=== almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away === asavu_ is now known as asavu [13:21] hi all [13:21] charm for statusnet is broken somehow [13:21] anyone managed to use it succesfully? [13:22] first i belive it's not idemponent... koz it fails on juju resolved --retry statusnet/X trying remove index.html that already was deleted [13:23] i think some automatical idemponency check for charms should be performed. otherwise charm store will be full of half-baked half-working charms. [13:24] and one question: can i use something other than m1.small on amazon? [13:30] oh. there is a bug reported against statusnet charm LP#914851 [13:30] <_mup_> Bug #914851: Fails to configure when title has a space in it < https://launchpad.net/bugs/914851 > [13:53] is it possible to install 2 services on 1 machine with juju? [13:54] each juju deploy runs new EC2 instance. it does not look smart [14:08] <_mup_> Bug #952262 was filed: Juju should use XDG directory layout < https://launchpad.net/bugs/952262 > [14:26] i found hwo to run other than m1.small! "default-instance-type" in config [14:58] antono: actually I think I found that it had some idempotency problems a while back yeah [14:58] antono: we do an automated install test right now [14:58] SpamapS: cool! [14:58] antono: http://charmtests.markmims.com/ [14:58] i going to create generic rails charm [14:59] antono: and a lot were broken by a recent change to config.yaml handling IIRC [14:59] antono: there's already a generic rails charm [14:59] oh, for real? :) [14:59] havent find it yet :) [15:00] can i clone all charms in one place with one command? [15:00] antono: try 'charm search rails' [15:00] i came from git background and not sure how to deal with submodules in bzr [15:00] antono: 'charm getall /path/to/charm/root' [15:01] wow. cool [15:01] oh git.. :-P [15:01] antono: note that charm getall is REALLY slow [15:01] SpamapS: is it possible to set somehow default repository [15:01] , [15:02] SpamapS: oh, bzr [15:02] :) [15:03] i'd like to get all charms somwhere to ~/.local/share/juju/charms and use it as default repository... unless other location specified [15:03] antono: I'd be happy to take a patch to charm-tools to provide a way to get them all branchd from git instead of bzr. [15:03] SpamapS: as for me bzr is ok :) [15:03] nice to learn something new [15:04] antono: a change landed last week that takes the default --repository from $JUJU_REPOSITORYY [15:04] bzr is about 5x easier than git to learn [15:04] SpamapS: yep. probably yes [15:04] It just makes it harder to rewrite history like git, so people get frustrated with it because they're used to that. [15:05] SpamapS: i found bzr rebase plugin :) [15:05] yeah, which is silly once you realize that the very act of doing 'bzr merge' is just like rebase, except it does not discard history. [15:06] SpamapS: does bzr merge put my commits on top of target branch? [15:07] antono: if you have branch A, and branch B, and you mege from B->A, then A will get one commit with all the missing changes from B. [15:07] SpamapS: and collocated branches is something really useful. i know they are here with bzr 2.5 [15:08] antono: there's a blugin for that I think now called cobzr [15:08] written, oddly enough, by the original architect of juju ;) [15:08] antono: http://labix.org/cobzr [15:08] SpamapS: not sure why but i'd like to have separate commits after merge [15:09] anyway. juju is way to go [15:09] i like it [15:09] but... some questions to you :) [15:09] antono: they're still there.. just wrapped up in a single commit. [15:10] does it allow to run some unit on specified machine? [15:10] i would like to run 3 app servers and 1 db on single amazon node [15:10] antono: two answers for that [15:10] antono: either a) use the local provider on that one machine (but this means you cannot scale beyond that one machine) [15:11] antono: or b) lets go on an adventure to see if the recent 'constraints' branch allows specifying machine ids.. [15:11] antono: there's also c) use my new experimental not yet landed 'charm splice' command to splice those 4 things into one charm. [15:12] SpamapS: thanks for your answers [15:12] * antono afk for 1 hour [15:18] d) all of the above? [15:30] marcoceppi: tell that to George Zip! [15:48] hmm. why bzr log does not pipe output to pager by default on ubuntu? [15:49] antono: because |less is easy? [15:50] oh noes. extra typing is totally wrong. same as expricit --repository requirement [15:54] antono: --repository is all but gone [15:55] SpamapS: nice :) [15:55] same |less should go by default [15:55] antono: antono |less is optional though.. if you want to search, | less.. if you just want to limit.. -l # [15:55] antono: its a different philosophy, but I personally hate things that push me into the pager like that. :-P [15:56] i love ubuntu for sane defaults... and i'ts clear why it's not piped by default in bzr. but on linux it could be default behavior. (i should rant in #bzr probably) [15:59] but if it goes to pager, how will I |grep -ri "stupid committer" to find all the dumb things someone did? [16:00] :) [16:01] marcoceppi: it would only go to pager if isatty(stdout) [16:02] its a valid, but annoying practice imo [16:02] and grep works ok with bzr-pager plugin... [16:04] SpamapS: what do you think about charm get lp:some/charm ? [16:05] antono: charm get is a bit of a hack... [16:06] antono: its mostly for getting charms from the official repo/dev series.. if you want a specific charm.. bzr branch works. :-P [16:06] antono: the whole 'charm get*' set of commands is supposed to go away once the full charm store backend service goes live [16:06] aha. i saw broken requests to store.charms... [16:07] how can i set default environment? [16:07] antono: in environments.yaml, you can set a root element 'default: xxxx' [16:08] antono: or JUJU_ENVIRONMENT [16:08] antono: sorry, JUJU_ENV [16:09] which is apparently undocumented. :( [16:10] SpamapS: me researching the code :) [16:13] At this point we have 2 JUJU_ environment variables affecting the cli.. and 1 in review.. we should probably make a dedicated environment variables document. [16:14] SpamapS: i think it should go to man juju [16:15] hahahahahaha [16:15] man page [16:15] thats rich [16:15] antono: I wrote a man page once. The dev team rejected it because it wasn't auto-generated :-/ [16:15] SpamapS: auto generated from source? [16:16] Is there any python solutions for such tasks? [16:16] yeah.. they want the man page to be generated from all the argparse --help output [16:16] nope [16:16] I'd have to write it myself [16:16] good idea [16:16] :) [16:16] sat down to [16:16] got disgusted with argparse [16:17] antono: btw, was the statusnet problem you saw bug 914851 ? I'm about to fix that (and add a test for it) [16:17] <_mup_> Bug #914851: Fails to configure when title has a space in it < https://launchpad.net/bugs/914851 > [16:18] SpamapS: yes, it is [16:18] Also i belive it doesnt clear /var/www on remove-unit [16:18] so next add-unit fails [16:18] need to check it [16:19] fails due to rm /var/www/index [16:19] now. it's obvious that just merging charms will not work [16:19] i mean just merge charms to run all on single machine [16:20] it should be some good conventions/guidlines for each charm [16:20] where should it put it's files [16:20] antono: what is "remove-unit" ? ;) [16:20] statusnet puts all to /var/www [16:21] antono: charms were never meant to be more than 1 per server [16:21] SpamapS: i can only say it's bad limitation [16:22] antono: the dev team wants to use LXC so the charms are completely namespaced from eachother [16:23] SpamapS: yeah. thats good idea. in this case it's ok to have charm per machine [16:23] and it's certainly will simplyfy things a lot [16:23] antono: unfortunately, thats complicated because LXC also namespaces the network.. so you have to figure out a way to make all the charms reachable from all the other nodes. On EC2 that won't work because the network is tightly controlled. [16:23] from charm-writer point of view [16:24] antono: t1.micro is pretty cheap... ;) [16:24] for testing. And for production.. m1.small's are the only sane thing. :) [16:25] i use m1.micro for my blog, mail, jabber, git etc [16:27] antono: right, none of those would conflict with the splice method. :) [16:29] antono: I actually think deploy and add-unit should take a '--machine' argument.. this is agreed upon as a more generick '--placement' option but its implementation is still pending. === hazmat` is now known as hazmat [16:35] oh cool, are you deploying status.net with the charm antono? [16:39] <_mup_> Bug #952355 was filed: Backward incompatible change: existing env boolean values cause total failure < https://launchpad.net/bugs/952355 > [16:42] jcastro: yes, just testing juju following instructions in your post [16:42] jcastro: probably i'll make a charm for http://rstat.us/ [16:42] koz it's better than statusnet [16:43] ah, cool! [16:43] and also support ostatus protocol [16:43] you can enter it in the charm contest! (See topic) [16:43] jcastro: yep, i saw [16:44] jcastro: you should also make juju documentation contest [16:44] i'd like to record some plain/text screencasts for juju [16:44] like this one: http://shelr.tv/records/4f53e3df8841ac0001000001 [16:45] so copy/paste will work and make things easier for newcomers [16:45] antono: rstat.us looks nice. :) [16:46] one more thing i going to charm is http://gitlabhq.com/ [16:46] should i put this to google doc before start? [16:46] antono: bugs.launchpad.net/charms [16:46] I have a bug for gitlab [16:47] gitlab would be quite awesome. [16:47] the google doc is not really authoritative.. its just richer and easier to organize than bugs.launchpad.net [16:47] SpamapS, I have a mind to just kill the gdoc [16:47] columns are good... [16:47] what do you think guys about obeying XDG spec for juju data? [16:47] antono: I saw that. Interesting idea. [16:47] i mean ~/.config/juju ~/.local/share/juju etc [16:47] https://bugs.launchpad.net/charms/+bug/931835 [16:48] here you go ^ [16:48] <_mup_> Bug #931835: Charm needed: Gitlab < https://launchpad.net/bugs/931835 > [16:48] jcastro: so i should just assign me and start? [16:48] sure! [16:49] done :) [16:49] i mean assigned ;) [16:49] hey nice [16:49] so gitlab needs mysql and ruby [16:49] we have those charmed already [16:49] jcastro: sqlite afaik [16:49] it has some slite specific sql for now [16:50] ah ok [16:50] no worries, just as easy [16:50] yep. [16:50] just as easy as git clone + bundle install [16:50] + some setup for apache/nginx [16:50] gitolite would be cool I think because we don't have a charm that is "I want my own git repo hotness" [16:51] jcastro: ruby isn't really "charmed" [16:51] jcastro: why should it be charmed? :) [16:52] i think it should be ok with apt-get install -y ruby1.9.1-full in install hook [16:52] antono: that bit.. "+ some setup" .. thats where the charm wins :) [16:52] at least such things work in other charms [16:52] antono: I don't think jcastro meant what we think he meant ;) [16:53] antono: indeed, just installing the components you need in the install hook is all that is needed [16:53] no I did not mean what you thought I meant [16:53] sorry [16:53] i have some chef experience [16:53] I was just thinking aloud when I saw their requirements and was like "we have that, that, and that." [16:53] it would be nice to reuse it in charms [16:54] you can absolutely do that [16:54] I am surprised no one has tried that yet [16:54] it has a lot of well tested idemponent recipes [16:55] chef/pupet recipes is a way to go for charms [16:55] like build blocks. tested. they have good collection of recipes [16:55] also it may increase portability of charms [16:55] koz for now they are only debian/ubuntu [16:55] antono: yeah, do that. :) [16:56] antono: there has been *zero* interest in porting juju to other platforms. [16:56] antono: people ask if its possible, thats it [16:56] I'd rather see a gitolite charm first though [16:57] People need a slick self-hosted alternative to github! [16:57] jcastro: right. same i do [16:57] i considered gitorious [16:57] antono: Maybe we're biased, being canonical people, but I think using juju means you don't have to care about which OS you're running. [16:57] but its hard to install [16:57] so charm would be nice [16:57] later i find gitolite [16:58] the nice thing is you can keep the charm up to date [16:58] you don't have to worry about if a package will make the ubuntu release or whatever. [16:58] yes. and it's much more hard to resolve all dependency hell in ruby land [16:59] is there any plans to push usage of augtool in charms? [17:00] charms in esence are very similar to less strict packaging systems like arch guys have [17:01] never heard of augtool [17:01] oh, the augeas thing [17:01] augeas, yes [17:01] almost silver bullet for configuration management [17:02] I don't see any reason why you wouldn't use it if you want to in a charm [17:02] in all this config zoo it provides unified api at least [17:03] hey, all! who participates charms contest? [17:03] there's been 3 or 4 so fat [17:03] err, so far [17:04] antono: I've used augtool .. I converted to creating files with appending though [17:05] antono: augeas is a neat idea.. and useful when you have multiple things influencing a config file that is complicated. But ultimately, its far more predictable to simply control the inputs, and generate the config in its entirety [17:06] jcastro: are you able to deploy stuff with r467 ? [17:09] SpamapS, I only use what you give me in distro bro [17:09] ahhhh [17:11] SpamapS: you are right. when every charm isolated in it's own container than it's easier to just fill config template with some values [17:11] but... augeas also have augtool get [17:12] wich is useful... not all this sed | grep | cut stuff [17:12] and more reliable than shell one liners... [17:13] antono: right, but what would you be extracting from config that you don't know as a charm author? [17:14] right, totally predictable... [17:15] antono: and even if charms need to both influence a config file, they should both collaborate around that, not try to avoid eachother. [17:17] hey peeps [17:17] where do we stand with subordinate charms? [17:24] james_w: still waiting [17:24] james_w: landing [17:24] I believe its all in review [17:25] m_3: did you run into the new brokenheaded way that juju is verifying *ALL* of the charms in a repository? [17:25] m_3: I can't deploy anything right now.. because nearly half of the charm sin my repo have a bad boolean [17:26] SpamapS: still on cli from earlier in the week [17:27] stay there [17:27] I"m going to raise a Critical bug [17:27] this is just *wrong* [17:27] I didn't have time to fix all of the charms this weekend... [17:27] task for next week! [17:28] dude, devopsdays has turned our submitted talks into a panel... juju,chef,puppet,crowbar,cloudformation,pallet,etc [17:28] whoohoo! should be entertaining [17:29] yeah sounds like fun [17:30] SpamapS: btw, how was your upgrade process for mactel? it's time to bump the lappy [17:30] <_mup_> Bug #952397 was filed: Juju is forcing *ALL* charms in a repo to be perfect to be able to deploy *any* < https://launchpad.net/bugs/952397 > [17:31] m_3: I upgraded a *long* time ago [17:31] m_3, man that sounds awesome [17:31] m_3: been working great [17:31] did my desktop earlier, but've been waiting on the laptop b/c of the travel [17:31] jcastro: picked up the gauntlet [17:31] m_3: its safe, go for it [17:32] I think I've got a great opener... love to run it by you guys this week [17:32] for sure! [17:32] m_3: tho I recommend unity-2d for battery life. :) [17:32] anyway, time to go enjoy my Sunday [17:32] <-- pool for me, cya guys [17:32] enjoy [17:32] ha nice [17:33] if you guys run into weird "ServiceConfigValueError" , its bug 952397 [17:33] <_mup_> Bug #952397: Juju is forcing *ALL* charms in a repo to be perfect to be able to deploy *any* < https://launchpad.net/bugs/952397 > [17:33] Hopefully we can get that fixed early tomorrow [17:34] yeah, maybe there's a nicer way to handle type-safety too === Leseb_ is now known as Leseb [21:14] Not sure what i doing wrong... but [21:14] juju deploy local:oneiric/mysql --repository . [21:14] 2012-03-12 00:13:57,830 INFO Searching for charm [21:14] Invalid value for cache: False [21:14] 2012-03-12 00:13:57,897 ERROR Invalid value for cache: False [21:14] [21:14] current repository contains output of charm getall [21:27] oh. it seems i was affected by LP#952397 [21:27] <_mup_> Bug #952397: Juju is forcing *ALL* charms in a repo to be perfect to be able to deploy *any* < https://launchpad.net/bugs/952397 > [21:34] hmm.. where should i report ami problems? can i use alternative ami? [21:34] 2 t1.micro hangs 10 minutes after start. system log shows [21:34] [ 167.094249] 1 multicall(s) failed: cpu 0