[00:42] SpamapS: around? [00:42] adam_g: yeah, whats up? [00:44] SpamapS: nvm.. was about to start with "am i losing my mind or...." turns out i am :) [00:45] adam_g: welcome to the club! [00:45] adam_g: wait till you attend one of our parties [00:47] * hazmat signs out [03:24] * nijaba is back after an unwanted rm -rf which included a mount --bind /usr/bin [03:25] SpamapS: I'm finishing sc_peer_copy (rsync/scp) [03:25] SpamapS: the hardest part is the test function [03:25] SpamapS: should be almost there [03:25] hazmat: thanks :) [04:34] hello [04:34] does anyone know how to write juju charms? [04:34] I'm working on one and had some questions [04:35] try to make a charm of gitolite: http://sitaramc.github.com/gitolite/ [05:19] shazzner_: it's pretty late for most of the team, so you might be better off asking tomorrow or posting to the list juju@lists.ubuntu.com [06:11] shazzner_: agreed with robbiew, I'm about to head to bed myself. Note that the instructions on that first page are basically your 'install' hook, and the "mirroring" instructions look like they might be good for peer relations (so you can scale out your gitolite servers on demand) [07:37] thanks guys, i'll try again tomorrow [08:42] mornin' [09:01] heya rog [09:18] hi all [09:20] mpl: hiya [09:21] hi rog [09:30] rog: btw, did you guys get man answer to gustavo (about the zk/ssh thing) a couple or so days ago? I thought I sent it, but I'm a bit distracted these days. and since I haven't gotten any reply... [09:30] s/man answer/my answer/ [09:31] mpl: yes, we did thanks [09:31] mpl: well, i did anyway [09:31] mpl: i hadn't noticed ssh.Client.Dial... [09:31] rog: ok, and did I seem to have understood correctly gustavo's plan? [09:32] mpl: i think so. [09:32] rog: I hadn't noticed it either, Dave had to point it out to me :) [09:32] imho it's not obvious enough in the documentation that this is the way to achieve ssh port forwarding. [09:33] one way at least. [10:46] hello there [10:47] heyho: hi [10:48] i've a problem with juju bootstrap can you give me a hand? [10:48] don`t ask just ask :) [10:49] ok, i'm getting the error: dict object has no attribute read [10:50] heyho: you will have to tell me your environment and yaml config along with error. please pastebin it [10:52] heyho, I've heard a couple of reports of something similar [10:52] heyho, you're running against orchestra, right? [10:52] yeah.. the problem is also that I'm using juju in a virtual machine and I can't copypaste it.. [10:52] heyho, apt-get install pastebinit :) [10:53] heyho, but I'd be most interested to know what OS and cobbler version you're running on the orchestra VM [10:54] i'm using ubuntu 11.10 oneiric server x64 [10:56] my environment.yaml is this: http://pastebin.com/SWiNejBc [10:57] heyho, the environents file looks fine [10:57] heyho, sudo cobbler --version [10:57] heyho, (no idea why it wants superuser powers to get the version) [10:58] i also think i'm using the latest version of orchestra (2.26) and cobbler (2.1.0) [10:58] is there any limitations if i'm running orchestra as a VM? [10:59] heyho, orchestra should be just fine in a VM [11:01] ok..I also configured cobbler and added new systemz but, while in my previous attempts I received an ssh error by juju status but the bootstrapped worked, recently i reinstalled completely orchestra and juju and now i get the error on juju bootstrap [11:03] heyho, it doesn't look like that issue's been captured in a bug yet [11:04] heyho, am I correct in thinking that the only difference you're aware of between that, and your previous setup, is the orchestra version? [11:04] I managed to get the full error I receive on bootstrap http://pastebin.com/JcF4989s [11:05] I think so [11:06] heyho, thanks very much for the error log; I'm quite sure I've seen someone else reporting that here, I'll make a bug and try to get onto it as soon as I can [11:06] yes [11:06] I saw an irc log with my same error [11:07] heyho, would the local provider be an OK workaround for you in the meantime? [11:08] I didnt understand your question.. [11:10] is maybe the problem related in the interaction beetween juju and the cobbler ks_meta? [11:10] heyho, I mean that -- since orchestra doesn't work and I can't immediately supply a fix -- it may be possible for you to experiment with juju using the local LXC provider instead of orchestra in a VM [11:11] heyho, it certainly is [11:11] heyho, but tracking down the precise cause and fixing it may take a little time [11:11] heyho, and so I'm keen to find you some way to be able to work with juju despite this bug [11:12] heyho, hence the local provider suggestion [11:12] I imagine and I thank you for your alternative, but I'm not aware how to use Juju as you suggested me.. What's the local LXC provider [11:14] heyho, let me try to find you a writeup -- basically it runs the "machines" as LXC containers, which are a bit like chroots [11:15] heyho, https://juju.ubuntu.com/docs/provider-configuration-local.html [11:15] Good morning all [11:16] heya niemeyer [11:17] heyho, btw, where did you get juju from? [11:17] from juju pkgs [11:18] heyho, thanks [11:18] I'm reading your link right now [11:20] heyho: you can also see http://askubuntu.com/questions/65359/how-do-i-configure-juju-for-local-usage :D [11:20] hello niemeyer [11:21] that's even better thanks [11:23] do i have to install juju on a physical machine I presume..but do I add pkgs repository or no? [11:24] hello to niemeyer [11:30] heyho: you can use the default pkg from oniric repo and it works too :) [11:34] <_mup_> Bug #905286 was filed: orchestra: dict has no attribute read < https://launchpad.net/bugs/905286 > [11:42] I'm getting this error at local juju bootstrap http://pastebin.com/b3JxLRKu [11:42] I previously rebooted networking service..do I need to reboot the entire system? [11:43] SpamapS: your comments for ch_peer stuff should have been addressed. Let me know [11:49] heyho, I'm sorry, I don't recognise that error -- probably best to just try a reboot, according to koolhead11's link, anyway :) [11:56] heyho_: You may also need to set net.ipv4.ip_forward=1 in you /etc/sysctl.conf [11:56] and then reboot the system [12:06] hello, I rebooted twice (first time I removed my bridge in interfaces) but I'm getting the same "network is already active" error . I also configured ip4 forward [12:13] heyho: am running juju on my physical machine running oneiric. I had same error you mentioned and i followed exact steps as given in askubuntu and it worked for me [12:15] niemeyer, re constraints spec... I'm becoming concerned about the overlap between ec2-instance-type and arch/cpu/mem [12:16] fwereade: It crossed my mind too [12:16] ok I'll retry again [12:16] fwereade: Are you finding any practical issues? [12:17] niemeyer, so far, only in thought experiments [12:17] fwereade: Ok, anything in specific that you're concerned? [12:18] niemeyer, well, consider an environment in which you specify cpu/arch/mem, and then try to set service constraints to an ex2-instance-type which is below those specs [12:19] fwereade: Ok? [12:19] niemeyer, according to the spec, the harshest constraints win, which will be the env-level "mem=16G" rather than the service-level "ec2-instance-type=t1.micro", for example [12:19] niemeyer, I feel this rather violates the principle of least surprise [12:20] fwereade: That's not what I had in mind when we talked about that [12:21] niemeyer, ah, sorry, I must have misinterpreted you :( [12:21] fwereade: What I had in mind is "the most specific constraint wins", which would highlight ec2-instance-type in this case [12:21] niemeyer, most specific in terms of scope? [12:21] fwereade: most specific in terms of the provider [12:21] fwereade: If we know the instance type, everything else is more vague [12:22] niemeyer, so I can't override an env-level m1.small with a service-level mem=16G? [12:22] niemeyer, I suppose it's reasonable [12:22] fwereade: The interaction between several levels is a good question [12:23] niemeyer, I suppose it's reasonable to say that once you've specified m1.small you need to override with m2.xlarge [12:23] fwereade: Maybe we should take the most specific on each level [12:25] niemeyer, ok, I think that could work [12:26] niemeyer, looks like lunchtime, I'll ponder a bit longer :) [12:26] fwereade: So in your example, mem on the service would win [12:38] niemeyer, if the user expresses that they want service machines to have 16G, I feel we should prbably give them service machiens with 16G regardless of underlying env constraints [12:42] I'm still getting the same error after 3 retries. I'm taking a break now..after that I'll try on another oneiric machine (this one had libvirt and virtual bridges already installed for virt-manager..I removed them but maybe there's still some configurations left that are causing the problem) [12:45] fwereade, for provider specific vocabularies, its not going to be clear what the default is in many cases [12:46] the juju default is, since its a detail of the provider [12:47] fwereade: Right [12:48] hazmat, the spec has defaults for all the constraints, what's not clear? [12:48] hazmat, ok, the defaults are often "unset" [12:49] fwereade, for provider specific vocabularies its not very clear what a reset to default implies, what's the default ec2-region, what's the default rack, what's the default cpu-firmware [12:49] hazmat, I imagined the default to usually be "don't filter on this at all" [12:49] hazmat, but the default ec2-region is us-east-1 [12:49] yup, but unset is can still a imply default value for the provider, but its internal to the provider [12:50] and not per se documented or discoverable [12:50] not much to be done about it i guess, i'm just curious/concerned about discoverability of these default values [12:52] hazmat, it seems to me that, tautologically, people who don't care about cpu firmware or rack placement... don't care about cpu or rack placement, and so they can get what they're given and like it ;) [12:52] hazmat, and indeed, as you say, maybe not much can be done anyway [13:34] Hello again. So I managed to bootstrap juju successfully (I had to disable previous installed virtual bridges); how can now deploy charms (ex: mysql) ? I need a local repository? I used charms getall and created a directory named jujulocal but i get this error "ERROR Charm 'local:oneiric/mysql' not found in repository /home/stefano/jujulocal" [13:35] heyho, is there a charm at /home/stefano/jujulocal/oneiric/mysql? [13:36] heyho, if they're just flat inside jujulocal, it won;t find them [13:38] No, It would be in /home/stefano/jujulocal/mysql without oneiric dir.. [13:38] do I need to change that in environment.yaml and bootstrap again? [13:48] heyho, (sorry, I'm semi-lunching) repository structure is: repo contains a dir for each series; each series dir contains charms for that series [13:48] heyho, so if you move everything from jujulocal to jujulocal/oneiric, you should be fine [13:50] that's what I just did. And I also changed data-dir in environments adding /oneiric [13:54] Now i successfully deplyed mysql and wordpress and I get this with juju status http://pastebin.com/zTCRXz84 Why are them in pending state (and not started)? === TeTeT_ is now known as TeTeT === lborda is now known as lborda_afk [14:10] heyho, if its the local provider, it has to build the container via debootsrap [14:10] when you first deploy [14:10] it takes a few minutes depending on the speed of your hard disk and net connection [14:10] its a one time operation, the packages and a template container are cached on disk [14:10] so subsequent deploys or add-unit are fast [14:14] so I deployes 3 services but they are still pending just because they need time, right? [14:15] how does juju assign a public ip to the services? [14:33] heyho, you should be able to see the activity of the container creation with lxc-ls which shows extant containers and runing containers [14:34] heyho, all containers/services deployed with local provider are only accessible from the host machine [14:35] heyho, effectively their services under local provider are behind a NAT, the only public ip is the hosts [14:35] heyho, of course other providers don't have this issue, but local provider in particular was intended for development and prototyping [14:44] hazmat, I think something went wrong because the report of my juju status http://pastebin.com/fUSyUL1Z is missing services' relations and ip assigned to them [14:47] heyho: it looks like they're still installing... takes a _while_ for the first local install [14:48] well.. its been an hr.. [14:48] so not that long [14:48] wow... ok [14:48] exactly.. [14:48] heyho, there's some logs in the $data-dir variable you specified in environments.yaml [14:49] that would hold additional error information [14:50] it contains only charms directories [14:51] maybe i should destroy environment and create a new one? [14:51] I'd recommend a cleanup ( destroy-environment, wipe lxc cache /var/cache/lxc ) [14:52] maybe deploy a single service next time around... once that's 'started' then you know the lxc template has been built out [14:52] ok I'm doing this [14:52] from there you should be able to bring things up quickly... or debug that single lxc instance [14:55] I get this error with destroy-environment "INFO Destroying unit containers... lxc-wait: bind : Address already in use" [14:55] you can watch the lxc images get built out on the filesystem at /var/lib/lxc I think [14:56] hmmm... ok, then something more complicated is going on here [14:56] maybe I should remove the images in /var/lib/lxc? [14:57] Do you have a stock libvirtbin install? i.e., with dnsmasq serving up the 'default' network for virsh? [14:57] when my local env is destroyed, /var/lib/lxc is empty [14:58] I had virt-manager installed but i removed it and installed libvirtbin again.. ok, i'm removing images in that directory [14:59] the local provider docs and inststructions definitely assume you haven't tweaked your libvirt setup... I'm a tweaker and've been hit by that before :) [14:59] cool... with a purge and re-install of libvirtbin you should be golden [15:00] easy enough to check that `ps auwx| grep dnsmasq` is running and `ip addr show` shows virbr0 correctly [15:03] ok I purged libvirt-bin.. dnsmasq is runnig and ip addr shows vbr-default-0 [15:03] vbr-default-0 looks strange [15:03] yep [15:03] virsh net-list --all [15:05] it's something like name:default state:active autostart: yes [15:06] /etc/libvirt/qemu/networks/default.xml contains the name of the bridge [15:07] yours might've been added by something else, hence the strange name [15:07] and "ip addr show" contains, besides lo and eth0, these two things: http://pastebin.com/9AsXFqVj [15:07] i should remove them? [15:12] heyho: I'd `/etc/init.d/libvirt-bin stop` and then remove any bridge interfaces that're there [15:13] heyho: then check the contents of /etc/libvirt/qemu/networks/default.xml to look like stock (http://paste.ubuntu.com/772295/) and restart libvirt-bin [15:14] i removed them with "virsh net-destroy default" and after that I tried to do "service libvirtd restart" with response libvirtd: unrecognized service! [15:15] don't remember the name of the service... looking [15:16] libvirt-bin [15:16] you'll need a default net for libvirt though [15:16] (BTW, I'd recommend EC2 over local provider at this point if that's an option) === lborda_afk is now known as lborda [15:25] so.. removing /var/cache/lxc isn't generally useful its just the debootstrap cache [15:25] the error about lxc-wait bind, is because there is already an lxc command currently running on the system [15:25] that device vbr-defat-0-nic is odd [15:26] ok I think i removed them [15:26] even the vbr-default-0 is odd too [15:26] now ifconfig looks clean [15:26] i reinstalled libvirt-bin [15:26] heyho, does the output of `groups` contain libvirtd for you? [15:26] i still have installed libvirt-0.. do you know what's that? [15:27] yes it contains it [15:27] libvirt0's ok to be there [15:27] its the libvirt runtime library files [15:28] ok so i basically have eth0 and lo.. but not virbr0 comparing in ifconfig.. is that ok? [15:28] yup [15:29] heyho, there was no files in your data-dir? [15:32] actually there were, now i deleted them.. In environments.yaml I should write the data-dir path excluding the distribution subfolder right? [15:32] like /home/stefano/jujulocal and not /home/stefano/jujulocal/oneiric [15:36] i tried another juju bootstrap resulting in this error http://pastebin.com/jzd6xz1i [15:37] <_mup_> Bug #905366 was filed: local provider needs a dedicated network < https://launchpad.net/bugs/905366 > [15:38] m_3, re last bug, how many bridges can you attach to a physical device [15:38] i'm trying to reboot the machine again, brb [15:39] heyho: grab the default.xml from http://paste.ubuntu.com/772295/ and then something like `virsh net-define default.xml` [15:39] i.e. the default net's gotta be created [15:39] hazmat: unknown [15:40] hazmat: it's actually not attaching to a physical device though, right? [15:41] there are bridges that do that, but not the virbr's we're using [15:41] hmm.. [15:42] yeah.. i thought there was a reason it was problematic .. but its not clear why.. if we're not attaching it [15:44] I've had 4 virtual bridges working at the same time w/o any problems (other than iptables spaghetti) [15:44] in addition to a physical-bridge [15:46] (sad attempt at euca HA) [15:49] ok, I succeded in juju bootstrap [15:50] now i need to getcharms and then deploy one of them right? [15:50] yes, get a charm and deploy a single service to to kick off the lxc template build [15:51] that'll be 'pending' for a few minutes while it installs [15:52] ok..and i also need to create a oneiric subfolder in my localdir right? [15:52] 'charm get mysql' is faster than 'charm getall' [15:53] yeah, so I 'mkdir -p ~/charms/oneiric' [15:53] 'cd ~/charms/oneiric && charm get mysql' [15:53] then your deploy lines will need a repo like 'juju deploy --repository ~/charms local:mysql mydb' [15:55] i.e., charms don't need to live in the local provider's 'data-dir' [15:59] well, that's a very effective explanation thanks [16:00] m_3: let's do a quick charm update later today even though we've been gone [16:03] jcastro: cool man.. we've got a colorado-canonical-crew holiday lunch, so maybe a mobile hang-out on the drive up? [16:03] that would be perfect [16:04] I'm thinking it's 5 minutes, tops [16:13] * SpamapS is having breakfast right now with the los angeles canonical crew... [16:15] * m_3 proscribes thorazine for that [16:16] whoops... s/proscribes/prescribes/ :) [16:20] * SpamapS proscribes rhymes [16:30] forgot your gang-sign [16:31] guys thank you very much for your support! With your help I managed to fix the configuration issues [16:32] heyho: woohoo!! [16:33] yeah! [16:34] heyho: was there something wrong with your system, or is there something we could do to make this process easier? [16:34] next big step will be for me to use juju on ec2 and orchestra! [16:35] there was a previous configuration of libvirt (because I used virt-manager) that I think it may have caused problems [16:35] even if i removed that [16:37] heyho: that sounds like something we could fix in juju [16:38] yeah, with my issues i think i made you think about a couple of things to improve.. expecially in Orchestra with that dict read problem [16:39] I found another guy had my same exact error reading a previous irc log [16:39] yeah I believe the fix for that is pending [16:40] SpamapS, heyho: I'll be on hat one just as soon as I can, I think it's a pretty serious problem [16:40] heyho, thanks for the extra information :) [16:42] No problem! And if you have other questions that could be useful to you, shout [16:46] As you surely have noticed, I'm pretty a beginner in the cloud computing world.. Do you think it would be better if I start trying to configure and use Openstack the hard way, or If I use orchestra to make, maybe, things go smoother? [16:47] heyho: orchestra is, I would say, the most complex target for juju. [16:47] heyho: local is second, and then ec2 is the simplest. [16:52] I also tried ec2 and juju bootstrap worked like a charm actually :D ... but I had issues in the juju status step.. But it's a big story that I maybe be explaining another time mainly because it's 17:52 pm here and I'm about to go home :D [16:53] heyho: was that ssh error during status [16:53] ? [16:53] heyho: well thanks for sharing all your troubles with us. :) [16:55] yeah ssh error problem [16:55] and thank you for your help [16:55] heyho: u need to generate a keypair :) and it will work [16:55] cheers!! [16:56] SpamapS: i was looking at owncloud, our repository still has owncloud1 [16:56] there newest release is owncloud2 [16:57] kollhead11, actually i generated it but had problems however.. I'll explain in details another time.. for now, cheers! [16:58] cool [16:59] Maintainer: Kubuntu Developers [16:59] Original-Maintainer: Jonathan Riddell [16:59] koolhead11: yeah, maybe file a bug? ;) [17:00] SpamapS: doing right away [17:07] Finally have a fully functional openstack environment… 2 servers with 6 cores each… Seems to work pretty well…. Now for the development of virtualbox as a provider in Juju… :-) [17:08] mchenetz: how did you manage only 2 servers? [17:08] SPamaps: That is the minimum… You can run everything but compute on one node and then compute on the other [17:09] mchenetz: so did you do that manually, or figure out some way to trick juju into installing more than one thing onto one box? ;) [17:10] I used stackps.org distro (Ubuntu 10 derivative) to install… Super easy… It was great to learn on… [17:10] stackops.org [17:11] used 2 nics… One for the service network and the other for public [17:12] now i am just setting up the Juju environment for EC2 compatibility and see if i can deploy to it... [17:20] jelmer: i think you helped me with the bzr thing [17:20] hi koolhead11 [17:20] hazmat: ping [17:21] jelmer: i will trouble you again in sometime. :) [17:21] lynxman: hey, you expired from charmers. I assume you'd like to stay in it right? (new memberships are getting much longer periods than we initially gave people.. not sure why we did that. ;) [17:22] SpamapS: yeah would love to stay in :) [17:22] SpamapS: if that's okay, of course! [17:23] lynxman: yeah, you've been quite active. :) [17:23] SpamapS: trying to, right now doing some juju client hackery that will make hazmat shriek ;) [17:24] * SpamapS makes sure everybody expires "never" :) [17:24] SpamapS: lovely, thanks man :) [17:24] I figure we can start expiring people when we get to 20 charmers. :) [17:25] medberry: aplogies to you, the one guy I did let expire. ;) [17:25] * koolhead11 listens [17:26] * koolhead11 rushes 4 home [17:26] laters [17:27] SpamapS, nope, I pulled myself off the list. You didn't let me expire. :^) [17:27] * medberry still had a day left at that point [17:34] lynxman, pong [17:35] hazmat: trying to do something a bit crazy, even if I know this will fail, would just like to try the concepr :) [17:36] hazmat: I'm writing an stunnel client charm, I'd like to deploy that using juju client then once its deployed connect to the zookeper node through the stunnel [17:36] hazmat: how do you see that? :D [17:37] lynxman, sounds interesting, the client is going over ssh now, but a generic stunnel charm would be interesting, its seems to me like its better targeted as a subordinate/plugin charm [17:38] since you'll need some local per charm state about the endpoint its setting up [17:38] hazmat: yeah, I think so, I'd like to try the basic concept for now and expand on that [17:38] hazmat: so my questions are, 1. how can I use juju client to deploy a charm without zookeper being present? 2. How do I tell the client where to find zookeper [17:39] 1) does not compute ;-) 2) its in the provider storage under a special named key 'provider-state' [17:40] lynxman: is there a reason you are shunning SSH? [17:40] * SpamapS shhhould shun sh words [17:40] re 1) it would have to be incorporated into juju's bootstrap [17:40] SpamapS: the idea is to be able to setup a machine in a remote location, then use stunnel to connect back to zookeper to tell it "I'm here and available" [17:41] * hazmat grabs some lunch [17:41] lynxman: ahh, you want what I was thinking.. a "just run the agent somewhere" provider [17:41] SpamapS: exactly! we talked with sabdfl also about it last week, be able to just run a charm in "standalone" [17:41] hazmat: enjoy :) [17:41] lynxman: I think with orchestra you could cheat and do this by defining a machine in orchestra, and then running the late command manually. [17:42] SpamapS: hmm how so? [17:42] lynxman: just put it in orchestra... tell juju to claim it, and then run the "late command" that gets shoved into ksmeta [17:43] lynxman: you still need zookeeper though [17:43] SpamapS: hmm thats the thing, I really don't fancy having a local zookeeper, that's plan B [17:43] Thats far too ingrained into juju to mock out. :) [17:43] lynxman: I'm 99% sure that its just not going to work without ZK [17:43] SpamapS: so maybe I can mock up something through cloud-init, of the sorts "install this stuff, then run the command to join zookeeper" [17:44] lynxman: you can certainly run ZK wherever you want and just point the client at it. [17:44] SpamapS: that'd be a good concept imho :) [17:44] as hazmat said.. just change the data stored in the provider state bits. [17:45] SpamapS: cool, will investigate into that then ^^ [17:45] SpamapS: hazmat: thanks for the feedback guys :) [17:45] lynxman: very interested in this, as it would be nice for the scaled-down services. [17:46] SpamapS: that's the direction I'm taking indeed :) [17:46] SpamapS: I'm keeping this under wraps for now, but #siteam will show something awesome in the next month if all goes right [17:49] lynxman: w00t [17:54] <_mup_> juju/ssh-known_hosts r442 committed by jim.baker@canonical.com [17:54] <_mup_> End-to-end [17:54] jelmer: hi again [17:55] hi [17:55] <_mup_> juju/ssh-known_hosts r443 committed by jim.baker@canonical.com [17:55] <_mup_> Merged trunk [18:05] jelmer: i modified a file and before that i did "bzr lp-login" and then bzr commit -m "added revision number in revision file" [18:05] i got message Committing to: /home/atul/bazaar/boa/trunk/ [18:05] modified revision [18:05] Committed revision 2. [18:06] so does that mean this committed file is still in my local system? i need to push it again [18:24] ooh it was bzr push :D [18:24] got it. [18:24] koolhead17: yes, commit only does the change locally - you have to push to get it onto launchpad [18:25] koolhead17: you can make "bzr commit" automatically propogate the change to launchpad by "binding" your local branch to the one on Launchpad [18:25] koolhead17: see "bzr bind" [18:25] jelmer: cool . let me check [19:21] SpamapS: ariund? [19:21] *Around [19:22] koolhead17: sort of :) whats up? [19:23] SpamapS: could you tell me that command for modifying the juju documentation, i want to remove "revision" part from yaml file, as user will get deprecated warning thing. [19:23] koolhead17: recently the docs were split out of the trunk.. [19:23] hazmat: are the online docs built from lp:juju/docs ? [19:35] SpamapS, no.. thats the last item till i can close that ticket out and announce [19:39] hazmat: still not sure of why we can't just pair the docs with the trunk. ;) [19:39] but I'd rather just see it move forward so koolhead17 and others can start updating them [19:39] SpamapS, exactly [19:40] SpamapS, if it doesn't work out in by precise release, we change it back to the old model [19:41] I think I do see one benefit [19:41] which is that it will have its own review queue [19:42] SpamapS: hey, so would you (or hazmat) know anyone who could port our juju client to OSX? [19:42] hazmat: my prediction is that it will eventually be split into lp:juju/docs-precise lp:juju/docs-q [19:42] or anyone else in this channel :D [19:42] robbiew: I don't know of anyone, but I second the request. [19:42] robbiew: I have 3 macs here (all running ubuntu.. but.. I can reboot ;).. I can probably figure it out. ;) [19:42] it should just work, just a question of getting it installed [19:43] python-zookeeper being the only non-trivial dep [19:43] hazmat: somebody recently found a dependency on python-apt that broke for obvious reasons on osx ;) [19:43] lol [19:43] SpamapS, they where using the local provider i thought [19:43] oh well thats just not gonna work ;) [19:43] SpamapS, the local provider clearly won't work on osx, and the code is not imported otherwise [19:44] yeah I think it should "just work" but we need to pick a single thing, macports or homebrew, and get it working [19:44] * hazmat looks for the wires to plug his mac [19:44] Probably need to focus on Lion [19:45] actually I can run Lion in a VM on either of my macs [19:45] so I don't even have to reboot [19:45] robbiew: sounds like something that needs a bug. :) [19:45] btw...I wasn't suggesting either one of you add this effort to your plate ;) [19:45] robbiew: seems like users who are on OS X will pick it up if we can help them out. [19:46] hello everyone [19:46] from what I see, the macports that lynxman submitted have been ignored [19:46] i guess bug 826498 sort of assumes the client works, but nothing explicit [19:46] <_mup_> Bug #826498: virtualbox machine provider for osx local dev < https://launchpad.net/bugs/826498 > [19:46] so maybe need to go with brew [19:46] +1 [19:47] another (evil) way to go is a direct installer [19:47] and they already have python-zookeeper (in their zookeeper dist) [19:47] after that its just pure python libs [19:47] hazmat: oh then maybe setup.py will work after brew installing that? [19:47] hmm.. [19:47] we could at least have a juju.ubuntu.com/OSX page then [19:47] I am working on a virtualbox provider and brew repo for juju [19:48] I tried last night but everyone was asleep, I'm working on a charm for gitolite: http://sitaramc.github.com/gitolite/ [19:48] SpamapS, it won't we don't have deps marked explicitly [19:48] mchenetz: RIGHT and you are also working on being *our hero* [19:48] ;) [19:48] hehe [19:48] :-) [19:48] welcome back shazzner_ [19:48] I figured out one issue but still have one question [19:48] hello :) [19:48] so part of installing gitolite requires a copy of your public key [19:48] SpamapS, easy to fix that though, so it is easy_installable [19:49] * SpamapS must feed himself, but is interested to help upon return. [19:49] and I'm not sure how to do this with the juju script [19:49] I know juju requires a key, but I'm not sure how to send this [19:49] or possibly to copy it from authorized keys on the server [19:49] just trying to figure out the best way to do this [19:50] <_mup_> Bug #905467 was filed: Allow juju to be easy_installed < https://launchpad.net/bugs/905467 > [19:50]