[01:53] SpamapS: When you get a chance can you take a look over the Minecraft charm merge? [01:53] https://code.launchpad.net/~marcoceppi/charms/oneiric/minecraft/trunk/+merge/97132 === freeflyi1g is now known as freeflying === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away === TheMue_ is now known as TheMue === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away === almaisan-away is now known as al-maisan === asavu_ is now known as asavu === asavu_ is now known as asavu === asavu_ is now known as asavu === asavu_ is now known as asavu [11:02] hi, can I experiment with juju on my single computer? [11:02] without using ec2? === al-maisan is now known as almaisan-away [11:03] according to the docs I cannot (it says lxc support is not ready) [11:03] but installing juju on precise pulled in lxc [11:03] so what is it? === almaisan-away is now known as al-maisan [13:00] zyga, lxc works, so you can use that [13:00] zyga, where did you see that it isn't ready? [13:01] james_w, in the official juju docs, although they keep talking about natty so they may just be old [13:01] let me pull the link [13:03] james_w, https://juju.ubuntu.com/docs/faq.html?highlight=amazon%20ec2 [13:03] james_w, "Does juju only deploy to the Amazon EC2 cloud?" [13:04] zyga, thanks [13:04] james_w, do you know of a tutorial that would allow me to learn about juju using my desktop and lxc/kvm? [13:04] james_w, I think those docs are out of date in other areas too: https://juju.ubuntu.com/docs/getting-started.html#running-from-ppa === marrusl_ is now known as marrusl [13:05] juju: https://juju.ubuntu.com/CharmSchool [13:05] err, zyga ^ [13:05] thanks [13:05] jcastro, where do we file/fix those doc bugs? [13:06] lp:juju/docs [13:06] is the ppa still recommended? [13:08] for oneiric? [13:08] yes, for precise, not sure. [13:08] for precise [13:08] i've been trying to run distro as much as I can [13:09] I'd say run the distro version in precise [13:10] so that you can report problems and they can be fixed [13:19] on https://juju.ubuntu.com/CharmSchool I see an example local environment [13:20] since I want to use distro packages what should I say in juju-origin: [13:21] Leaving it out should just work [13:34] http://pastebin.ubuntu.com/883240/ [13:34] virbr0 is already up [13:34] it was up on boot [13:35] my /etc/network/interfaces contains static configuration for eth0 and lo, if that is relevant [13:40] * zyga tries to reboot [14:18] so rebooting did not help [14:18] what should I do when juju fails to boostrap with "2012-03-14 15:16:01,188 ERROR Command '['virsh', 'net-start', 'default']' returned non-zero exit status 1" [14:32] zyga, can you check that the account you are using is in the libvirtd group please [14:33] jamespage, it is not [14:33] jamespage, should that be automatic? [14:33] zyga, no [14:33] ah, I see [14:33] sigh [14:33] juju should tell me that [14:33] zyga, that should be in the docs and juju should tell you - please raise a bug [14:34] reporting now [14:34] zyga, TBH as soon as you add that group it will be able to see the default network (which should already exist) [14:34] it does [14:34] sigh [14:34] can I do this without logging out? === al-maisan is now known as almaisan-away [14:37] https://bugs.launchpad.net/ubuntu/+source/juju/+bug/955110 [14:37] <_mup_> Bug #955110: juju should tell me that I'm not in libvirtd group when running juju bootstrap

< https://launchpad.net/bugs/955110 > [14:38] zyga, I don't think so [14:39] newgrp libvirtd [14:40] james_w, wow, I learned something very cool today, thanks [14:42] nice [14:47] zyga, hmm - the docs do actually refer to this - https://juju.ubuntu.com/docs/provider-configuration-local.html [14:47] but it would be nice for juju to check === zyga is now known as zyga-afk [15:42] hi, it's me again and I'm having trouble with juju lxc [15:43] I'm am trying to perform the mir for juju, but can't get it to work the way I thought it should work [15:43] s/mir/mir review/ [15:44] you may recall last time I tried to do this inside a vm. today I am trying on bare-metal [15:47] here are the commands I ran: http://paste.ubuntu.com/883423/ [15:47] looking at juju debug-log, I see: [15:47] 2012-03-14 10:27:59,548 Machine:0: juju.agents.machine INFO: Started service unit wordpress/0 [15:47] jdstrand, what's the current status output? [15:47] (it took a while to bootstrap) [15:48] jdstrand, bootstrap on local doesn't do very much, it does most of the work creating the first unit [15:48] and subsequent units in that env are fast (lxc-clone) [15:48] hazmat: http://paste.ubuntu.com/883430/ [15:48] hazmat: I meant debootstrap [15:49] hazmat: I did notice some firewall denials on the host, which thought could be the problem. so I did: sudo ufw allow in on virbr0 from 192.168.122.0/24 [15:50] hazmat: I have not exposed this yet, cause the state is still pending [15:51] jdstrand, hmm [15:51] jdstrand, can you ssh into the unit [15:52] jdstrand, you did the firewall manipulation after launching the unit? [15:52] er. deploying the charm [15:52] hazmat: I only noticed the firewall denials after deploying the charm, so yes, after [15:52] I did not try to do the charm again [15:53] hazmat: is there a convenient way to get the ip address for the machine? [15:53] s/machine/container/ [15:54] (remember I'm still a juju noob :) [15:54] jdstrand, yeah.. if the container didn't external connectivity during its setup, that would be problematic for it.. inside of $data-dir/units there should be symlinks to the unit logs which would be helpful to see [15:55] hazmat: what is the proper way to blow away this charm, destroy-service? [15:55] jdstrand: to get the ip btw, I usually just 'grep dnsmasq /var/log/syslog' [15:56] SpamapS: ah right, this is just a libvirt thing. thanks [15:56] * jdstrand is not new to libvirt [15:56] Right it uses the libvirt networking [15:57] ssh ubuntu@192.168.122.145 worked [15:57] jdstrand: next thing to check is whether or not juju is running (should be, via an upstart job) [15:58] I think.. I'm not sure if that landed before or after 457, which is what is in precise [15:58] SpamapS: on the host, right? [15:58] jdstrand: there should be a unit agent running inside the container [15:58] jdstrand: and there would be a machine agent running on the host [15:58] $ ps auxww|grep juju [15:58] root 372 0.0 0.6 34972 14036 ? Ss 15:28 0:01 /usr/bin/python -m juju.agents.unit --logfile=/var/log/juju/unit-wordpress-0.log [15:59] (that's in the container, so yes) [15:59] yay [15:59] jdstrand, yes [15:59] jdstrand: so you should also have a charm log visible on the host at ${data-dir}/${USERNAME}-${environment-name}/units/wordpress-0/charm.log [16:00] the agent is running on the host as well [16:00] jdstrand, there's a machine agent juju.agents.machine on the host that's deploying the containers + unit agents [16:00] hazmat: remind me, why doesn't debug-log show install anymore? [16:01] SpamapS: no charm.log, only container.log and unit.log (a symlink) [16:01] SpamapS, argh.. i thought that was fixed, it hasn't really ever done it reliably to be honest. its worth a bug report [16:01] jdstrand: oh I think unit.log is the one. [16:01] yup [16:01] hazmat: there's an open bug [16:02] bug #760818 [16:02] <_mup_> Bug #760818: debug log is initialized too late in the unit agent startup, to catch install/start hooks < https://launchpad.net/bugs/760818 > [16:02] sheesh that's old [16:03] * hazmat needs caffienates [16:05] SpamapS: what am I looking for in unit.log? [16:06] jdstrand: that should be the record of the install of wordpress [16:06] I do see some zookeeper errors [16:06] the timeouts are just ZK being pedantic [16:07] it ends wth initiated connection to server [192.168.122.1:56835] [16:07] session establishment complete on server [192.168.122.1:56835] [16:07] Closing zookeeper sessionId=0x1361181aacd0007 to [192.168.122.1:56835] [16:07] there is nothing about wordpress [16:07] (except ZOO_INFO@log_env@662: Client environment:host.name=jamie-local-wordpress-0) [16:08] interesting.. I guess I'm forgetting where the charm log ends up [16:08] SpamapS: perhaps destroy-service and try again now that my firewall is in order? [16:09] jdstrand: you can just 'deploy local:wordpress wp2' and keep the old one around for comparison.. [16:10] jdstrand: I'm deploying here so I can jog my memory [16:10] SpamapS: so, the first time I used: [16:10] juju deploy --repository=/usr/share/doc/juju/examples local:oneiric/wordpress [16:10] this time I should use: [16:10] juju deploy --repository=/usr/share/doc/juju/examples local:oneiric/wordpress wp2 [16:10] correct? [16:11] jdstrand: correct [16:11] 'deploy' command finished successfully [16:11] jdstrand: naming the service after the charm is just a convenience [16:11] it still says pending [16:11] * jdstrand check firewall logs again [16:13] did have some ipv6 ping denials in the forward chain... [16:13] * jdstrand disables firewall and does wp3 [16:13] jdstrand: its entirely possible that master-customize step failed.. which would make all subsequent deploys fail. thats in ${data-dir}/user-env/units/master-customize.log [16:14] hazmat: been meaning to poke you about that.. master-customize seemes to not detect errors very well [16:14] SpamapS: it looks ok [16:15] apt-get update [16:15] id ubuntu [16:15] etc [16:15] Container Customization Complete [16:15] (of course, I think we knew that cause I could ssh into the machine) [16:16] SpamapS: after 'deply' what am I supposed to see in 'juju status'? [16:16] jdstrand: well, thats just the thing, one of the steps in the middle can fail and it still starts the containers [16:16] deploy* [16:16] jdstrand: you should see the unit wordpress/0 with a state: started [16:16] or, wp3/0 [16:16] <_mup_> Bug #955209 was filed: charm.log uses 'Error' for all stderr messages < https://launchpad.net/bugs/955209 > [16:17] SpamapS: yeah, I've never seen that :( [16:17] SpamapS: I'll be happy to paste master-customize.log, but it looks ok [16:18] http://paste.ubuntu.com/883485/ [16:18] jdstrand: is that more like what you have? [16:19] hazmat: would be nice if destroy-environment cleaned up datadir some.. I just realized mine has a ridiculous amount of cruft in it [16:20] SpamapS: yes, that is exactly what I have for wordpress/0 [16:21] SpamapS: wp2/0 and wp3/0 now have a public-address [16:21] jdstrand: but still pending? [16:22] SpamapS: master-customize.log: http://paste.ubuntu.com/883488/ (would be great if I could set a local mirror) [16:22] SpamapS: juju status: http://paste.ubuntu.com/883491/ [16:22] oh, I just ran it again and it started! [16:23] both wp2 and wp3 [16:23] yay [16:23] SpamapS: how long does it typically take for a charm to go to 'started'? [16:24] jdstrand: the local provider is particularly slow because it has to download everything as individual packages and unpack them... [16:25] jdstrand: with EC2, you have the Ubuntu cloud image, which has a lot of deps already installed, so the time between saying 'deploy' and the install hook runite a bit less [16:26] jdstrand: On my very-slow-disk macbookpro, I said 'deploy' some 8 minutes ago, and its justn ow instaling the wordpress package. :-P [16:26] SpamapS, hazmat: thanks for your help and explanations. seems my bare-metal woes were firewall related. I will continue this later (I have calls now) [16:27] SpamapS: ok, thanks. that is helpful [16:27] SpamapS: it would be nice if there was a command to see the progress of the deployment... [16:27] anyhoo-- call! [16:27] * jdstrand steps away === zyga-afk is now known as zyga [16:35] jdstrand: btw, the request for allowing a local mirror is here; bug #897645 [16:35] <_mup_> Bug #897645: juju should support an apt proxy or alternate mirror for private clouds < https://launchpad.net/bugs/897645 > [16:36] * SpamapS raised it to High since its such a common request [16:36] SpamapS: awesome, thanks :) === Furao_ is now known as Furao === Leseb_ is now known as Leseb [17:45] m_3, I just did a review call with msm about our events and stuff. [17:45] all our major events are handled [17:46] m_3, but we need to know your plans for all the language confs your hitting this summer [17:46] so she can ping the organizers and get you what you need. [17:46] so if you could ping her this week that would be grand. [18:05] I did a review-of-sorts of https://code.launchpad.net/~rockstar/juju/setuppy-fixes/+merge/96902. I'm not an approved reviewer so I marked it as a comment in hopes that it will be helpful to the person that reviews it for real. [18:11] Hi guys, is it possible to call a custom cloud-init script from a juju charm? [18:12] I have some custom initialization I would like to use on all of the hosts I create, and cloud-init's config files could save me from writing boilerplate code. [18:21] mars: we're working on supporting the "policy" use case in a somewhat different way [18:22] SpamapS, ? [18:22] Sorry, I don't understand what the 'policy' use case is [18:22] mars: basically a new feature called "subordinate charms" will land soon that will allow you to define policy for the box and deploy it along with other charms [18:23] mars: what you're talking about is policy.. it goes across all your machines.. you want users or monitoring or whatever... [18:23] mars: so, you'll write a charm that enables that, and then say 'deploy mystuff' and then 'add-relation service1 mystuff' and 'add-relation service2 mystuff' and they'll be deployed together [18:24] benji: thanks btw! [18:24] ok [18:24] SpamapS, I was looking to do something a bit different [18:25] There are few problems with LXC Lucid containers that I was hoping cloud-init could help smooth over [18:25] SpamapS: my pleasure [18:25] And one of my charms needs to have a custom repository set up in order to fetch the sourcecode: again, cloud-init makes that very simple [18:26] mars: lucid is not a tested target for juju.. we don't test on python2.6.. so its possible the agents won't even start up [18:27] mars: so do shell scripts. :) [18:27] hmm, good to know [18:27] mars: seems like your install hook is a simpler place to setup your custom repository than cloud-init [18:27] SpamapS, yeah, I have written that in shell, and in Python fabric. But I wanted to skip the boilerplate 'bootstrap' targets. [18:28] I do think we need to look at adding some kind of way to easily include shared code in charms. Right now the best way to do that is with packages in PPA's.. but that seems too complicated for the simpler use cases [18:39] SpamapS, so Google has a couple references to Juju using cloud-init internally. If that is true, then I assume there is no way to extend or call cloud-init with my additional configuration? [19:08] mars: juju uses cloud-init for orchestra and ec2, but its not fundamental to juju.. [19:09] mars: so if you do any customization around cloud-init .. you won't be able to repeat on the local provider, which does not use cloud-init [19:10] ah [19:10] ok [19:11] mars: I agree that boilerplate is bad.. juju needs to do more to allow cross charm sharing of code. Right now.. packages are the best way to do that. [19:11] mars: note that there is precedent for this in the charm-helpers-sh package [20:30] <_mup_> juju/increase-session-timeout r450 committed by kapil.thangavelu@canonical.com [20:30] <_mup_> increase default session expiration timeout from 10s to 20s (same ping count/timeTick) [20:31] <_mup_> juju/force-upgrade r463 committed by kapil.thangavelu@canonical.com [20:31] <_mup_> remove cli short switch, force should be explicitly spelled [20:32] <_mup_> juju/scheduler-peek-list r473 committed by kapil.thangavelu@canonical.com [20:32] <_mup_> additional error scenario testing around revamped scheduler [20:45] <_mup_> txzookeeper/managed-watch-and-ephemeral r47 committed by kapil.foss@gmail.com [20:45] <_mup_> new managed client subclass that tracks watches and ephemerals [21:18] so, I rebooted my machine that had local working right, and now get: [21:18] $ juju status [21:18] could not connect before timeout [21:18] 2012-03-14 16:05:21,185 ERROR could not connect before timeout [21:18] nothing is running-- not zookeeper, not the agents, not the containers [21:19] I tried 'juju bootstrap' again, and it told me it was already bootstrapped [21:19] is there some procedure I missed for rebooting a machine with local lxc and juju? [21:21] jcastro: will do [21:44] SpamapS, hey, you around? I'd like to talk about our team's difficulties with the debian build for the python charm helpers [21:44] my team's, I should say :-) [21:46] gary_poster: yeah [21:47] gary_poster: been fighting the email demons all day, but I have a todo to review that whole merge proposal. :) [21:48] Thanks SpamapS. Yeah, I've been fighting my own demons today too. The short of it is that we've lost a lot of man hours on trying to get the debian packaging working on that. Could we do one of (a) toss it to you, (b) make a separate package and toss it to you, or (c) have some close attention from you to pair with someone to get this packaged? We'd love to increase our packaging fu, but that's more time for you I suspec [21:48] t. [22:01] gary_poster: toss to me [22:01] gary_poster: will try to carve out some time today, otherwise I'll take a look tomorrow firs tthing. [22:02] SpamapS, ok, many thanks. gmb, do you have something you should merge into that mp, or should SpamapS just look at the original stuff from bac? [22:02] gary_poster, SpamapS: I think the original stuff from bac is sufficient. My additions either a) had no effect or b) broke everything. [22:02] lol [22:02] I'll make sure it's in a pristine state. [22:03] ok, thanks again SpamapS, and thanks gmb === jrgifford is now known as Guest18667 [22:25] SpamapS: is there any trick to restarting a machine with a 'local' configuration? [22:26] SpamapS: the host that is. if I reboot the host all is gone on reboot [22:26] SpamapS, I've updated the branch to remove my cruft. Best of luck; let me know if you need any help so that I can be conveniently sick that day. [22:26] jdstrand: in theory, it should all come back up.. as it should have installed an upstart job to start the machine agent and zk [22:26] jdstrand: in practice, I have not tested that. ;) [22:26] (ie, zookeeper, twistd, agents, instances and anything else I forgot is not started) [22:27] SpamapS: I guess I will file a bug [22:27] jdstrand: on my system, I see a juju-clint-local-file-storage.conf and a juju-clint-local-machine-agent.conf .. but nothing that starts ZK [22:28] gmb: remind me where the branch is? [22:28] SpamapS: I don't have any /etc/init/juju* on the host [22:29] jdstrand: should be put there after 'bootstrap' .. [22:29] SpamapS: I can say after employing my workaround in bug #955540, I was actually able to use juju wholly within a vm [22:29] <_mup_> Bug #955540 was filed: juju-create hard-coded to use 192.168.122.1 < https://launchpad.net/bugs/955540 > [22:29] <_mup_> Bug #955540: juju-create hard-coded to use 192.168.122.1 < https://launchpad.net/bugs/955540 > [22:29] jdstrand: its entirely possible that landed after r457 [22:30] SpamapS: is there a new version of juju expected in precise soon? [22:30] jdstrand: yes there will be a FFe filed as soon as one last feature is done. [22:31] jdstrand: thanks for the thorough reviews.. lots of stuff to work on. [22:32] SpamapS: heh-- at this point I am just trying to get it to work so I can understand it and pke at it :) [22:32] but np [22:33] its pretty cool once I got things working [22:33] it's [22:34] jdstrand: I think we need to make it more visual.. more feedback between the agent coming up and state: changing to started.. so you can watch more [22:36] SpamapS: that would be helpful. I'm also finding the lack of man pages with toy and real world examples to be a hinderance (that will be in my MIR review) [22:36] I feel like I can actually do enough to test it. but to deploy it in the real world I have a lot too learn [22:36] (particularly around zookeeper security) [22:38] jdstrand: I wrote a man page 6 months ago but the devs didn't want to keep it up to date. :-/ perhaps I should add it back to the debian package and just commit to updating it as part of the packaging. [22:38] jdstrand: https://bugs.launchpad.net/juju/+bugs?field.tag=security [22:38] jdstrand: https://bugs.launchpad.net/juju/+bug/813773 [22:38] <_mup_> Bug #813773: Juju should have security rules/acls for every path in zk < https://launchpad.net/bugs/813773 > [22:39] SpamapS: excellent, thanks. are these bugs planned on being fixed in precise, or maybe 12.04.1? [22:40] jdstrand: I'm not aware of plans to complete any of those no [22:41] SpamapS: re man page> I find them immensely useful, and so do others (as you know, you wrote one :). there are so many urls to go to it is confusing to know what to use. most of the ones I saw were out of date (didn't mention lxc) [22:41] heck, the man page could say: for more information, please see http://...' [22:41] but something should be up to date :) [22:42] I'm sure there are plenty of docs that are up to date-- I just didn't know where to find them is all [22:45] jdstrand: good point. I think there's going to be a big documentation push very soon. [22:45] SpamapS: that is great to here [22:57] SpamapS, lp:~gmb/charm-tools/add-charm-helpers [22:58] gmb: awesome, thanks [23:02] jdstrand, that should work with the ppa [23:02] re reboot [23:02] not sure that it does with the last precise upload [23:02] which was right before the restart work landed [23:05] "I'm sure there are plenty of docs that are up to date-- I just didn't know where to find them is all" [23:05] hah [23:05] soon, soon! [23:06] ah cool, a bug for murder, that should be interesting [23:07] SpamapS, you still working? Because if you're going to do an initial review of the subway charm I'd like to shoulder surf [23:31] anyone seen an issue where, after deploying a number of services, a service is missing its machine units? [23:31] http://paste.ubuntu.com/884080/ [23:32] ^ rabbitmq gets deployed but a machine never provisioned or associated with the service [23:32] ive been seeing this somewhat frequently but randomly over the last week [23:51] jcastro: yeah still working [23:51] SpamapS, I found some things for him to improve [23:52] so no need to review it [23:52] adam_g: for your issue, have you looked through the provisioning agent log? [23:53] jcastro: aight [23:55] SpamapS, https://bugs.launchpad.net/charms/+bug/944246 [23:55] <_mup_> Bug #944246: Charm Needed: Subway IRC client/server < https://launchpad.net/bugs/944246 > [23:55] yeah! I helped! [23:56] SpamapS, also I owe the team, I used the word idempotent when talking to him [23:59] aahhhh!!!!!! [23:59] * SpamapS responding to idempotent