[05:34] jcastro, traveling to pycon today [05:34] jimbaker, where's the fun at? ;-) === almaisan-away is now known as al-maisan [06:39] hazmat, i was just downstairs in the bar area... regrettably calling it an early evening :) [06:39] jimbaker, me too [06:39] reworked the hook scheduler significantly on the plane [06:46] hazmat, a good place to do that [07:08] adam_g: when you get that error, that means you have used an old version of juju to spawn a node with a 'juju-origin:' that pulls in a newer version of juju. [07:08] adam_g: older versions don't start the agent with the right arguments, so it fails. === al-maisan is now known as almaisan-away [11:47] hi, good afternoon [11:47] any juju guru here that i can ask some question? [11:52] for a charm we are writing, we need to download code that resides on some private branches of bazaar. So we need to identify as a bazaar user, providing the public key. I was thining in sending the key in a config var in the charm, is this a good idea or there is some other way to do it? [11:54] yolanda, for now i would recommend doing it that way [11:55] ok, so i create some config var for bazaar user name, public key... and just use it in the hooks, right? === almaisan-away is now known as al-maisan === JanC_ is now known as JanC === al-maisan is now known as almaisan-away [16:02] how complete is the diaspora charm? === asavu_ is now known as asavu === asavu_ is now known as asavu [17:11] m_3: ping [17:11] SpamapS: ping [17:12] negronjl: ssup? [17:12] m_3: Can you review this: https://bugs.launchpad.net/charms/+bug/950968 for me to see if I can promulgate it ? [17:12] negronjl: pong! [17:12] <_mup_> Bug #950968: Charm needed: gearman < https://launchpad.net/bugs/950968 > [17:13] mmmm gearman [17:13] SpamapS: begging for reviews :) [17:13] negronjl: I am a huge gearman fan [17:13] SpamapS: https://bugs.launchpad.net/charms/+bug/950968 [17:13] <_mup_> Bug #950968: Charm needed: gearman < https://launchpad.net/bugs/950968 > [17:13] negronjl: I have code in gearmand actually. :) [17:13] * SpamapS wrote --round-robin, and 1/2 of the tokyocabinet persistence queue [17:14] SpamapS: you wanna review it or you want me to? [17:14] SpamapS: cool... Now I know who to complain about it to :P [17:15] negronjl: yeah, in the last 2 years gearmand has kind of gone down a rocky road with the addition of C++ [17:15] I think its for the best.. but boost has just made everything harder to build. :-P [17:16] SpamapS: So far, I have nothing bad to say about it. It has actually helped me in a few occasions [17:16] m_3: there's no problem w/ 2 reviews.. :) [17:16] negronjl: first simple problem.. you forgot to remove README.ex from gearman-worker [17:16] SpamapS: ahhh... boost.. haven't used it in about a year....it was good to compile and build my stuff ... a horrible pain to port it anywhere else [17:17] SpamapS: i thought i had fixed that ... checking [17:17] negronjl: also pecl is not safe.. no cryptographic verification. :-/ [17:18] * SpamapS curses the pedantic license quibble that has kept php5-gearman out of debian [17:19] negronjl: all in all though, I love the way this looks. Its kind of more of an example to be built on, but it gets the concepts of gearman quite well. :) [17:19] negronjl: mediawiki actually has a gearman worker btw.. for doing image thumbnailing or something [17:20] SpamapS: I just fixed the README [17:20] SpamapS: I just forgot to commit [17:21] SpamapS: not much I can do about pecl though ... If I take it out, not much functionality remains :/ [17:22] SpamapS: One thing that I can do to build on the charm is to provide a few tests/examples in the README file so, the user has something to test with ... thoughts ? [17:22] negronjl: you can pin to a single version (0.7.0 is really old btw) and build the tarball without the pecl command's help [17:22] negronjl: also, gearman-server is the perl version. gearman-job-server is the C one [17:23] SpamapS: I thought it better to use the perl version as it would be (in my mind) more portable and easier to deal with [17:23] negronjl: yeah I think it might make sense to actually have the charm exit 1 at the end of install, and have a comment like "echo THIS DOES NOT DO ANYTHING WITHOUT A WORKER" [17:23] negronjl: that may actually be true. :) [17:24] negronjl: the C version is about 5x more memory efficient, and about 2x faster ... and has built in persistence (but kind of lame persistence) .. but the perl version is way more stable. [17:24] SpamapS: If I were to exit 1, the charm would look broken to a somewhat new Juju user ... it may complicate things ... My thought is that, if you are going to deploy gearman, you probably know that you need server, workers, clients. [17:28] SpamapS: I just added a Testing sections with examples in cli, python and php to the gearman-server and gearman-worker README files [17:30] negronjl: yeah its a crazy idea. But ultimately, if you deploy that worker charm, you might expect that you have workers now.. which concerns me.. it doesn't actually explain how I go from that, to deploying my worker (I'm assuming you expect them to fork it?) [17:30] negronjl: one thought is to make it do something.. like have a 'log' function and have it log to disk. [17:31] SpamapS: Let's leave it in a branch for now as I can't think of an easy way to have it register something. The idea is for the workers to register their functions ( which we won't know ahead of time ). [17:31] SpamapS: For now, I agree that this is a sample template charm for gearman [17:31] SpamapS: Something where people can build upon [17:32] yeah, perhaps its time we started pushing for tags in the charm metadata so we can flag things as templates [17:32] SpamapS: that sounds good [17:32] SpamapS: I should change the bug from new-charm to new-template [17:32] SpamapS: or new-charm-template [17:32] negronjl: I'd still like to see it in the charm list... [17:33] negronjl: I just want it to be clear to a user that this charm needs some extra love before it will do anything useful for you. [17:33] SpamapS: Is there a more subtle way of letting them know ? [17:33] negronjl: another idea is to have config options kind of like m_3's rails charm that will just checkout and run a worker [17:33] SpamapS: Ahh...that I can investigate [17:34] SpamapS: I'll check that out over the weekend .... but, that sounds like the better option for now. [17:35] yeah that would be kind of cool.. juju deploy gearman-worker 'imageresizer' --set worker-branch git://mygitserver/foo [17:35] SpamapS: yup... that's what I'll do ... thx for the tip [17:35] * SpamapS reboots hoping to observe the miracle of starting up again in about 2 minutes [17:42] \o/ [17:49] SpamapS: welcome back to the land of the living === TheMue_ is now known as TheMue [20:17] --constraints CONSTRAINTS [20:17] Hardware constraints for the service [20:17] --constraints CONSTRAINTS [20:17] Hardware constraints for the service [20:17] worth two pastes [20:17] AWESUUHMMM [20:31] juju.charm.errors.ServiceConfigValueError: Invalid value for cache: False [20:31] Did we make another backward-incompatible break recently? [20:32] ahh yes we did [20:32] cache: {default: false, description: Whether to cache pages or not, type: string} [20:32] I have that deployed in my environment. [20:32] Seriously.. we're going to have to take action, this is getting ridiculous === hspencer is now known as hspencer[brb] [21:38] * SpamapS completes the russian roll on mocker and is nearly ready to submit namespace-from-env for a 3rd round of reviews.... [21:52] SpamapS: There are numerous charms that are now broken due to the stricter checking in config.yaml. [21:52] SpamapS: I did fix a few of them yesterday but, I'm afraid we may have to (re)check them all to ensure that they work with the current juju revision [21:55] I thought we had already found them all [21:55] there was supposed to be some time where it was just a warning. :-( [21:55] and further.. it should not break running environments! [21:55] * SpamapS cannot even run status on an env with a "bad' charm [22:32] darn.. forgot the core team is at pycon.. no chance they'll be doing reviews today. :-P [22:52] jamespage: https://code.launchpad.net/~james-page/charms/oneiric/cassandra/10x-plus-updates/+merge/96849 .... Reviewed and approved! ... Thanks!