[01:28] jcastro: :D who won the contest? [01:28] :D === al-maisan is now known as almaisan-away === almaisan-away is now known as al-maisan [13:27] niemeyer: are you accepting bugs on the store yet? Or is it still sort of in progress? [13:38] jcastro: Certainly accepting them [13:38] jcastro: What's up? [13:39] http://pastebin.ubuntu.com/913016/ [13:40] jcastro: Hmmm [13:40] the .charm file is 0 bytes [13:41] jcastro: Yeah, just noticed that.. very weird [13:41] jcastro: Theoretically it doesn't link the charm in the final location while it's not completed successfully [13:41] jcastro: Will investigate, thanks [13:49] lynxman: mira, want to do a charm review? [13:49] jcastro: throw it my way :) [13:49] https://bugs.launchpad.net/charms/+bug/961819 [13:49] <_mup_> Bug #961819: New "sbuild" charm for build environments < https://launchpad.net/bugs/961819 > [13:49] and https://bugs.launchpad.net/charms/+bug/956259 needs a final review [13:49] <_mup_> Bug #956259: Charm needed: znc < https://launchpad.net/bugs/956259 > [13:51] oh noes, party over, someone call the strippers and tell em not to show up, jono's here [13:51] lol [13:52] jono: on a real note i got some more css goodness for ya for accomplishments-system , just finishing up some of the tweaks now [13:52] some pictograms and such [13:52] lynxman: you have promulgation powers right? [13:53] imbrandon, oh thanks so much, pal! [13:53] :) [13:56] jcastro: I don't know if I do, if I don't I'll bug you ;) [13:57] I'll be around today to prog if needed [13:58] lynxman: ok when you finish these up ping me and marcoceppi [13:58] That's the fun part though, typing in promulgate and simultaneously hoping it works and wondering if there's a better word for that action [13:58] it'd be nice to ring the bell 2 times today [13:59] jcastro: sounds good :) [13:59] negronjl like nginx, we'll have him check the new drupal one. :) [14:00] marcoceppi: appflower needs a round #2 and it should be done [14:01] jcastro: perfect, if juju bootstraps then I'll give it a quick go [14:01] ditto for gitolite [14:01] * marcoceppi really wants to try out gitolite [14:01] this week gentlemen, we promulgate! [14:01] i would soo need that in bash completion [14:01] :) [14:01] Thankfully it is [14:02] prom [14:02] lol [14:03] jcastro: i dident mirror the drupal one to LP, let me know if i need to [14:04] jcastro: fix on the way [14:04] imbrandon: it needs to be on LP to be promulgated [14:04] kk, i'll do that now then [14:05] we'll need to figure out a way to autoimport github charms onto LP if we want that to scale. [14:05] history dont matter does it, i can just bzr init it clean right ? [14:05] but m_3 is at a conference so we can deal with it later. [14:05] yea i can help him with that too maybe a bit, for Penton I wrote the hg<-->git two way mirror for our repos [14:06] we used hg internaly but all contractors used git :) === al-maisan is now known as almaisan-away [14:06] git and hg shared hashes thou, not sure if bzr does [14:07] LP can already pull git repos [14:07] oh really ? [14:07] it can only mirror the master branch though [14:07] do i just need to import it ? [14:07] thats fine [14:07] http://askubuntu.com/questions/13613/git-on-launchpad [14:07] i forgot github has an svn interface too [14:08] So you can just push everything to github and every 4-6 hours it'll be imported on to LP [14:08] nice, thast perfect, and i work in named branches and merge to master as the "gold" copy anyhow [14:08] as most git ppl do [14:09] so works out perfect [14:12] Hum, deploying from the store gives me an error [14:12] I just told niemeyer, he's on it [14:12] ah, cool [14:13] something something needs to be a zip file? [14:13] marcoceppi: What's the error? [14:13] marcoceppi: Just to make sure we're talking about the same thing [14:13] jcastro, marcoceppi: Should be fixed in 5 minutes or so [14:13] i get the same thing jcastro did a bit ago [14:13] niemeyer: http://paste.ubuntu.com/913053/ [14:14] marcoceppi: That's it, thanks [14:14] I was getting all gutsy and trying to deploy from the store :) [14:14] Will be fixed in a few [14:14] marcoceppi: Sorry for undermining the excitement :-) [14:14] marcoceppi: HOld it for a moment, though! [14:14] :-) [14:14] heh, I was already cheering when I hit enter. [14:15] marcoceppi: do i need to make a project to import from another vcs ? [14:15] imbrandon: I have no idea [14:15] Never actually used it [14:15] just push it by hand for now [14:15] LP ui really does suck, i've been using it for 6 years now and still dont "get it" [14:16] we can sort the automated bits later [14:16] kk [14:17] i just thouht about that, 6 years, wow, its been a long time, wonder where that whiprush dude i used to give hell to is nowa days [14:18] imbrandon: less talk, more pushing. :) [14:18] its pushing [14:18] we're running out of days! [14:18] shish [14:18] bzr = slow [14:18] There's your whiprush and whip crack! [14:19] hahah [14:19] ahh crap its slow cuz its pushing .git [14:19] ctl+c [14:19] ugh [14:24] ok done [14:24] and linked to bug [14:25] Amd I weird because I like LP's UI? (except blueprints) [14:25] heh [14:25] I find the cross-project-distro bug tracking to be supremely useful [14:26] i dont think its ugly, ui was probably the bad word, i dont get most of the workflows [14:26] and linking branches/merge proposals to bugs is really nice. [14:26] imbrandon: add a bug.. toggle a status... whats not to get? [14:26] SpamapS: yea most modern trackers do that, gitosis , github, bitbucket, and i'm sure others :) [14:26] they're all *really* short [14:27] SpamapS: the basics sure [14:27] and please, don't bring up github's *ridiculous* issue tracker [14:27] I'm coming around to git.. but that thing is a joke. [14:27] heh, i love it personaly ;) [14:27] to each their own i guess on that one [14:27] heh [14:28] its great for small projects I'm sure [14:28] Charms still require two reviewers, right? [14:28] marcoceppi: *no* [14:28] never have [14:28] *what* [14:28] marcoceppi: ~charmers needs 2 +1's [14:28] marcoceppi: but if you are in ~charmers, you have the power [14:28] SpamapS: rails isnt what id call a small project nor bootstrap, or h5bp [14:28] Why have I always thought that charms require two reviewers [14:28] * SpamapS hears "YOU GOT THE TOUCH" in his mind and starts googling for leather pants prices [14:28] we've been doing 2 reviewers for the sake of completeness, I think people just got into the habit of doublechecking each other's work [14:29] haha [14:29] which is fine by me also [14:29] imbrandon: I suppose by having no features.. it forces developers to just focus on each issue rather than being able to prioritize stuff away [14:30] ok going to get some food and shower/shave , back in 20 min for critique if there is any by that time :) [14:30] SpamapS: yea, sometimes too much optiions is not a good thing [14:30] Cool, well looks like appflower is ready then! [14:30] SpamapS: and you can prio somewhat [14:31] anyhow brb food time [14:31] imbrandon: though many would call LP's features pretty minimalistic when compard to bugzilla [14:31] very very true, that thing is horrific [14:33] but really its not the number of features about LP , nor the ui specificly , i mis spke at first, but more the workflow as in sometimes , hell all the time its not clean on what needs to be done or if something does or how to do it, or click then then that to do this, FSK!!! give me ONE "FIXED!" button [14:35] Found a but with charm proof [14:35] marcoceppi, imbrandon, jcastro: It's fixed [14:36] party time [14:36] niemeyer: rockin ty [14:36] SpamapS: but on the other hand i'm a huge fan of minimalitic UI's too, sometimes to a fault [14:36] ok really gone for food, brb [14:36] ok but we need to wait for it to be deployed? I'm not familiar with the store workflow [14:37] jcastro: it worked for me, just delete the file in ~/.juju/cache/ [14:38] Have we fixed the mysql charm in the charm store yet? [14:38] It was broken as of Friday [14:38] Would a charm having an optional config option with no config-changed hook be a blocker? [14:39] Or a fix after prog situation? [14:41] niemeyer: this is great, I've deployed a bunch of stuff from branches too [14:42] Yeah, charm store rules already [14:42] niemeyer: so "source" and "publish" are 12.10 targets then? [14:42] so much faster [14:42] marcoceppi: all configs are optional.. I think its ok to have configs that only get applied in install as long as the config.yaml says so in the option's description. [14:42] also, I already want "juju search blah" [14:42] SpamapS: well it doesn't get applied anywhere :) [14:42] I think it was a future feature half added [14:42] anywho, I'll mention it in the branch and promulgate [14:42] marcoceppi: thats just a bug. I wouldn't block promulgation on that if it were the only thing wrong. [14:43] * marcoceppi *nod* [14:43] just file a bug for it on the spot so we don't forget [14:43] blockers are "does nothing", "explodifies", or "pwns you" [14:43] cool, appflower is ready. about to promulgate [14:44] its the app FLOW-er [14:44] endorsed by Flo-rida [14:44] jcastro: Yeah [14:44] jcastro: I intend to provide some helpful API endpoints well before that, though [14:44] jcastro: The first thing I want to expose is the error messages [14:45] jcastro: Every imported charm already comes with status [14:45] jcastro: We should have a way to show the import result, so that any failures may be known by the author [14:47] I just realized the store spec doesn't mention search [14:47] should I file a bug on that? [14:47] basically like how apt does it [14:47] jcastro: Yeah, please feel free to file it [14:48] jcastro: It'll certainly come [14:48] SpamapS: getting an error on promulgation "ERROR:Branch has not been pushed." with nothing else [14:49] marcoceppi: promulgate uses the remembered push location [14:49] marcoceppi: bzr push lp:~charmers/charms/appflower/trunk && charm promulgate should work [14:49] Cool, thanks [14:49] zomg jcastro SpamapS marcoceppi , yalll cant miss this http://jujucharms.deviantart.com/ [14:50] hit #3 on google [14:51] maybe we can get the artist to make us some keychains for promulgatee's [14:51] heh [14:53] <_mup_> Bug #972515 was filed: Charm store needs search functionality < https://launchpad.net/bugs/972515 > [15:04] SpamapS: did you promulgate Subway? [15:06] jcastro: yes [15:07] ok anything else on Friday? I need to chase the people down for swag remember [15:12] jcastro: code.launchpad.net/charms .. you can see all the charms that have been pushed/promulgated recently [15:12] jcastro: "Mature" means promulgated :) === tobin is now known as Guest31547 [15:47] the old way of deploying charms doesn't work anymore? [15:47] 2012-04-03 11:44:15,105 ERROR Bad charm URL 'local:precise:precise/splice-glance-keystone-mysql-rabbitmq/': no series specified (URL inferred from 'local:precise:precise/splice-glance-keystone-mysql-rabbitmq/') [15:48] sorry, that's the error: 2012-04-03 11:43:30,886 ERROR Bad charm URL 'local:precise/splice-glance-keystone-mysql-rabbitmq/': no series specified (URL inferred from 'local:precise/splice-glance-keystone-mysql-rabbitmq/') [15:48] I have default-series as well in my environments.yaml [15:50] lynxman: repository should still work the same [15:51] SpamapS: yeah I don't need to specify the series in the repo url anymore, it's wrong in the docs in juju.ubuntu.com as well [15:51] lynxman: you never had to specify series in the repo url [15:51] ?? [15:51] SpamapS: you had to at some point [15:51] SpamapS: and it still says so in the docs [15:52] SpamapS: https://juju.ubuntu.com/docs/user-tutorial.html#deploying-service-units [15:52] SpamapS: see the "oneiric/somecharmname" in the juju deploy line [15:58] lynxman: I've been using 'local:foo' for a long time.. but I agree the docs is not awesome [15:58] SpamapS: I always was using the doc method, doesn't work anymore :) [16:00] SpamapS: these docs are all in juju/docs right? [16:01] lp:juju/docs is where juju.ubuntu.com/docs is generated [16:34] marcoceppi: are you on precise? Can you try to deploy zookeeper from the store? [16:35] Yea, let me fire up my laptop [16:36] deploy zookeper? [16:36] yeah [16:36] 2012-04-03 12:36:39,085 ERROR Error processing 'cs:precise/zookeeper': entry not found [16:36] is what I get [16:37] Oh, but you should be able to do that from an oneiric machine, so long as you set the series up correctly. I'll try anyways [16:37] I'm all precise right now [16:37] but if you want to test it on oneiric if you have it handy that would be useful too [16:38] I get the same error on my precise laptop [16:39] Maybe only oneiric series is in the store atm? [16:39] SpamapS: is that because we haven't done something in launchpad for the precise series or is it a bug in the store? [16:39] marcoceppi: there's a precise version in the charm browser, not sure how that relates to what's in the actual store [16:40] I'm not sure either [16:43] marcoceppi: hey Marco, did you comment on the kusabax bug with your feedback? :) [16:44] shazzner: ah, I don't think I submitted it yet. Let me get back to you after lunch [16:45] marcoceppi: gratci :) [17:48] Yes only oneiric is in the store right now... [17:48] but I don't think that should be the case [18:06] jcastro: How's the store coming? [18:07] jcastro: Is it working well now? [18:07] yep [18:07] I've deployed a bunch of things [18:07] jcastro: Super [18:24] niemeyer: yeah, the charm store is working beautifully. :) [18:25] though we still need a switch-charm command so people can fix things without re-deploying [18:47] SpamapS: it also just made it painfully obvious that we don't have many precise charms, heh [18:53] I may have found a bug with boolean configs: http://paste.ubuntu.com/913459/ [18:54] Am I missing something or should this be filed? [19:04] marcoceppi: hey check this out: https://help.ubuntu.com/community/EC2-VNC [19:04] I was thinking this might be a cute charm [19:05] smells like a charm [19:05] I think it would be cool to have config options to be desktops [19:05] so you could test derivatives, etc. [19:06] problem with that, is you couldn't just sudo apt-get remove *buntu-desktop since meta packages don't remove their dependencies [19:06] Would have to find a way to cleanly remove those each time. Maybe use aptitude instead [19:06] ugh [19:06] <3 [19:06] #blameapt [19:07] Would be a cool charm though [19:07] What would you call it? [19:07] hah "juju deploy ubuntu" [19:07] perfect! [20:25] marcoceppi: I do believe that is a (trivial to fix, but extremely critical) bug in juju set [20:25] SpamapS: thanks, I'll open a bug report [20:26] SpamapS: you in distro mode or charm mode? [20:26] daddy wants promulgation [20:27] jcastro: I'm in "recover from 3 days away from email" mode, but I can be interrupted to do important things. :) [20:27] https://bugs.launchpad.net/charms/+bug/912050 [20:27] <_mup_> Bug #912050: Charm Needed: OpenERP < https://launchpad.net/bugs/912050 > [20:28] and maybe https://bugs.launchpad.net/charms/+bug/906176 ? [20:28] <_mup_> Bug #906176: Gitolite charm < https://launchpad.net/bugs/906176 > [20:28] I can grab one of thoes [20:28] oh both [20:28] or* [20:28] ok [20:29] didn't I give you two today already? [20:29] I did only one :) [20:29] I don't recall a second one [20:29] I already poked at openerp so I will focus on that one [20:29] https://bugs.launchpad.net/charms/+bug/964936 [20:29] <_mup_> Bug #964936: Drupal Charm: superchared Drupal charm with nginx,, apc, php-fpm, all setup to scale to the moon and be Best Practices. < https://launchpad.net/bugs/964936 > [20:29] marcoceppi: how about that one? [20:29] Oh, I haven't looked at that one yet [20:30] let me grab it while I wait for imbrandon to wake up [20:30] Nick has a semi-abandoned drupal charm already [20:31] so depending on what you think maybe this can just supercede it? [20:31] * SpamapS votes for it to supersede the others that weren't done by people who make drupal sit up, roll over, and beg like imbrandon does ;) [20:32] marcoceppi: I've got some other merge proposals too if you want to clean those up [20:32] jcastro: yeah, I saw those, shouldn't take too much time to get them updated [20:32] SpamapS jcastro so if this checks out, just charms/drupal ? [20:35] marcoceppi: yes [20:35] that should always be the most current drupal [20:35] sounds good [20:37] SpamapS: https://bugs.launchpad.net/charms/+bug/903361/comments/3 [20:37] <_mup_> Bug #903361: Charm needed: Alice IRC < https://launchpad.net/bugs/903361 > [20:37] Should the open-port be in the install hook or the start hook? [20:37] I think start because you don't want to open the port until the service is started right? === ajmitch_ is now known as ajmitch [20:40] marcoceppi: alice irc is ready. :) [20:40] sweet! [20:41] I'm going to just go in to charm mode for the rest of the evening since I'm done at work [20:41] then switch to omg tonight [20:42] marcoceppi: lynxman snagged znc and sbuild, but he's slow, we could do those today too [20:43] https://bugs.launchpad.net/charms/+bug/966484 [20:43] <_mup_> Bug #966484: New Charm: Kusaba X < https://launchpad.net/bugs/966484 > [20:43] is also another shazzner production [20:46] yeah, I've got a review in the draft for that [20:47] <_mup_> Bug #972829 was filed: maas-server setting requires port even when using http default < https://launchpad.net/bugs/972829 > [21:13] SpamapS: ok so even if we didn't make the precise store "open" yet, shouldn't "juju deploy precise/zookeeper" work? [21:14] jcastro: Are there any blog/G+/whatever posts about the store yet? [21:15] niemeyer: I have a draft, but my problem is I can't seem to deploy any precise charms [21:15] jcastro: Was just thinking about mentioning it online, but it'd be good to point it to something that is not an https API endpoint :-) [21:15] jcastro: What happens? [21:15] Error processing 'cs:precise/zookeeper': entry not found [21:15] but I don't know if this is a store bug or a charms bug [21:16] jcastro: Hmm.. maybe it doesn't actually exist in the store? [21:16] niemeyer: I have an epic blog post ready that I have been working on, I just need to make sure my examples work [21:16] http://jujucharms.com/charms/precise/zookeeper [21:17] the problem is that we don't have many precise charms yet [21:17] SpamapS has some plan for this [21:17] right, but the ones that are in there I can't seem to deploy [21:18] ah [21:18] yeah, that's not my main concern right now [21:18] we'll just copy the oneiric ones over, run the charm tester, and see what breaks [21:18] jcastro: Uh.. there are apparently *no* official precise charms [21:19] jcastro: Yeah, that's why I was against having that web site as it is in the first place, but that's history [21:19] jcastro: what niemeyer said [21:19] https://code.launchpad.net/charms/precise [21:20] oh ok, I can file a bug on that [21:20] has no lp:charms/appflower for example [21:20] like https://code.launchpad.net/charms/oneiric does [21:21] hmm ok [21:21] so ... either wait for the grand master copy over to precise, then mention the store [21:21] jcastro: that's what SpamapS was going to sort out...I believe [21:21] or I can rejig the announcement to talk about oneiric charms [21:21] I like the latter option, although it takes more effort :/ [21:21] and then say something like "the store won't switch over until precise releases, but you get the idea." [21:21] ack [21:22] robbiew: the post is more about policy than actual commands [21:22] cool...then I say sooner is better [21:22] I even say "this command doesn't exist yet, but you get the idea." [21:22] ok, on it. [21:22] give me 15. [21:22] jcastro: Yeah, +1 on going with Oneiric for the post [21:22] jcastro: Otherwise you'll be waiting on an unbound timeframe [21:23] that's should be motto of this cycle [21:23] juju deploy unbound-timeframes [21:23] \o/ [21:31] That ec2-vnc thing didn't quite work for me :/ [21:42] jcastro, while that charm exists for precise, it hasn't been promulgated such that its the branch tip [21:43] jcastro, the charm browser was written when that notion was in its infancy, so it takes the expedient approach of if the branch name matches, show it for that series. [21:45] because its clearly at the official branch namespace for that package, even if it hasn't been promulgated as the official branch tip yet. [21:45] jcastro: all your readmes are up to date [21:45] marcoceppi: thanks! [21:45] hazmat: thanks for clearing that up [21:46] jcastro: you said Alice IRC was ready for review+prom? [21:46] yessir [21:46] other than the location of the expose [21:46] which we wanted clarification on, but I don't think it's a deal breaker [21:46] naw, that can be a patch later [21:47] marcoceppi: next time shazzner and SpamapS are around I'd like to talk about putting him in ~charmers [21:47] he's got enough under his belt soon [21:47] same with patrick-hetu I think [21:48] 07 [21:53] marcoceppi: thanks for your feedback, I'll get to work on it! :) [21:54] shazzner: thanks for the charm! This is going to be perfect for the office :) [21:55] jcastro, niemeyer its a significant problem that the workflow for the charm store, is rather non obvious for a charm contributor of 206 charms on launchpad, 146 aren't properly marked for a series. [21:55] got to run to the hackerspace but I'll get the changes in tonight [21:55] jcastro: I need to clarify one thing with the website-relation-joined hook, I don't think there's a `unit-get private-address` can someone confirm that it exists? === shazzner is now known as shazzner-away [21:55] that whole part confuses me [21:56] I think there's only public-address and for the private address people have been using `hostname -f` [21:57] From my understanding you can only get unit data that's shown from the status, so under a typical unit you have agent-state, machine, and public-address [21:57] hazmat: A charm contributor doesn't have to be aware of that workflow at first.. SpamapS, m_3_, and jcastro have to be aware of it [21:57] it's not ideal, it involves me hand checking all the time, but we can discuss that at UDS [21:57] hazmat: So far there was very little reason to follow the arguably boring procedure [21:57] hazmat: Now there's a good one [21:57] it certainly won't scale, heh [21:58] niemeyer its more than that, and this applies even to a ppa charms, which won't be curated [21:58] marcoceppi, 'private-address' works as well [21:59] hazmat: What's the "more than that" in this case? [21:59] hazmat: Thanks, should older charms be updated to use private-address instead of hostname -f ? [21:59] niemeyer, its more than just the curated collection of the official distribution, being manually managed [21:59] niemeyer, ie. it definitely applies to charm contributors [22:00] marcoceppi, yes [22:00] hazmat: Sorry, I don't understand what you're trying to say.. [22:01] hazmat: Those "it" are missing context.. it is more.. it definitely applies.. etc [22:01] niemeyer, 'its' a significant problem that the workflow for the charm store, is rather non obvious for a charm contributor of 206 charms on launchpad, 146 aren't properly marked for a series. [22:02] hazmat: Do we have a single person that contributed 206 charms!? [22:02] shazzner is well on his way, heh [22:03] niemeyer, we have 59 contributors.. most of whom haven't gotten the process right such that they have a charm in the charm store. [22:03] hazmat: Why!? All of them have a charm in the charm store, unless their charms are broken [22:04] niemeyer, how many charms in the charm store? [22:04] hazmat: mthaddon is the one with the numbers, and he's asleep by now [22:05] hazmat: You can find out-of-date numbers in the email I sent to the list some time ago, though [22:05] niemeyer, most of the charms pushed to lp at a correct address don't have series returned by getbranchtips [22:05] niemeyer, i have the current numbers.. [22:05] hazmat: Branches don't need a series to be in the charm store [22:06] niemeyer, then why aren't the precise charms there? [22:06] hazmat: That is, they don't nee to be marked as a series [22:06] hazmat: They are there.. they are just no under cs:precise/* namespace, because they've not been blessed by SpamapS/m_3 [22:10] hazmat: % wget --quiet -O- https://store.juju.ubuntu.com/charm/~shazzner/precise/kusabax | wc -c [22:10] 5646 [22:10] niemeyer, yeah. saw that one [22:10] hazmat: Well, what are you bring up then? [22:10] bringing [22:10] niemeyer, so its only the official ones that need to be marked/promulgaged against the series [22:10] ? [22:11] hazmat: Yes, it's only the official ones that have to be made official. (!) [22:14] ic [22:14] so these charms at the official branch locations which haven't been marked as tip.. http://paste.ubuntu.com/913770/ [22:14] jcastro: sbuild charm looks good, don't have powers though :) [22:14] er. promulgated [22:14] ^^ marcoceppi want to promulgate that badboy? [22:15] yeah! [22:15] sup jcastro [22:15] heya bkerensa [22:15] store is live, your charm is in it brother! [22:15] http://www.jorgecastro.org/2012/04/03/the-juju-charm-store-will-change-the-way-you-use-ubuntu-server/ [22:15] ^^^ I need help spreading the word on this [22:17] jcastro: I will push it on the social webz [22:17] jcastro: any news on contest? :P [22:18] tomorrow-ish [22:18] we've published like 3-4 today already [22:18] jcastro, nice blog post [22:18] hazmat: thanks! [22:20] http://www.reddit.com/r/linux/comments/rrvec/why_the_juju_charm_store_will_change_the_way_you/ [22:20] upvotes please! [22:20] http://www.reddit.com/r/Ubuntu/comments/rruma/why_the_juju_charm_store_will_change_the_way_you/ [22:20] I have it there as well [22:21] marcoceppi: alice redirect fixed and pushed. [22:21] jcastro: interesting article, I might have to become interested in this juju stuff at some point [22:21] jcastro: sweet, just about to prom sbuild [22:21] jcastro: ok enjoy traffic (Reddit, Stumbled, Facebook, Twitter, G+, Hacker News etc) done [22:21] AlanBell: almost as if I planned it all along [22:22] bkerensa: where's the HN link, HN I've never been able to get onto [22:22] jcastro: http://news.ycombinator.com/item?id=3795097 [22:24] bkerensa: I don't think HN will roll to a reddit link [22:24] is there a way to edit it to point to the canonical url? [22:24] hah, I made a pun [22:24] jcastro uhh oh [22:25] jcastro: fixed here http://news.ycombinator.com/item?id=3795109 [22:26] jcastro: yeah if you can ship that stuff it would be better... my fiancee wants me to pickup tourist swag on my trip home from UDS... and unfortunately Ill be stuck in Oakland for a entire day after UDS [22:26] :D [22:26] so shopping I must [22:26] no worries, I have a juju shirt with your name all over it bud [22:26] k [22:26] jcastro: well the cup was what I was hoping for :P but ok [22:26] :D [22:27] how well does juju work with lcx containers? [22:27] AlanBell: fine [22:27] AlanBell: You can use lxc for testing charms [22:27] and you do that on your own computer right? [22:27] yeah [22:28] could someone host a heap of computers running lcx for juju? [22:28] so kind of an amazon cloud that scales down small [22:28] with a different cost model [22:28] AlanBell: yeah so what you really want, is an "ssh provider" that ssh'es into your linode and does all the magic [22:28] I wouldn't use LXC for production, to be honest [22:28] jcastro: I am thinking of being such a provider [22:29] AlanBell: You could deploy Open Stack if you wanted to go *that* route [22:29] jcastro: "since we haven’t deployed the charms to Precise yet" [22:30] jcastro: "deployed" may be confusing there [22:30] "opened" perhaps? [22:30] marcoceppi: possibly, but I don't think juju instances need the hard isolation from each other of virtual machines, or the preallocation of memory [22:32] jcastro: yeah, or "moved" or similar [22:32] sbuild is promulgated [22:32] jcastro: Why didnt you do the talk at PuppetConf? Adam Gandelman and Marc Cluet were talking juju? [22:36] alice-irc promulgated [22:37] bkerensa: I don't think I was working on juju that much yet, we spread the load. [22:37] bkerensa: but no worries, I'll catch Luke and Co. at OSCON [22:38] marcoceppi: \o/ [22:39] * marcoceppi heads home. [22:39] Catch you guys in a bit [23:09] AlanBell: what you probably want is openstack + lxc [23:12] maybe, I will have to look into it [23:12] I think that there might be a way to do a juju specific cloud provider that doesn't bill by the number of instance-seconds [23:13] AlanBell: juju is more about using an API to manage compute resources. What you are talking about is more having compute resources and making it available via an API.. which is what openstack does. :) [23:13] sure, but providing it in a way that works well for juju [23:13] economically as well as technically [23:14] AlanBell: frankly, LXC is probably *not* ready for a business to be built on top of it, and you are probably going to spend less on the slight inefficiency of kvm than you will bringing it up to snuff. [23:14] AlanBell: LXC is going there.. but its not there yet. :) [23:15] maybe openstack is the way to do it [23:16] but you have to define your instance sizes ahead of time, I think with LCX you don't [23:18] I kind of want to say "you have 4GB of ram, one public IPv4 address, an IPv6 subnet, it is all set up for juju, spin up as many instances as you like within that" for a fixed monthly hosting fee [23:21] AlanBell: LXC isn't an "instance", its a container [23:21] yeah, juju services then [23:22] AlanBell: sorry, but why would I care about ipv6 subnets? I want availability, computing resources, RAM, disk IO.... not ips. [23:22] well they have to talk to each other [23:22] 127.0.0.1 works fine if its all on one machine. :) [23:23] oh I guess that would work with lcx [23:23] I want to understand what you're getting at. [23:23] Do you want to use juju to setup a whole bunch of stuff on one physical machine? [23:23] juju uses loads of machines, and amazon bills by the second [23:24] or hour or whatever [23:24] thats only because we haven't been clever yet. [23:24] jcastro set up a wordpress blog that cost $85 per day in hosting fees [23:25] AlanBell: and now its down to quite a bit lower than that.. I think maybe $2 or $3 / day [23:25] AlanBell: because we got a lot more clever. :) [23:25] AlanBell: and thats a blog with > 1 million hits per day, btw [23:25] sure [23:26] which is now elastically scalable with one command.. 'juju add-unit omg-wp' [23:26] it is a fairly big wordpress install, and a lot was learned [23:26] but still to get wordpress up you need a server for mysql and a server for apache [23:27] AlanBell: I know what you're getting at. Allowing people to over-subscribe machines (or just use bigger machines and higher density) is definitely on the radar. [23:27] and a server for the controller and a server for a loadbalancer if you want to scale it [23:28] AlanBell: except.. we moved the load balancer onto the omg-wp nodes [23:28] I don't think it is a massive technical problem, I just think the amazon model of charging by instance-hour doesn't really work so well for juju [23:28] AlanBell: it works great for juju. :) [23:28] works great for amazon too :) [23:29] AlanBell: the ZK node can be a t1.micro [23:29] bet they love it [23:29] oh I thought all the juju instances had to be the same size? [23:29] AlanBell: the only situation where it doesn't work well is for sites with no actual traffic. [23:30] AlanBell: no, you can (painfully) change your instance type between deploys. [23:30] yeah, "no actual traffic" is where every site starts out though [23:30] AlanBell: all the instance types for one given service do currently need to be the same. [23:31] so what you need is a smooth migration path from "no actual traffic" to "argh, spin up more instances" [23:31] AlanBell: so they use t1.micros and as they get bigger, reboot them into m1.small's, then medium, and up and up.. ;) [23:31] t1.micro can be rebooted into any other instance size. [23:32] AlanBell: I do want to be able to put a few things on one instance though.. I'm with you there. [23:33] I would kind of like to start building stuff like openERP instances with juju, but no way am I going to tell a customer they need 3 servers from day 1 [23:33] plus another 3 for a test/dev environment [23:34] AlanBell: 3 t1.micros would be $51/month ... [23:34] they're not servers, they're VMs :) [23:35] we have a bunch of these http://www.hetzner.de/en/hosting/produkte_rootserver/ex5 which we divide up with KVM [23:36] AlanBell: right, so throw openstack in front of that and you have an API. :) [23:36] yeah, thats what I am thinking [23:37] AlanBell: and just chop them up into tiny bits like they do w/ t1.micro and you're set for juju users. :) [23:38] yeah, could go quite a bit smaller than the t1.micro [23:39] AlanBell: 600MB is pretty tiny :) [23:39] AlanBell: I suppose it would work for haproxy though. :) [23:40] oh, I thought the t1.micro was 1.7GB, is that the small one? [23:40] thats m1.small [23:40] t1.micro is borrowed CPU .. and something just under 700MB of RAM [23:40] ah right, 600mb is pretty small [23:41] yeah, not going to get much on that after the operating system [23:41] Yeah, 80MB for the kernel, 200+MB for in use VFS cache.. you basically just have enough for a single process low-memory type program. [23:42] and then you still have the issue that sometimes there just won't be any CPU for you [23:42] however the point of this thought experiment is to have lots of tiny VMs so that you can start your juju thing and do the horizontal scaling and then when you get big, move to Amazon [23:42] AlanBell: I *love* that idea. :) [23:42] hey guys [23:43] jcastro called me and asked if you can respond to questions on http://news.ycombinator.com/item?id=3795109 while he is at dinner [23:43] SpamapS, constraints land3ed [23:44] anyhow, night all o/ [23:48] SpamapS, ie. you can verify instances easily per service [23:49] arg.. vary [23:56] <_mup_> juju/relation-hook-context r520 committed by jim.baker@canonical.com [23:56] <_mup_> Better tests [23:57] <_mup_> juju/relation-hook-context r521 committed by jim.baker@canonical.com [23:57] <_mup_> Merged trunk