[00:00] cowmix: hm [00:00] cowmix: its definitely *far* better than when that statement was written [00:00] cowmix: I think for certain use cases, its a good choice in production [00:00] cowmix: https://bugs.launchpad.net/juju/+bugs?field.tag=production [00:01] cowmix: those are known issues that should be considered before putting it there. Perhaps we should update the docs to point at that list. [00:11] .win 15 [00:40] SpamapS: thanks... I'm *very* excited about juju.. I'm trying out the 12.04 beta as soon as its out [00:47] cowmix: beta1 is out. :) [00:47] cowmix: I'd recommend using juju from the ppa though.. we should land a new version in beta2, and the PPA is much closer to that. [00:50] Awesome.. when 12.04 is fully baked.. use the release version then? [00:51] It's interesting because I have some friends that do HUGE EC2 deployments on Ubuntu and they didn't know about juju.. I turned them on to it and they seemed jazzed. [00:52] Canonical really needs to pump this up harder.. it solves an issue everyone is trying to solve on their own. [00:55] cowmix: a little over a year ago, juju didn't exist.. so.. give us some time. :) [00:56] cowmix: there's a charm contest going on right now.. you should get some of them to enter it. :) [01:10] SpamapS: is Canonical dog-fooding it now in their own infrastructure? [01:39] anyone got time for a really basic question [01:46] cowmix: here and there, yes.. nothing important, as we only run that on released LTS's, and lucid can't run juju [01:47] sloth_: just ask, and hang out for an answer, somebody will get back to you. Also we watch the juju tag on askubuntu.com a lot, so asking there might be a good idea [01:49] cool. so I'm totally new to juju (and not a Ubuntu guy normally). I'd like to use it provision a mongo cluster on amazon. I'm not sure if I have to create a ubuntu EC2 instance to use as the provisioning host (that wouldn't be part of the 3 machine replication set) or if I can do it from my mac or where to start [01:49] I looked at the introduction to juju but I suspect I'm missing a few pieces of knowledge on how to get to where it starts [02:02] <_mup_> juju/status-changes r484 committed by kapil.thangavelu@canonical.com [02:02] <_mup_> collapse relation display in units unless error, all relation names are multi-valued [02:04] i'm going to hit a devops meetup tomorrow, we're hitting a few conferences this year [02:21] <_mup_> juju/status-changes r485 committed by kapil.thangavelu@canonical.com [02:21] <_mup_> status uniques when the same service has multiple relations to the same endpoint [02:42] <_mup_> Bug #959884 was filed: Improved status output < https://launchpad.net/bugs/959884 > === bradm_ is now known as bradm [03:03] <_mup_> juju/force-upgrade r465 committed by kapil.thangavelu@canonical.com [03:03] <_mup_> finish upgrade-charm --force cli tests [03:04] <_mup_> juju/force-upgrade r466 committed by kapil.thangavelu@canonical.com [03:04] <_mup_> add to ignores [03:07] are pkg task sets addressable, ie. can i install the cloud/server task set into a container, or is it more of a preseed thing [03:08] ah.. tasksel [03:11] hmm.. doesn't want to work without a ttyl [03:59] <_mup_> juju/series-from-charm r464 committed by kapil.thangavelu@canonical.com [03:59] <_mup_> charm series is used as service system constraint [04:19] <_mup_> Bug #959914 was filed: Charm series is used for service series constraint. < https://launchpad.net/bugs/959914 > [06:35] SpamapS: Uhh lots of suggestions and I dont know where to begin :) I guess I will bug jcastro for guidance tomorrow :P I must be off to bed now [06:39] bkerensa: Sorry.. I am really excited about your charm which is why I'm being so picky about it. I will likely be a user. :) [06:40] bkerensa: bug me for help too, I'm happy to assist. :) === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away === jelmer_ is now known as jelmer === almaisan-away is now known as al-maisan === amithkk is now known as ubuntu === ubuntu is now known as marcoceppi_ === jml` is now known as jml [11:29] whats the state of juju on lucid? [11:30] might be a funny question but I'd like to see whether I can get apache bigtop running with the hadoop charm === d0od_ is now known as d0od === marcoceppi_ is now known as amithkk === amithkk is now known as ubuntu === ubuntu is now known as amithkk === fenris is now known as Guest87935 [13:07] jamespage, it hasn't been tried in some time, the libzk libs need a more recent version to avoid some bugs there afaicr. we may have picked up some py 2.7isms [13:08] hazmat, ack [13:08] it was a 'if i get time' type activity TBH === al-maisan is now known as almaisan-away [13:53] hi all, i'm trying to use juju in openstack, but i receive an error in the machine every time i try to create one, any one knows about this problem? === almaisan-away is now known as al-maisan [14:07] hi, anyone can help with juju and openstack? [14:18] yolanda, hi, what's the error? [14:18] hi, james_w, just my openstack instance is created, but the status is "error" [14:20] yolanda, hmm, I'm not sure how to find out what caused the error [14:21] yolanda, could be a number of things [14:21] james_w, i see that the image used is ami-00000049, is that ok? [14:21] it might be best to take a look at the instance directly in openstack to see if you can get more info [14:22] jamespage, how can it be done? i only can see the instance when i do an euca-describe-instances, and see that status [14:22] yolanda, does it show error there as well? [14:22] no [14:22] yolanda, hmm [14:23] euca-get-console-output might show you a bit more [14:23] mm, let me see [14:23] euca-get-console-output i-00001b40 [14:23] UnknownError: An unknown error has occurred. Please try your request again. [14:24] yolanda, oops [14:24] not a very clear error :) [14:24] yolanda, thats a openstack is broken error [14:25] jamespage, but i'm running some other instances in openstack right now === carif_ is now known as carif [14:25] i'll try to create that one manually [14:26] yolanda, you might be able to ssh to it - please try that [14:26] might be able to get some good debug out of it yet... [14:26] jamespage, cannot ssh, i can't see any internal ip [14:26] yolanda, what is its status? running? [14:26] i have 3 instances, i can see the internal ip for the previous two, but this doesn't show any ip [14:26] status is error [14:27] the privateIpAddress field is empty [14:27] yolanda, so basically it compelelty failed to start [14:28] yes, it seems [14:28] will try to create same image manually [14:30] ok, same image created with euca-run-instances work [14:30] so it's something with juju process [14:32] btw, what i want is to test some juju charm, if i cannot do it with openstack, what can i use? i did some tests with a personal EC2 account, but i need a large machine, and they started charging me an important bill, so i stopped it [14:33] yolanda, I think I know what it might be - lemme just check [14:33] ok, jamespage, thanks [14:36] yolanda, can you try with ami-00000048 please [14:36] jamespage, that can be specified with the default-image-id param, right? [14:37] yolanda, it can [14:37] yolanda, have you tried using the local provider for juju? [14:38] jamespage, yes, but gives another error: error: Falló al iniciar la red default [14:38] error: Requested operation is not valid: network is already active [14:38] Command '['virsh', 'net-start', 'default']' returned non-zero exit status 1 [14:38] 2012-03-20 15:38:29,472 ERROR Command '['virsh', 'net-start', 'default']' returned non-zero exit status 1 [14:39] yolanda, ah - I know what this is [14:39] i saw that i had to logout/login, but i already did it [14:39] yolanda: what does virsh list-network give you [14:40] unknown command "list-network" [14:40] sorry net-list [14:40] not list-network [14:41] yolanda: did you logout/back in after installing juju and especially libvirt-bin? [14:41] yolanda, can you check that the account you are using is in the libvirtd groups as well please [14:41] SpamapS, yes [14:42] Nombre Estado Inicio automático [14:42] ----------------------------------------- [14:42] default activo si [14:42] sorry but it's spanish output [14:42] no problemo, yo hablo un poquito ;) [14:42] genial :) [14:44] jamespage, how can i check it? [14:44] yolanda, type 'groups' [14:45] groups [14:45] yolanda adm dialout cdrom plugdev lpadmin admin sambashare libvirtd [14:45] that? [14:45] yes - looks OK [14:45] 'libvirtd' being the critical bit... [14:45] hmm [14:46] I wonder [14:48] yolanda, I think thats a bug - juju parses the output of net-list and looks for 'active' [14:48] 'active' != 'activo' [14:48] damn [14:48] so how can i fix it? is it possible? [14:49] wait [14:49] please don't say: use Ubuntu in english :) [14:49] yolanda: dpkg -l juju [14:49] That bug was fixed... we set LANG=C before calling net-list [14:50] Deseado=Desconocido/Instalar/Eliminar/Purgar/Retener [14:50] | Estado=No/Instalado/Config-files/Desempaquetado/Medio-conf/Medio-inst/espera-disparo/pendiente-disparo [14:50] |/ Err?=(ninguno)/Requiere-reinst (Estado,Err: mayúsc.=malo) [14:50] ||/ Nombre Versión Descripción [14:50] +++-=================================-=================================-================================================================================== [14:50] ii juju 0.5+bzr398-0ubuntu1 next generation service orchestration system [14:50] SpamapS, I was just looking at that [14:51] yolanda, I would recommend using juju from the team PPA - that is a very old version [14:51] (the one that is in oneiric) [14:52] jamespage, how can i config it? in which repo is? [14:52] yolanda, https://launchpad.net/~juju/+archive/pkgs [14:52] let me see [14:52] i wonder if that's the cause of all my problems [14:53] yolanda, not all of them but this one - probably [14:53] SpamapS, good catch [14:54] krondor, how's it coming along? [14:54] krondor, hey do you guys plan to nginx in the moodle charm? [14:54] let me test [14:59] i updated it, now i have this error: [14:59] Starting storage server... [14:59] could not connect before timeout [14:59] 2012-03-20 15:58:41,768 ERROR could not connect before timeout [15:01] SpamapS: gmb said he talked to you about packaging for python-shelltoolbox. will your packaging work for a lucid build? [15:01] We really should just push a newer juju into 11.10 and fore-go the proposed testing requirements. :-P [15:02] bac: it might, but you can't deploy charms on lucid at the moment, so thats sort of moot. ;) [15:03] jcastro: it's going well so far. We're going apache because the thought is most moodle users would need the auth modules. [15:03] yolanda, please try 'juju --verbose bootstrap' - you might need to juju destroy-environment first [15:03] SpamapS, +1 its next to useless how it is ATM [15:03] SpamapS: our buildbot slaves will be running lucid containers and they need to use the package too, that's why i ask. [15:04] jcastro, lookings at the jetty/spdy stuff ATM - have a few ideas [15:04] bac: I can just use python-support.. that will work back to lucid [15:04] jamesjpage, no timeout error now, seems i got it! [15:04] i'll check with status [15:04] SpamapS: that would be great [15:04] * SpamapS goes afk for a bit [15:06] more recent versions set the locale to C explicitly before calling out to libvirt [15:06] via env vars [15:06] jamespage, right so I was thinking, something simple, that would end up being "this is a fast way to mess with SPDY" and then tell people about it. As a technology showcase, nothing like serious or something people would actually use. [15:07] jcastro, how about jenkins on SPDY? I'm going to give it a whizz [15:08] jamespage, SpamapS, thanks a lot for the help [15:08] yolanda, np [15:08] I noticed something with hooks firing. My db-joined hook takes quite a while to execute. But as soon as I add-relation it shows as "state: up", if it eventually fails it shows the failure then. So, I know the relation hook fired but it won't wait for end of execution of the relation hook. Is this a bug or a intended usage? [15:10] jamespage, ok so that is sexy, yes, absolutely! [15:15] marcoceppi: that's a known issue [15:16] m_3: Cool, I'm fine with just adding relation, exposing, juju status until the port shows as open (thus the end of the script is done) [15:19] marcoceppi: yeah, we've had several discussions of this problem, but no solns afaik [15:33] jcastro: mayber I can let them pick ngninx or apache w/ a config directive. Was thinking the same with postgresql / mysql. === al-maisan is now known as almaisan-away [15:36] krondor, yep [15:38] <_mup_> juju/trunk r486 committed by kapil.thangavelu@canonical.com [15:38] <_mup_> merge setuppy-fixes from paul [a=][r=hazmat,clint] [15:46] m_3: I wish I knew more about how juju launched hooks :\ one day I'm going to start digging in to the source code [15:47] marcoceppi: open-port is actually supposed to be the signal that your service is ready. [15:48] SpamapS: yeah, that's what I've been using, it's just a wee bit confusing to see the relation state as up when the hook is still running, because then it goes from up to error half way though [15:48] Merely commenting on how it'd be nice if it went from state: "adding", to "up" or "error" [15:53] I am running into ssh host key errors after I destroy my environment and try to bootstrap a new one [15:53] marcoceppi: yeah, I think a more appropriate word would be "active" [15:53] ninjix: thats common [15:53] ninjix: there's an open bug to have juju ssh use a generated known_hosts file to avoid that.. [15:53] lmorchard, hey I heard you were having some juju problems, anything I can do to help? [15:53] how do you guys run your ssh_config? [15:54] did you set StrictHostKeyChecking off? [15:54] ninjix: for production, I just deal with the pain. For testing I wildcard the test hosts and do StrictHostKeyChecking no [15:54] ;) [15:55] the way I do that, which might be a bit evil, is by wildcarding by amazon region. All my production instances are in us-west-1 .. I test in us-east-1 [15:55] ok, cool. just wanted to make sure I hadn't missed something basic [15:56] SpamapS: btw, thanks for your help over the last few days. I've now got juju and cobbler driving our Proxmox cluster [15:57] dev team is going to very pleased when I unveil this new cloud environment [15:58] this is quite cool [15:59] ninjix, when you're not so slammed it'd be nice to send this to the list. [15:59] yeah, I'd love to see a writeup of that [15:59] sure [15:59] ninjix: *NICE* [16:00] http://charms.kapilt.com/~patrick-hetu/oneiric/openerp-server [16:00] look at this readme [16:00] it's already better than all the stuff we wrote in the intial set of charms. :) [16:00] patrick and jseutter are killing it. :) [16:01] jcastro: I'm so ashamed of mediawiki now.. I almost want to just nuke it and re-do the whole thing. ;) [16:01] huh? [16:01] jseutter, it's exciting to see you guys submitting these charms [16:01] jcastro: ah :) [16:02] shazzner's working on gitolite too [16:03] my service units keep appearing with .localdomain why are they not using the domain delivered by DHCP? [16:08] * SpamapS digs through the code to see where that value comes from [16:09] * SpamapS goes a 4th level of indirection deeper and wonders what he has wandered into [16:12] UGH [16:12] output = subprocess.check_output(["hostname", "-f"]) [16:12] ninjix: ^^ [16:12] that is just so wrong.. :-P [16:18] hmm... ok I'll work with the -f arg === bcsaller1 is now known as bcsaller [16:38] just had 99% successful wordpress test run. Only thing that didn't work was that the 000-default site was left enabled [16:38] is that normal? I'm running today's PPA [16:39] ninjix: its more complicated than that [16:40] ninjix: basically the wordpress charm expects you to use the 'public-address' to access it.. [16:40] ninjix: I believe marcoceppi is working on fixing that [16:40] ninjix: glad to hear you got it deploying though! :) [16:40] marcoceppi: were you going to try and tackle the Host: header weirdness w/ wordpress? [16:40] the public address started working once I a2dissite 000-default [16:41] ninjix: oh, hm, that sounds different [16:41] jcastro, that was painful - looks like upstream managed to not release spdy as part of the lastest jetty distro [16:41] I had to hack it in [16:42] nice [16:50] SpamapS: this seems to be Apache2 strangeness. The FQDN site looks properly configured but Apache stops using it as soon as I re-enable the default [16:51] ahh... this is happening because the 000-default is using the same FQDN as the juju created site [16:51] ninjix: sounds like a bug in the charm really.. should just dissite the default [16:52] ninjix: https://launchpad.net/charms/+source/wordpress/+filebug [16:52] If you wouldn't mind. :) [16:52] no problem [17:00] SpamapS: That's a different issue with using HAProxy in front of WP [17:09] marcoceppi: I just appended the a2dissite to my local oneiric/wordpress/hooks/install [17:09] marcoceppi: will the next juju add-unit wordpress pick up the change? [17:11] ninjix: no [17:12] ninjix: you have to use upgrade-charm [17:13] does that imply that the my current env has the wordpress charm cached on the bootstrap instance? [17:13] ninjix: yes, there's a webdav server that hosts the charm bundles [17:13] :) [17:14] actually, SpamapS I had a question about add-unit. When I add-unit it executes the charm without any real idea that it's an additional unit, correct? So it would be just as if I had done a deploy + add-relation + expose + whatever else I did to the previous unit [17:14] marcoceppi: right [17:14] marcoceppi: to make units aware of one another use peer relations [17:14] figured, follow up question [17:14] and answered. [17:15] marcoceppi: though that one will proceed all the way through install -> started before being aware of the peer relationships [17:16] marcoceppi: which has been a problem in the past for charms that need to work slightly different between 1 and 1+ nodes ;) [17:16] Lastly, is there a way within a non-peer relation hook to tell if there is another unit? [17:16] I'm trying to avoid having the db-relation-joined hook run a MySQL import when it already exists from the first instance [17:17] marcoceppi: I'd use the database itself for that... not rely on juju for it [17:17] I guess just mysqladmin create and capture a non 0 exit or something? [17:17] marcoceppi: there's a few different ways to do that... right... what SpamapS just said... check the db for a schema version or some content [17:26] <_mup_> juju/subordinate r522 committed by kapil.thangavelu@canonical.com [17:26] <_mup_> update unit deploy signature [17:31] I am getting an SSH key error when I execute upgrade-charm [17:32] ssh works manually to the remote_host value [17:57] jamespage: ping [17:58] negronjl, pong! [17:58] ninjix: all commands have to ssh to machine 0 so they can talk to zookeeper securely.. [17:58] jamespage: Here is what I have so far for a combined tomcat6 and tomcat7 charm ( lp:~negronjl/+junk/tomcat ) [17:58] jamespage: if/when you get a chance, I would appreciate some feedback. [17:59] negronjl: +junk?! ;) [17:59] negronjl, yep - I'll take a look tomorrow [17:59] SpamapS: It's a work in progress and I don't want to confuse anyone that may think that this is an actual usable charm yet. [17:59] jamespage: thx [18:00] SpamapS: I'll put it in the proper place and put it through the proper process once I get done with it. [18:00] negronjl: roger. :) [18:25] <_mup_> juju/relation-id r488 committed by jim.baker@canonical.com [18:25] <_mup_> Pass through relation id [19:12] SpamapS: how does one add a second mysql unit as a slave? [19:13] I noticed the mediawiki charm utilizes mysql slaves === hspencer is now known as hspencer[afk] [19:51] ninjix: currently the mysql charm only supports one-way replication, so you 'juju deploy mysql slave-service' and then 'juju add-relation master-service slave-service' [20:17] starting get the hang this. So it is the relation call that determines which hooks fire [20:17] ninjix: right [20:19] so a master-master creating hooks that understand how to handle 'juju add-relation master-1 master-2' and then a 'juju add-relation master-2 master-1' [20:20] that's going to dial up the complexity of the charm [20:22] maybe I can add a hook for mysql-mmm pkg [20:23] ninjix: you'd only have one 'add-relation' [20:23] ninjix: they're bi-directional [20:23] ninjix: you might be able to achieve it just by doing that now though.. hmm.. hadn't thought of trying that [20:24] how would it discriminate between between simple slave and master-master with only one add-relation === hspencer[afk] is now known as hspencer [20:31] ahh... answer my own question. Just learned how to use the : syntax :) [20:48] very cool. I'm now scaling the database service with total ease. [20:49] ninjix: I think thats our next t-shirt ;) [20:49] jcastro: ^^ [20:49] really liking the low barrier learning curve too [20:49] man nice dude [20:50] ninjix, send me a mail, we should send you a juju shirt, jorge@ubuntu.com [20:51] me like schwag [21:37] hello [21:38] quick question, what could be up when after deploying a charm the newly created machine still has a 'not-started' state [21:38] and I can't ssh into it yet [21:38] just need to be patient? [21:52] well I destroyed it, then recreated it [21:52] now it works, weird [22:09] jcastro: when does this contest end? I have a busy week and probably cant do work till this weekend on the charm [22:09] bkerensa: the 23rd [22:10] bkerensa, it's ok we can be flexible [22:10] the deadline is the 23'rd [22:10] but then there's a week of reviews, etc. [22:10] cool [22:12] this is the first one so it'd be dumb to make it "pencils down everyone!" [22:30] <_mup_> juju/relation-id r489 committed by jim.baker@canonical.com [22:30] <_mup_> Completed relation id refactoring [23:13] I think if you already submitted the charm.. [23:13] and you are just dealing with my nit-picks.. [23:13] your entry will be considered positively ;) [23:13] bkerensa: ^^ [23:14] kk