=== jono is now known as Guest43238 [00:25] I am looking for pointers on how to use juju OSX client with a local VagrantBox [00:26] I see that I have to juju switch manual (the ~/.juju/environment.yml contains a "null" instead of "manual", though( [00:27] lemao: I think the 'null' provider is renamed 'manual' in a development branch. probably 'null' still works for whatever you've got installed.. [00:28] sarnold: ok. It seems that my initial problem is that vagrant boxes use port 2222 instead of port 22 for ssh. How can I change the default ssh port in the environment? [00:28] lemao: hrm, I don't know for certain, but probably ~/.ssh/config could be omdified to use the port 2222 [00:29] sarnold: true. Forgot about that. It would be nice though to have a self-contained configuration for juju ... Let me try that for now [00:42] lemao: there is also this workflow https://juju.ubuntu.com/docs/howto-vagrant-workflow.html [00:42] lemao: yeah, I know what you mean... [00:43] lemao: did you get it sorted out? [00:43] and the folks at altoros whipped up https://github.com/Altoros/juju-dev-box [00:43] though sarnold may have gotten you what you need too :-) [00:44] heh, the Altoros thig looks pretty cool [02:28] sarnold: sorry, stepped out for dinner. Let me take take a look at the urls [02:39] sarnold, arosales: I was trying something slightly different: using juju OSX to interface with a Vagrant JujuBox, but just vagrant ssh may be the simplest route. [02:39] sarnold, arosales: but the link is superb since it covers the next question: how to create charms in this env === vladk|offline is now known as vladk === vladk is now known as vladk|offline === CyberJacob|Away is now known as CyberJacob === zchander_work is now known as zchander [08:17] hey guys, does `unit-get public-address` give the public address? or that command doesn't exist? === CyberJacob is now known as CyberJacob|Away [09:18] jamespage, what should i do about keystone and postgresql? what branch should i use as base? === vladk|offline is now known as vladk [10:06] yolanda, hmm [10:06] one second - just reviewing branches atm [10:31] marcoceppi, if you have time - https://code.launchpad.net/~openstack-charmers/charm-helpers/ssl-everywhere/+merge/209301 [10:31] I think that is the last big branch for this cycle; once we have that landed I can start landing ssl-everywhere and icehouse branches for the charms === Overmind is now known as overm1nd [11:14] jamespage: ack, will review in 30 [11:14] What would be the best way to test for an existing relation (e.g. a relation to a MySQL charm is required), so I can keep the charm stopped until the relation established? [11:17] marcoceppi, thanks [11:17] zchander, you can query relations from any relation context - so you could write a check to see if a) mysql relation is present and b) mysql has provided credentials [11:17] which would indicate its all good to work [11:19] jamespage: Thanks. Going to search for that… (maybe I am going to ask a über-noob question, but where can I find documentation about all available relation/etc context keys I can query from my shell scripts?) [11:20] zchander, juju docs (http://juju.ubuntu.com/docs) [11:20] the commands you need are [11:20] relation-ids [11:20] relation-list -r [11:20] and relation-get -r attribute [11:21] relation-list gives you a list of related service units. [11:22] zchander, this is written in python but it give you and idea - http://bazaar.launchpad.net/~charm-helpers/charm-helpers/devel/view/head:/charmhelpers/core/hookenv.py#L295 [13:05] so i'm working on a Meteor charm. Meteor is basically a node app that requires mongodb (won't run without it, as far as i know). my thought is to make the "start" hook a no-op until/unless the mongodb relation is joined. is there a better way? [13:07] tvansteenburgh: you could create a dot file, like .db-added, which the start hook checks, if it exists, start the service, if it doesn't exit 0 [13:07] tvansteenburgh: so every hook could call hooks/start which wouldn't actually happen until it was ready [13:07] atlernatively, just having the service start in the db-hooks is fine too [13:08] ok cool, thanks [13:10] and all the hooks run as root right? [13:12] yup [13:12] is there a preferred place for installing stuff (stuff that doesn't come from apt) [13:12] like /opt or other? [13:12] or no one cares? :) === roadmr_afk is now known as roadmr [13:23] tvansteenburgh: well, IS says /srv/ [13:23] tvansteenburgh: but anywhere outside of $CHARM_DIR is good, follow whatever the community expects [13:24] the community of the service you're setting up* [13:24] ok [14:04] frankban: around? Thanks for the reply. Can we keep this conversation in the bug, please? [14:04] frankban: specifically then, I think think mterry's question is that for juju-quickstart installed on Ubuntu from the official archive, why is its behaviour not to install juju also from the same archive? [14:05] marcoceppi: any chance meteor would run in node app [14:05] arosales: maybe, haven't tried [14:05] tvansteenburgh: ^ [14:06] tvansteenburgh: we do have a node.js framework charm, "node-app" that can wrap node apps . . . [14:06] frankban: and if we could do that for the on Ubuntu from archive special case, then it could be done via dependencies in packaging, and so eliminate a need for a sudo call for that. [14:07] rick_h_: ^^ [14:07] arosales: meteor uses node to run the app, but you still need the meteor framework [14:07] rbasak: otp the moment but will catch up [14:08] tvansteenburgh: ah ok so mabye not a good fit for the node-app charm, but thought I would check [14:08] arosales: if the app was already bundled i guess you could run it on node/mongodb [14:11] Just a ‘quick’ non-juju question (not directly)………. I am strugling to replace a line in a php file (config) file using sed (in bash). [14:19] rbasak: have time to chat real quick? I'm not 100% following the question there [14:19] rbasak, rick_h_ : I'll copy observations on the bug. I think the point is that quickstart wants to set up a working juju installation independently from how you obtained the application. E.g. in saucy the PPA is required because the juju version we support (and other dependencies) is not in the official archives. In general, the idea is that we set up the code sources we require, and put there the things we want to ins [14:19] tall. In trusty, quickstart currently installs the juju-core located in universe (since it is the most recent version), and that's fine, but we'd like to have the ability to keep that under control, so that we can guarantee the same user experience on different times and series. [14:20] rbasak: I think we're walking a line of company wishes vs packaging rules/standards and might need to get a better idea on those rules [14:22] rick_h_, frankban: sure. Hangout? [14:22] rbasak: sure thing, sec I'll get a url [14:23] rick_h_: I need to reboot. Five minutes. Please can you see if you can find me and send me an invite from within the app? I can't c&p the URL to my hangout machine :( [14:23] rbasak: https://plus.google.com/hangouts/_/7acpidcv17rkrcgi24kle76rpk?authuser=1&hl=en [14:23] rbasak: will do [14:24] frankban: will redo in 5min [14:25] sorry about the confusion there [14:25] np [14:28] frankban: https://plus.google.com/hangouts/_/7ecpjtb43orhuo5sivafe2e6m8 === hatch__ is now known as hatch === vladk is now known as vladk|offline === Darkwing_ is now known as Darkwing [15:21] lazyPower: hey, mind if I ask a little question? [15:21] sure, whats up jose? [15:22] erm, give me a couple mins so I can try this again and make sure it *is* failing [15:28] jose: ack. Ping when ready [15:34] lazyPower: disregard, looks like yesterday tiredness was playing a trick on me :) [15:34] jose: fatigue is a developers 'best friend' [15:34] :P [15:39] lazyPower: now an actual question: what should be done in case a charm has no copyright file? just file a bug and assign the maintainer? [15:42] jose which charm doesn't have a copyright? [15:42] it must be an old one [15:42] marcoceppi: correct, 2011, cf-mongodb, cf-mysql and cf-redis afaict [15:43] haven't looked further [15:43] they don't have a README nor an icon or categories, but those I can fix [15:43] jose those are negronjl charms I think [15:43] yep [15:44] you can ping him and have him update with copyright files [15:44] there shouldn't be many others floating around [15:44] ok, I found another 2 charms [15:44] I'll send him an email in a while, then :) [15:45] jose any other you find just ping me with the names [15:45] cool [15:45] thanks [16:00] jcastro: hey, don't forget to push your own fix to the postfix charm (Markdown), I have no permissions [16:06] lazyPower: btw - haven't forgotten about giving you feedback on relations / docs, just ended up being massively busy last night, feel free to give me a poke if I don't get round to it ;) === vladk|offline is now known as vladk [16:23] marcoceppi: to push a fix for the juju gui I need to have signed Canonical's Contributor Level Agreement, right? [16:24] wesleymason: perfect, no rush :) === roadmr is now known as roadmr_afk [17:25] <_sEBAs_> hi! [17:26] <_sEBAs_> im trying to connect juju with our openstack using lxc as a hypervisor [17:27] <_sEBAs_> but im getting an error at the time to get the container at the bootstrap [17:27] <_sEBAs_> open.go:278 failed to write bootstrap-verify file: cannot make Swift control container: failed to create container: juju-cece0b9817a68cba4780784bf0663e45 [17:27] <_sEBAs_> Im stuck in this like a day now [17:27] <_sEBAs_> someone knows whats the problem here? [17:28] <_sEBAs_> i sow some issues related to this, where i have to do a "juju metadata generate-image......" [17:28] <_sEBAs_> but I thought juju bootstrap was doing that already :P [17:31] _sEBAs_: are you doing the bootstrap with sudo? [17:31] <_sEBAs_> ow [17:31] <_sEBAs_> not [17:31] <_sEBAs_> jose: I will try that! [17:32] afaik when bootstrapping to lxc you need to do sudo, otherwise juju won't have enough permissions to create the container [17:32] <_sEBAs_> but in the openstack provider? [17:32] maybe yes? [17:32] give it a try [17:33] <_sEBAs_> ok, I would thanks!! [17:38] <_sEBAs_> now something changed, 2014-03-26 17:35:11 ERROR juju.cmd supercommand.go:300 failed to GET object provider-state from container juju-cece0b9817a68cba4780784bf0663e45 [17:39] <_sEBAs_> caused by: Authentication response not received in 1m0s. [17:39] <_sEBAs_> thats wired [17:40] <_sEBAs_> i think i have all the info for authentication right [17:55] <_sEBAs_> i'm trying to introduce private cloud with juju here in a big company at Brazil, but I'm more than a day now stuck in this problem [17:56] <_sEBAs_> jose, jcastro or marcoceppi some light, please! :) [17:58] _sEBAs_: maybe sudo was expecting a password? [17:59] it actually shouldn't require sudo, but I'm wondering why it's not able to create the container [18:00] maybe others may be able to guide you better, I've personally not used openstack :) [18:00] _sEBAs_: run juju bootstrap with --debug [18:01] hi guys. Is it mandatory to have a Swift infrastructure to une Juju ? [18:02] <_sEBAs_> jose: maybe, but it isn't asking for one [18:02] bada: if you're doing openstack, yes [18:02] jose _sEBAs_ you don't need sudo for any prodvider other than local [18:02] <_sEBAs_> marcoceppi: im running bootstrap with --debug [18:02] <_sEBAs_> greate! [18:02] <_sEBAs_> i was expecting something like that [18:03] marcoceppi : OK and it should use x509 certificat ? Because i have an error x509: cannot validate certificate for xxxxxxx because it doesn't contain any IP SANs [18:03] bada: so juju creates a x509 cert, iirc, for secure communication with the bootstrap server [18:03] bada: when are you getting this error [18:03] marcoceppi : when i run the bootstrap command [18:04] marcoceppi : i'm not the only one https://bugs.launchpad.net/juju-core/+bug/1261780 [18:04] <_mup_> Bug #1261780: go 1.1.2 TLS-enabled client does not accept our CACert

[18:04] yep [18:04] :) [18:05] note that we've removed CACert from the certificates packages [18:05] if you want to trust CAcert, you'll have to install the certificate manually === roadmr_afk is now known as roadmr [18:06] sanrnold : how and where ? [18:08] bada: even though this is about intermediate certs I suspect it would work for top-level certs as well: http://yob.id.au/2013/02/06/trusting-new-ssl-certificates-in-debian.html [18:09] sarnold: Ok i'll try this. Thanks [18:10] <_sEBAs_> marcoceppi: can i show you the debug log of the bootstrap ? [18:11] _sEBAs_: yeah, just make sure to scan it for any private information first [18:21] make [18:23] <_sEBAs_> marcoceppi: http://pastebin.com/UFmBgAG7 [18:23] <_sEBAs_> marcoceppi: thanks in advance ;) === vladk is now known as vladk|offline [18:37] allomov, greetings :-) [18:37] Hello %) === allomov is now known as allomo === allomo is now known as allomov === allomov is now known as code1n [18:49] <_sEBAs_> marcoceppi: any ideias? maybe my openstack credentials? === CyberJacob|Away is now known as CyberJacob [19:25] i deployed my service (forgot to run `juju debug-log`) and my service is now in error state. wondering about the best path forward to determine what went wrong. [19:26] destroy the service and rerun with debug-log? [19:28] tvansteenburgh: juju ssh [unit], logs are stored in /var/log/juju [19:28] jose: thanks! [19:28] sure :) [19:30] Q about virsh and maas [19:30] http://askubuntu.com/questions/292061/how-to-configure-maas-to-be-able-to-boot-virtual-machines/295976#295976 says to use the url: qemu:///system [19:30] <_mup_> Bug #295976: LTSP Fat client unable to mount network shares in fstab [19:30] is /system the real system name [19:30] or literally ///system [19:33] I expect that 'qemu://' is the scheme and '/system' is the path [19:34] just like file:///etc/blah/blah [19:34] resolves to /etc/blah/blah [19:35] Ok so real machine dns name [19:35] cool === timrc is now known as timrc-afk [19:37] I think /system is a path on a default name (probably localhost) [19:37] otherwise qemu://dns_name_of_machine.com/system [19:37] hum ok [19:37] trying to make power on work for a local guy [19:38] think thats whats blockin me [19:38] Well, I'm at my limit; nothing more to contribute :) [19:38] sorry [19:39] haha thanks anyway i have some more ideas to guess and check [19:39] good luch [19:39] luck [19:40] i have resorted to naming all my juju machines prefixed with turd_ [20:01] <_sEBAs_> here is someone using juju with openstack? and working? [20:04] <_sEBAs_> I tried everything and i'm with two errors, "GET object provider-state from container juju-..." is given and apache time out, and the second one "failed to write bootstrap-verify file: cannot make Swift control container" [20:22] openstack is complicated [20:26] <_sEBAs_> Fishy_: tell me about it [20:27] <_sEBAs_> took me like 3 days to install and configurate with lxc hypervisor ¬¬ [20:30] been trying for 3 days to get juju status to return [20:30] for maas [20:30] still failure [20:30] no useful logs [20:32] <_sEBAs_> Fishy_: i feel you bro [20:33] <_sEBAs_> im stuck with an error trying to connect openstack to juju [20:33] <_sEBAs_> ops juju to openstack [20:33] we need a juju emotional support group [20:44] <_sEBAs_> hehe [20:49] sounds like you need an openstack emotional support group [20:49] ;) [20:54] OT, but does anyone know why a maas configurd bind9 dns proxy would fail 66% of the time, but succeed 33% of the time.. on a dns request to a random website like google.com ? [20:54] i set a forwarder to a real working dns server [20:56] Fishy_: that's, fishy.. ;) what does the logs in /var/log/maas/ show? [20:57] logs there are: celery.log celery-region.log maas.log maas.log.1.gz oops pserv.log pserv.log.1.gz rsyslog txlongpoll.log txlongpoll.log.1.gz [20:57] Fishy_: huh, none of those are helpful [20:57] Fishy_: is there a dns-y log in /var/log ? [20:58] nope [20:59] oo i think I see it in syslog [20:59] Mar 26 15:57:47 cgta-dev-6 named[27714]: validating @0x7fa468022410: com SOA: got insecure response; parent indicates it should be secure Mar 26 15:57:47 cgta-dev-6 named[27714]: error (no valid RRSIG) resolving 'google.com/DS/IN': 192.168.78.202#53 [20:59] Fishy_: that seems to be your issue [20:59] dnssec-enable no; and dnssec-validation no; worth a try! [21:01] marcoceppi: fixed it yay thanks [21:02] <_sEBAs_> \o/ [21:03] <_sEBAs_> marcoceppi: i sent an email to explain my issue with juju and openstack, any help would be appreciated :) [21:04] _sEBAs_: thanks, a bit busy at the moment, will look a little later today. Sorry! [21:04] <_sEBAs_> to the juju group :) [21:04] _sEBAs_: perfect [21:05] <_sEBAs_> marcoceppi: I imaging that!! sorry to bug you, and thanks! === timrc-afk is now known as timrc === roadmr is now known as roadmr_afk [21:34] So I deployed a charm to hp-cloud that provides a web page. Without exposing the charm I am able to view the web page. Can someone explain why that is? According to juju status exposed: false and no open ports [21:40] mbruzek: no [21:40] mbruzek: I don't think so, the firewall needs to have ports opened [21:40] thumper I am checking the charm... I don't see firewall config, apache config yes, but no firewall configuration [21:41] mbruzek: juju handles the firewall by the charm opening ports [21:41] thumper, yeah but open-port is not called in the hooks that I can see [21:42] that may well be the problem [21:44] thumper, So I would not expect that I could have got to the 80 port on this system if there are no open-port commands. Is it possible to confiure a charm to be open without calling open-ports? [21:44] I'm not sure, but I don't think so [21:44] perhaps jcastro knows [21:45] or marcoceppi [21:47] thumper: does juju reuse sec groups? [21:47] that could explain it [21:47] * thumper shrugs [21:47] if a sec group was left behind machine-0-secgruop [21:47] whatever [21:47] etc [21:47] yadda [22:15] hey lazyPower, mind a PM? [22:15] jose: not atm, brb [22:15] ok! [22:46] Is there any way to get juju working on Debian currently? Here's what I get when I try to bootstrap with the local environment: http://sprunge.us/ZCMC [22:50] smarter: you /can/ but it requires a bit of work [22:51] smarter: and I don't think it'll work with local at all, only a "real" cloud [22:51] I tried earlier, and got pretty far, but had to stop because of time constraints [22:53] I see, thanks for the information [23:03] hey marcoceppi, still around? I heard back from negronjl [23:03] jose yo [23:03] who dares mention my name :) [23:03] :P [23:03] <3 negronjl [23:03] marcoceppi, the cf-* charms should be shot ... they are old stuff :) [23:03] negronjl: shot as in unpromulgated? [23:04] marcoceppi: yup [23:04] negronjl: ack, will remove from store [23:04] marcoceppi, thx man [23:04] I also mentioned the cloudfoundry ones [23:04] those too? === CyberJacob is now known as CyberJacob|Away [23:45] when I "accept an commision" a new node.. should MAAS use WOL to wake that node up and start the process? [23:46] right now i need to boot each node 3 times.. one to get registered.. once to get comissioned.. and a third time to actually do what he is suppsed to