to read and provides unneccessary information

[21:42] tagged with cli-ui [21:44] jono: I wonder if this would be better served as juju summary, or something similar. juju status has always been tradtionally presented in a way that was designed for both humans and machines to consume [21:44] * marcoceppi collects thoughts for the bug report [21:44] but with the API [21:44] that handles the machine consumption [21:45] marcoceppi, I don't think machine friendly output is human friendly [21:45] I recommend the primary way we recommend people check status provides a people-friendly format [21:45] right [21:45] no reason why juju status -yaml or something could not present another format [21:45] so now that we have an api that machines can consume [21:45] make status nice for us! [21:46] Is the API actually out for juju-core? [21:46] I mean, other than the websockets thing that kapil posted to the list a few weeks ago [21:46] one difference right now, is the format param doesn't control the volume of information [21:47] just how it is presented [21:47] marcoceppi, I agree that exposing json or yaml formatted status would be awesome so when I subprocess it from python I can consume it [21:47] but that should be a switch, not the default UI [21:47] I'm getting: ERROR [('SSL routines', 'SSL3_GET_SERVER_CERTIFICATE', 'certificate verify failed')] whenever I try to run juju deploy [21:48] any ideas how I should troubleshoot? What is the url that "etherpad-lite" for instance actually tries to load/ [21:48] ^ in juju deploy etherpad-lite [21:49] jono: thumper I think git does a good job of this when viewing the juju log. There's things that control format, then things that control data volume "prettyprints" that might be interesting to have in a status command [21:49] gQuigs: Can you try juju deploy with a -v flag for more verbose output? [21:49] (then put it in a pastebin somewhere like http://paste.ubuntu.com) [21:49] marcoceppi: oops, I forgot to post that: http://pastebin.ubuntu.com/5691834/ [21:50] gQuigs: Are you behind a firewall or proxy? [21:50] it seems like it must be an SSL issue on my machine... because using openssl works fine to store.juju.ubuntu.com [21:51] marcoceppi: technically yes, but not anything that interesting (standard NAT/Firewall box) [21:52] https://blueprints.launchpad.net/juju-core/+spec/s-cloud-juju-core-plugin complete [21:52] gQuigs: O.o your SSL is being MITMed? [21:53] gQuigs: I wasn't able to replicate the error on my machine using juju-0.7, I'm not sure what would be causing that error except twisted bad certs locally or an actual mitm [21:54] If I'm being mitm it's juju specific.. so I guess that leaves bad certs.. [21:56] marcoceppi, yeah, I think there are a few options for improving it [21:59] gQuigs: There might be something else, but that's the first time I've encountered that error in particular [22:07] yup, ca-certificiate reinstall fixed it. sorry for the noise [22:17] gQuigs: no problem! Glad that's all it took to resolve [22:18] any ideas why juju-0.7 errors with this during deploy: ERROR [('PEM routines', 'PEM_read_bio', 'no start line')] [22:21] verbose output here: http://pastebin.ubuntu.com/5691915/ [22:22] bah, why couldn't it at least include the filename that failed..? [22:25] dpb1: found the source of your bug [22:26] * thumper considers [22:26] thumper: ohh? easy fix? [22:27] still looking [22:28] interestingly, the HasLen check in the tests succeed, as (nil, HasLen, 0) [22:28] paraglade: you're the second person to be having SSL issues trying to connect to the store today [22:29] dpb1: yeah, one line fix, and I found the line [22:29] not sure if anything else will break [22:29] paraglade: The the errors are different, it looks like txaws is having problems loading CA certs [22:30] marcoceppi: awesome :) [22:30] paraglade: If it's anything like the last users issue, trying reinstalling ca-certificiates as that resolved the issue for them [22:31] well I just deleted the .juju/cache and my local data-dir and I am not able to deploy [22:31] s/not/now/ [22:31] sorry [22:31] thumper: ya, that is a problem. It strikes me as wrong to put the fix in the charm since it's not proper json text. But I can see the quandry. [22:32] paraglade: so, it's resolved? [22:32] https://www.ssllabs.com/ssltest/analyze.html?d=jujucharms.com < welp [22:32] dpb1: I'm looking at it now [22:32] kyhwana: jujucharms.com isn't the charm store [22:32] marcoceppi: hrm, http://paste.ubuntu.com/5691944/ [22:33] oh. what -is- the charm store? :) [22:33] thumper: k. if you need anything just ping. I'll be afk in a bit [22:34] dpb1: I think it'll be fine [22:35] sarnold: I believe it's store.juju.ubuntu.com but I'm not 100% certain. I'd have to dig through the source code again to verify [22:35] marcoceppi: yup. my mysql node is starting up now on a local provider [22:35] paraglade: ah, didn't realize you were using the local provider [22:37] https://www.ssllabs.com/ssltest/analyze.html?d=store.juju.ubuntu.com&s=91.189.95.66 [22:38] kyhwana: I don't personally run the store, so I can't comment, but this line "This server supports insecure suites (see below for details). Grade set to F." seems to be why they've decided to give it an F ranking [22:38] kyhwana: if you' [22:39] d like you can open a bug against the charm store with that information. You'd at least get a response from those maintaining it [22:39] marcoceppi: yeah, the anon cipher suites don't provide any authentication [22:41] * marcoceppi looks for charmstore project [22:44] kyhwana: if you want to open a bug about it, do it against the juju-core project. It looks like the charmstore is housed inside that project. [22:46] marcoceppi: alright, will probably have to wait till I get home [22:46] kyhwana: at the very least they'll be able to respond with why it is as it is === defunctzombie_zz is now known as defunctzombie [22:57] dpb1: fix submitted for review [23:01] juju-core on arm ping [23:06] kyhwana, thanks [23:06] mwhudson, better ping luck with that on #juju-dev [23:06] oh i didn't know about that channel [23:07] mwhudson, ping cheney (aussie) about it.. he's got a bank of arm test runner [23:07] runners [23:07] tz delta isn't quite good for it atm [23:08] ok [23:08] i'll email again then [23:08] cheers :) === wedgwood is now known as wedgwood_away [23:19] kyhwana, marcoceppi fixed re ssl [23:20] clear cache on page to verify [23:21] hazmat: hmm? [23:22] kyhwana, https://www.ssllabs.com/ssltest/analyze.html?d=jujucharms.com [23:22] hazmat: oh, awesome [23:23] tho https://www.ssllabs.com/ssltest/analyze.html?d=store.juju.ubuntu.com&s=91.189.95.66 is still showing F [23:25] kyhwana, sadly i don't have access to that one, i'll raise a flag to those that are [23:26]