=== flaviamissi_ is now known as flaviamissi [05:54] hazmat / SpamapS ( i know neither of you are around yet, this is for later ) : can their be custom "hybrid" providers , e.g. unholly marages between like google storage ( that provides s3 api and even uses s3cmd for its cli tool ) and ec2 or an openstack compute , and if so how would one such env.y look ? [05:54] s/their/there [06:14] imbrandon: it would need to be written, but it wouldn't be hard to make happen [06:14] imbrandon: I've suggested a few times that we should consider decoupling storage from compute [06:15] imbrandon: also I think we can remove the S3 dependency entirely [06:16] imbrandon: just fire up something for storage on node 0, and use ec2 groups to find it. [06:39] SpamapS: awesom [06:42] SpamapS: next question is do we do that in go or python, i'm kinda curious to get started with some go [06:43] SpamapS: btw i did a lil screencast bout blitz'ing omgubuntu today , go like or plus one it or whatever its called :) [06:43] :) [07:15] imbrandon: nice [07:15] imbrandon: so we're stuck in a weird spot right now [07:15] imbrandon: python dev is supposed to cease... [07:15] imbrandon: but go is still not really useable [07:16] imbrandon: they'll be usable soon.. and caught up I'm sure in a "months" measurable time frame [07:20] =o [07:20] yeah, I am =o too :-P [07:20] SpamapS: python dev will cease wha? [07:21] bkerensa: yeah, all new features will go into the go version [07:21] oh [07:21] but.. [07:21] so python-dev will not be carried as a package anymore? [07:21] I dunno.. they seem very far behind. [07:21] bkerensa: no we'll keep shipping python-dev in Ubuntu.. ;) [07:22] oh l [07:22] ok then :D [07:22] bkerensa: or are you meaning, python-juju ? [07:22] That version is the only working version of juju right now, so it will continue to be the one that goes into Ubuntu until the go version can do everything it can. [07:22] python2.7-dev [07:22] :P [07:23] right ok well I'm too tired to know if we're talking in circles or I'm just out of brain power [07:23] so.. I'll just sleep. :-P === OutOfControl is now known as benonsoftware === almaisan-away is now known as al-maisan [10:15] bkerensa: we're only talking of python dev ceasing in relation to juju core, not ubuntu as a whole, juju is being rewritten in go [10:15] SpamapS: gnight , see ya when ya wake :) === al-maisan is now known as almaisan-away === daker_ is now known as daker [12:32] hola [12:45] Hello all [12:46] * koolhead17 hola backs Daviey :P [12:46] niemeyer, hi [13:00] niemeyer, heyhey [13:00] niemeyer, is it meeting time? [13:02] fwereade__: It was 14UTC, but if everybody is ready we can do it now [13:02] niemeyer, I'm easy, just hadn't picked up on being 2 hours off instead of 1 === ahs3` is now known as ahs3 [15:19] fwereade__: hey, did we forget to announce subordinates on the mailing list? [15:21] jcastro, if yu mean constraints, then ...er, probably, there hasn't been anything since the warning mail a couple of weeks ago [15:22] hmm, did I miss the mail on subordinates then? [15:23] jcastro, ...and if you do mean subordinates then I'm afraid I have no idea, I've been up to my eyebrows in constraints exclusively, sorry [15:23] ok I think I am mixing you up with someone else. :) [15:23] but hey a quick "hey guys this is finished, here are the docs" wouldn't hurt for constraints either. [15:24] ah, subordinates, that's hazmat and bsaller right? [15:25] hazmat: ping [15:25] jcastro, thanks, good plan [15:31] hrm i deleted this branch long ago, http://jujucharms.com/~imbrandon/oneiric/quickdrop [15:31] should it not get cleaned up ? [15:31] when? [15:31] on cron ? [15:32] as in the branch has been gone for weeks [15:32] did you just now remove it? [15:32] oh [15:32] sounds like a bug then [15:32] maybe 2 weeks [15:32] sec [15:32] kk [15:32] I'll file it [15:32] kk ty ty :) [15:32] it's lp:~charmworld for the web UI btw [15:32] ahhh kk [15:33] btw i installed hub, makes git work with github like bzr does lp , soooo nice [15:34] git clone drupal6 == clones my fork of drupal6 or , git clone jcastro/drupal6 == for yours :) === Furao_ is now known as Furao [16:33] jcastro, pong [16:33] yeah.. i don't think we've announced the subordinate capability [16:33] they landed end of last week [16:33] * jcastro nods [16:33] we should [16:33] wouldn't hurt to post it this week [16:33] I've already written puppet as a subordinate charm [16:33] constraints was just announced! [16:34] and their even documented.. https://juju.ubuntu.com/docs/subordinate-services.html [16:34] Yeah the docs are very good [16:34] * imbrandon is working on his first sub charm now, newrelic [16:34] though the 'juju-info' relation is not super discoverable in them [16:35] hmm.. good point [16:35] oh yeah we need to make an RDS sub charm [16:35] and an ELB sub charm would be cool too [16:35] hell yea [16:36] also a mysql proxy sub charm with the 'is this a write or a read' ruleset that can then send writes or reads to masters/slaves [16:36] <-- full of ideas, out of time ;) [16:36] SpamapS, why those be sub charms? their both standalones [16:36] rds and elb [16:36] hazmat: you're going to run an entire instance, to make API calls? [16:37] hazmat: external services work great as subs [16:37] SpamapS, or i propogate ec2 credentials everywhere? [16:37] or, should anyway [16:37] hazmat: thats why Amazon has the sub-creds stuff [16:37] sub keys [16:37] * SpamapS forgets the acronym [16:37] jcastro, so i'm not entirely clear that we ever allow 'deletion' of a charm [16:37] iam [16:37] hpcloud has it too [16:37] niemeyer, does the charm store ever delete a charm because its upstream branch in lp is gone? [16:38] hazmat: Nope [16:38] hazmat: I happen to agree with you, but short of placement: local .. we really shouldn't waste an instance on stuff like that. [16:38] doh [16:38] hazmat: the charm in question is mine, i renamed it and the old one is gone for good [16:38] we need like, a daily delete cron or something [16:38] like for weeks now [16:38] i'm sure there are others though [16:39] deletes.. the bane of all syncing systems [16:39] http://jujucharms.com/~imbrandon/oneiric/quickdrop [16:39] --delete-after :) [16:39] Do we care if they're deleted or not, maybe just "don't display"? [16:40] well it dont display now, i found it via the interfaces [16:40] well dont display on the main cs [16:40] clicked interfaces, then http, then there i was :) [16:40] heh [16:41] jcastro, imbrandon the problem is someone may have already deployed it [16:41] oh [16:41] and at that point deleting it is.. is really bad for a user of the charm [16:41] but yes ultra hide seems ok since thet are essentially namespaced by dev, so even someone with the same name charm would be ok later [16:42] ok so I would think we'd do that for cs:foo, but not cs:somedude/foo [16:42] hazmat: not knowing its un-maintained might be as bad though [16:43] and if the branch is gone , whats the hurt in deleting it, it cant be deployed anyhow [16:43] imbrandon, true... for the browser we could hide it from listings and search, but show it on direct nav [16:43] and perhaps highlight its unmaintained status.. [16:43] SpamapS, you around? [16:44] hazmat: that's what I was thinking [16:44] hazmat: yea i'd say hilight its unmaintained too [16:44] and even put that in juju status if able to [16:44] jono: I am! wassup? [16:44] "branch not found, there's a good chance this might be unmaintained, here's a link to the official one in this same namespace..." or something [16:44] SpamapS, I have these nodes that just arrived at my place, I am going to bring them in this afternoon, but will need a hand getting them in the venue [16:44] can you help me? [16:44] jono: he's not there, you want zul or Daviey [16:45] jono: I arrive Thursday morning [16:45] heh [16:45] ahhhh np [16:45] jono: when I said I'm "around" .. I'm .. like.. fat, and spilling all over my chair. ;) [16:45] zul, Daviey can either of you guys help? [16:45] SpamapS, lulz [16:45] they should have made openstack closer to with uds so i coulda went :) [16:46] hahah [16:46] imbrandon: it has become the pre-drinking event to get your liver ready for UDS [16:46] hahah [16:46] jono, i'm around [16:46] jono, i can rope in folks as needed to help out [16:46] i'm good then, no need to see jono bottle dance after the first time [16:47] although i am hoping he drags along a les paul or some such this time since its not that far :) [16:47] hazmat, that would be cool [16:48] I am going to go and see if they fit in my car in a few mins [16:49] heh.. jono is juju's roadie [16:49] hah [17:02] man, we need to revive the effort to have '--set' passable to deploy. That was a great idea. [17:07] --set for like, initial config options? [17:08] yeah [17:08] +1 [17:08] Though I see a lot of really bad practice with that [17:08] people writing "this will only be applied at install time' [17:09] I am pondering making config-get in 'install' a warning in charm proof. configs should really only be applied in config-changed [17:09] otherwise they lose their magical powers to morph the service as needed [17:09] SpamapS, well they could get defaults in install, you can set config at deploy [17:10] its a fail though for charms that require config at deploy imo [17:10] hazmat: that doesn't make any sense though. It can be changed, so it should be handled *only* in config-changed. [17:10] even if it is the default [17:10] install is for stuff that *always* has to happen [17:10] SpamapS, true and we do execute config-changed always before start [17:10] exactly [17:10] SpamapS: wouldn't it eventually be handled in config-changed even during first install [17:11] marcoceppi: yes [17:11] so, yeah [17:11] so anything that you do 'config-get' for in install, should be moved to config-changed [17:11] * marcoceppi goes back through his charms [17:11] I'm sure we've all made the mistake :) [17:11] SpamapS: tells SpamapS minus one year there is a json interface :) [17:12] :) [17:13] inorite [17:13] SpamapS minus one year had no config-get [17:13] hehe [17:16] jcastro: going to send the shirts today...to our west coast warehouse, aka jono's house [17:16] nod [17:16] I'm bringing the micro right? [17:16] yes [17:16] I'm bringing another, but we need a murphy's law machine [17:16] ;) [17:17] two if jcastro is responsable for one [17:17] * imbrandon dips his head back into the code [17:22] imbrandon: you've published things on OSX right? [17:22] imbrandon: maybe we should look at the OSX juju work? [17:22] yea [17:22] and yea [17:23] https://juju.ubuntu.com/OSX [17:23] any help here would be <3 [17:23] ohh awesom the hardpart is done [17:23] sweet [17:23] yea i'll get this on wrap sometime in the next 24 [17:24] you have 10 days actually. :) [17:24] infact looking at this, i can get it down to one command [17:24] but getting it all sorted before then would be wicked [17:24] jcastro: great cux my laptop only runs osx atm , that might change at usd [17:24] but lol [17:24] nah i'll be dual booting [17:25] but yea for sure in 10 days, i can get this down to one command i can allready tell [17:25] sweet [17:25] imbrandon: if we need to host some stuff on officialish resources I can help you with that [17:25] we _need_ to work awesome on OSX [17:25] nah we want it in the official brew trree [17:25] oh ok, well, whatever you kids are into [17:25] that way a user littrally says [17:25] brew install juju and is done [17:26] that would be nice [17:26] and no sef respecting devops dosent already have brew installed, but if not its just one more command [17:26] self* [17:26] but yea, i'm on it [17:27] oh wow these are old too, this is a few months out of date [17:27] sweet ok, /me is happy now [17:28] infact jcastro i may take you up on the hosting thing [17:28] the more i think about it i can also produce dmg installers [17:29] i forget i'm a "official" apple dev and can code sign packages :) [17:29] lol [17:29] I will get you anything you need [17:29] sweet yea, it would only be the equiv of a .deb [17:30] but with 10.8 you they have to be code signed with an approved key [17:30] damn appple walled garden, but lickily i have such key for macos and ios for atleaste the next 10 months [17:31] ( $100 bux a year each , but i just renewed ) [17:32] SpamapS: rember that ,its the "Seem exclusive but not Be exclusive", like a stripper :) [17:33] yeah, thats what new-charm is :) [17:33] I think anyway [17:33] :) [17:33] SpamapS: when you get a sec, I'd love to go over precise charms [17:34] m_3: any glaring problems? [17:34] SpamapS: nope, not at all [17:35] SpamapS: just leaning towards _not_ making any big changes through launchpad this week [17:35] jcastro: wont happen in 10 days but i can get it into the app store too ( approval process takes longer than 10 days ) [17:36] SpamapS: just push and demo against cs:~charmers/precise/ for what's needed [17:36] m_3: hrm [17:36] m_3: where's your sense of adventure? [17:36] SpamapS: I don't have a [17:36] ha! [17:37] don't have any specific problems with upgrading across the board other than lots of moving parts and the risk of breakage [17:37] demo faeries(sp?) and such [17:38] more charms "just worked" than I expected [17:38] well that's a good sign at least. [17:39] jcastro: yeah, the process went really well... I'm just pretty skiddish when it comes to changes during big demo time [17:40] skittish? hell, I need more coffee [17:40] switch to IV, works better i found [17:40] well hey, if we want to flip the switch next week then that means we can promulgate the next three in the queue right? :) [17:42] m_3: the MaaS demo isn't using cs: is it? [17:43] robbiew: ^^ ? [17:43] imbrandon: I bet [17:43] SpamapS: it sounded like they were excited about showing the charmstore [17:43] It would be nice, not a requirement [17:43] tbh, we'll probably script the deployment anyway [17:44] well, if we don't have to fix it now we have to fix it by release [17:44] so we might as well fix it now [17:44] and walk folks through it [17:44] SpamapS: in discussing it last week, it sounded like they were ok with cs:~charmers/precise/ though [17:44] 'juju deploy cs:~charmers/precise/hadoop' -vs- 'juju deploy hadoop' [17:44] I wouldn't worry about affecting the demo...hell, we might use local charms just to avoid any surprises [17:45] * robbiew goes to ship boxes...back l8r [17:46] SpamapS: ok, well it sounds like we should pull the lp switch to promote them to precise then [17:55] m_3: charm school tomorrow for QA folks! [17:57] jcastro: yup... is there anything in particular that peeps wanna see? [17:57] nope, it sounds more like it'll be a Q+A of their questions, [17:57] yeah... cool [17:57] jcastro: are there few enough to do it in a hangout? [17:57] jcastro: or should I spin up juju-classroom [17:58] I asked for G+ but they want phone+IRC I guess [17:58] so maybe it's more than 10 [17:59] jcastro: don't see anything about phone... is there a conference thingy? [18:00] not obvious to me, I'll follow up over email, you'll be CCed [18:00] jcastro: cool thanks [18:01] jcastro: re the openerp6.1 split... there can be as many openerp charms in the store as people want [18:01] right I get that, what I am wondering is if it makes sense to do that [18:01] sorry, not in the store... int he repo [18:02] oh right [18:02] in the store, we need to have one [18:02] imo [18:02] I'd like them to just play together and figure out how to combine them [18:03] but I understand that may be unrealistic [18:03] so either combine, or pick one to keep in the store and one to keep in a separate repo [18:03] jcastro: cool with you? [18:04] sure [18:04] I just didn't want them to accidentally duplicate work [18:04] right [18:21] http://www.rackspace.com/blog/next-generation-rackspace-cloud-servers/ [18:21] anyone try juju on this yet? [18:21] marcoceppi: new toys ^ [18:22] jcastro: I signed up for their beta through my company, never heard back from them. [18:24] jcastro: I need to setup awsome proxy first but I'll give it a try [18:24] marcoceppi: strange I got a reply the next day [18:26] SpamapS: I'll hit lunch then we can G+? We can talk charm store too with m_3 if we want [18:26] jcastro: can we G+ first? [18:27] jcastro: I'm ready right now [18:27] SURE [18:27] win 7 [18:27] fire it up [18:27] doh [18:27] Ok I'll initiate [18:27] SpamapS: invite the others too if people want to hang out [18:27] btw, but 983313 [18:27] I am feeling very "light at end of the tunnel let's hug alot." [18:27] bug 983313 [18:27] <_mup_> Bug #983313: New Charm: puppet + puppetmaster < https://launchpad.net/bugs/983313 > [18:28] jcastro: me too [18:28] but with mayonaise [18:28] i'll hang, if nothing else to listen [18:29] imbrandon: its mostly logistics around our demo Thursday.. right after the hugs [18:29] lol [18:29] k /me goes back to fixing the lxc [18:39] SpamapS: hey [18:40] SpamapS: looks like it worked... aliases are pointing at precise [18:40] oh they are? [18:40] SpamapS: yay, ok, so I'll start manual changes for ones that're not supposed to be there in precise [18:44] test out 'juju deploy ' if you get a chance... here're the ones we expect to work http://ec2-23-20-58-1.compute-1.amazonaws.com:8080/ [18:44] I'll do the same after fixing some links [18:47] SpamapS: perhaps we need another '--fix' round from charmtools? /me looking to see if bzr keeps the actual branch or the alias [18:52] m_3: alias [18:52] m_3: or at least, it should [18:53] parent branch: bzr+ssh://bazaar.launchpad.net/+branch/charms/subway/ [18:54] jcastro: woot i now have the ability to blitz with 625 concurrent users ( standard free acct is 250 ) [18:54] looks like others liked the video too :) [18:55] Brandon, [18:55] Cool screencast! We've added one-time +250 credits to your account so you free plan gets bumped to 500 concurrent users. Of course, with referral credits you can get this up to a max of 750! [18:56] from blitz.io in my email [18:56] :) [18:56] imbrandon: that scool [18:56] hellz yea :) /me is even more happy now [18:56] i already had 125 referal credits [18:56] so 625 now :) [18:59] Ok, folks PRECISE is the new dev focus for charms. [18:59] charm promulgate will now promulgate to precise [19:00] lp:charms/* is now precise [19:00] all branches have been copied forward [19:00] oneiric, may ye rest in stability.. aaaahhhmmeeen [19:00] SpamapS: ok so on precise "juju deploy mysql" will do precise? [19:01] yep [19:01] its like we planned it this way :) [19:03] SpamapS: jcastro:technical difficulties... one sec [19:03] m_3: clearly :-P [19:28] SpamapS, are you not at ODS? [19:29] koolhead17: no, I go on Thursday [19:29] SpamapS, okey. We have Juju/Maas?openstack demo there? [19:45] niemeyer: an hour ago we copied all the oneiric charms to precise, but they're not showing up as cs:precise/xxxx ... what criteria does the charm store use to determine if it should import a charm? [19:46] SpamapS: Can you give me a sample of a charm that isn't showing but should? [19:47] SpamapS: If you copied all of them at once, it might be that it's still importing [19:47] SpamapS: I can't check that without help from someone with superpowers [19:48] hello there. I'm trying to juju ssh to an running instance but I'm getting this error: Incompatible juju protocol versions (found 1, want 2). Is this a known error/bug? Is there a workaround? [19:49] I'm using juju: 0.5+bzr531-1juju5~precise1 fwiw [20:00] f/win 3 [20:09] SpamapS: so maybe metadata[maintainer]... and have promulgate create-or-add-to a "-maintainer" group. Ok, I guess that's simpler than I was thinking [20:10] s/-maintainer/-maintainers/, but keep the metadata field singular so there's a single point-of-contact [20:17] I'm getting a... bzr: ERROR: Permission denied: "~charmers/oneiric/znc/trunk/" when trying to backport a charm fix to oneiric [20:18] SpamapS: who were you working with in lp with super powers? [20:18] (the precise lp:charms/znc pushed just fine though) [20:25] jcastro: hey, can you send me the final of the charmschool handout? I wanna make sure the node charm is named / behaves the same... I've used two versions over time [20:26] m_3: in your inbox! [20:27] jcastro: thankyouverymuch [20:29] It's an aubergine world :) [20:34] negronjl: hey man, please test out 3-node mongodb replsets on precise when you get a chance... it's front-and-center in the juju handout we'll be using at ODS and I haven't run it on anything but oneiric (lp:charms/mongodb now points to lp:~charmers/charms/precise/mongodb/precise). [20:35] m_3: I'll do that sometime today [20:35] m_3: Let me finish some MaaS testing first [20:35] negronjl: thanks! [20:35] jcastro: crap... typo [20:36] jcastro: "juju add-unit -n20 hadoop hadoop-slavecluster" should read "juju add-unit -n20 hadoop-slavecluster" [20:37] jcastro: at least it's not really a big one at all [20:37] didn't you review this like 3 times already? [20:37] I reviewed an early version... the google doc [20:37] that was the final draft I sent you, I don't know what's actually on the paper at ODS right now [20:38] m_3: any losa in #launchpad-dev can do it. thedac helped me out [20:38] SpamapS: thanks [20:38] m_3: jamespage reviewed it too [20:39] m_3: I am hoping what I sent you is just wrong. :) [20:39] matsubara: re your problem, are you by any chance spawning oneiric machines with 'juju-origin: distro' ? [20:39] jcastro: no biggie... it's a small one. We can catch it before the nxt batch gets printed up. nothing to do before ods [20:41] niemeyer: I tried 'statusnet' and 'subway' [20:41] SpamapS, this is my environment.yaml for that env: https://pastebin.canonical.com/64366/. answering your question, no, I spawned a precise machine and have no juju-origin: distro in my environment.yaml [20:42] SpamapS, that environment was bootstrapped on 2012-02-29 and have been running since then [20:43] matsubara: ugh, you may have to re-bootstrap [20:43] SpamapS, really? how about the services I already have running there? re-deploy them? [20:43] well, to be fair, I only one service running on that env, it's the MAAS jenkins instance [20:44] but even so I wasn't expecting that a juju version change would require me to re-deploy my services :-( [20:44] should I file a bug about this? [20:45] SpamapS niemeyer: I just noticed that things are aliased to lp:~charmers/charms/precise//precise and not /trunk... is that going to cause problems? [20:45] SpamapS, I can get access to the ec2 instance by ssh directly to its public dns address. Would that help? [20:46] matsubara: I don't believe there's any migration code to morph your version 1 zk tree to version 2 [20:46] m_3: yes thats likely the problem, DOH [20:47] matsubara: we won't do that once the release is stable.. but during dev re-factoring ZK had to be done unfortunately [20:48] SpamapS: hah idea for charm school. How about towards the end we let people who set up their laptops deploy to the cloud right there [20:48] SpamapS: like, it's one thing for you to be deploying to it, it'd be sweet if people from the audience can just deploy something [20:48] jcastro: yeah that should work [20:49] if len(u) < 5 || u[1] != "charms" || u[4] != "trunk" || len(u[0]) == 0 || u[0][0] != '~' { [20:50] return "", nil, fmt.Errorf("unwanted branch name: %s", name) [20:50] m_3: curses [20:50] } [20:50] m_3: we'll need to have a special argument of some kind for branch-distro, or update the charm store. [20:51] silly launchpad.. why do pushes store in 'trunk' but branch-distro stores in $series? :-P [20:51] SpamapS: we could move them again to charms/precise//trunk right? [20:51] I just re-promulgated statusnet as 'trunk' [20:52] waiting to see if it appears as cs:precise/statusnet [20:52] we really could do it manually if necessary... the hard part is the aliases [20:53]