[00:00] SpamapS: btw , you have to really strech it to say omg isnt using juju anymore :) heh i posted that showing the fact it was using 3 suborniates, only one other exists in the store :) [00:00] SpamapS: yea but dident it just use the service name not the proto [00:01] e.g. it sees mysql and local:mysql and cs:precise/mysql all as the mysql service [00:01] I did not say they'r enot using juju [00:01] they're using juju BEAUTIFULLY [00:01] I'm saying they're not using the charm store [00:01] SpamapS: oh for sure, but its the FIRST, and also still very much down in my spare time [00:01] imbrandon: the service name? no. It uses the charm name to figure out if there is a new charm to upgrade to [00:02] so there are things to improve etc etc for sure, but also try to not cowboy stuff either :) [00:02] imbrandon: yeah don't sweat it. I'm frustrated with us for not making this easier, not you for not contributing more. ;) [00:02] oh yea, i dident mean it like that eihter [00:02] kinda came out wrong :) [00:03] but ya know what i mean, and yea juju is still very veyr young too, as well as this is the first one, trust me it could be ALOT worse [00:03] i think considering those two things its fantasitic [00:03] personally :) [00:09] SpamapS: omg has picked up quite a bit of traffic too since we all started [00:09] a few months ago on it, like almost doubled actually [00:09] sustained [00:09] imbrandon: I'm not surprised. Its *WAY* faster now. [00:09] that matters to people [00:09] as google has shown [00:09] like the surge from 12.04 never backed back off [00:10] and its even increaced a bit [00:10] :) [00:11] SpamapS: got him on with that new advertiser too [00:11] http://bsa.ly/k2s [00:11] only a few days ago though so not much inventory yet, and the impressions is aoff [00:11] off* [00:12] but still impressive, and may make him some more money :) [00:12] I like the whole MaaS but I keep running into nodes getting stuck in comissioning [00:12] even when their clocks are sync'd [00:12] so I'm tryign to run initial trials with a bunch of small desktops before I go and rebuild my cluster [00:14] but once I get that ironed out and the comissioning rock solid, this will definately be handy [00:18] :) [00:18] EvilMog: thats the plan [01:38] woot, blitz.io t-shirt showed up today [01:38] :) [01:38] more swag heh === m_3_ is now known as m_3-backstage [15:13] ok, really appreciating that I can run mixed series environments right now [15:13] byobu-classroom internals work a little better with the default settings from oneiric [15:14] guess maybe that was the switchover from byobuscreen to byobumux [15:14] really would like to just have tmux-classroom at this point [15:15] * m_3 fingers relearning split adjustments for the irc talk [15:26] oh cruft, i got that class today [15:27] * imbrandon goes to polish the presentation [17:02] morning [17:03] I have juju set up on osx and I've configured the ec2 variables for my openstack install (type: ec2), but I"m getting connection errors. (euca2ools works with the same configuration) [17:04] I used this as a reference for the environments.yml file: http://www.w3.org/TR/html4/strict.dtd [17:04] oops [17:04] http://askubuntu.com/questions/65364/how-can-i-configure-multiple-deployment-environments-for-juju [17:06] imbrandon: so `juju deploy cs:oneiric/byobu-classroom` works pretty well as long as there're only 20 or so users [17:07] imbrandon: it sets up an ajaxterm in ec2 with user/pass of guest/guest... precise's doesn't work though so you have to subsequently ssh from there to a precise instance if you're showing precise-specific cli stuff [17:07] * m_3 add a note to fix byobu-classroom to work with byobumux instead of byobuscreen [17:08] or actually just frickin write tmux-classroom [17:10] how do I enable debug logging in juju? [17:10] jkyle: `juju debug-log`? [17:10] nice, yea i was gonna ask you if that was charmed [17:10] brb afk [17:10] jkyle: did you get your multiple environments straightened out? [17:12] debug-log doesn't seem to do it [17:12] oddly [17:12] debug-log runs in a seperate term while you do the commands [17:12] keep it running [17:12] didn't know they were crooked [17:12] the issue is a failure to connect [17:12] jkyle: there're some bugs with what it logs too [17:13] I opened the connect.py at the error point and noticed it has decent logging to the debug facility [17:13] e.g. log.debug("Failed using url: {0}".format(url)) [17:13] jkyle: hmmm dunno [17:13] and I'd like to enable that [17:14] does your dns work on your openstack install? I know there've been issues with that in the past [17:14] this is a running install [17:14] has dozens of users and vm's on it [17:15] my endpoints are ip's, e.g. http://12.50.28.2:8773/services/Cloud [17:16] I've got bacon, egg_white_s, and raw spinach upstairs... the latter two are bacon-offset(TM) [17:16] jkyle: is your bootstrap node directly accessible from your client? (i.e., with a public address) [17:17] :) [17:18] jkyle: nothing works from the client if it can't talk directly to the bootstrap node... iirc, haven't used juju against openstack in a bit [17:18] * m_3 gonna go eat.. ttyl [17:19] juju bootstrap 1 ↵ [17:19] 2012-06-23 10:19:34,212 INFO Bootstrapping environment 'pao1' (type: ec2)... [17:19] Connection was refused by other side: 61: Connection refused. [17:20] can't create bootstrap node [17:20] like I said, failing to connect to the endpoints ;) [17:25] maybe it's a version thing, does juju only work with essex or something? [17:28] jkyle: hmmm it sounds like it's not reading your environment [17:28] `juju bootstrap -e` [17:28] or [17:28] ah, my other openstack install seems to behave more [17:29] nope, timed out. but not connection refused [17:29] ~/.juju/environments.yaml should have a top-level 'default: ' if you've got multiple environments [17:29] don't know where things are stored on the mac client [17:29] is it required that you have an s3 endpoint? [17:30] dunno for openstack... lemme see what I have [17:30] right, I have default:pao1 and it's using that definition [17:30] too bad juju doesn't have a nova provider yet. [17:31] ec2-uri, s3-uri, default-image-id are all there [17:31] pretty sure the first two are requied [17:31] jkyle: it's there just in review... not in trunk yet [17:31] yeah, looks like juju requires s3 [17:31] I'm not running an s3 service [17:31] there'll be a 'honolulu' release in a couple of weeks with that provider [17:32] oh [17:32] yeah, that's a problem [17:32] big problem, I don't think we're offering that on any of our clouds [17:33] there may be workarounds to get an s3-like service up (rados and stuff like that) [17:33] you can run rados from euca-tools, then use _that_ endpoint url for juju [17:33] haven't seen that actually working before though... just ideas thrown around [17:34] trick is to configure to accept the same creds as nova [17:34] why is s3 needed? [17:34] caching charms mostly [17:35] you want the set of charms you've deployed to be sort of frozen [17:35] then you can deploy additional units of a service using the _same_ version as the already-deployed ones [17:35] two months later [17:39] * m_3 gonna get rid of the classroom setup... brb [17:39] well, this is disappointing. [17:59] so heres my question, is there a way to force a username and password onto newly comissioned nodes so that I can log into them during their comissioning phase and sync their times and otherwise force them to complete comissioning? [18:02] EvilMog: would keys work? it's easy to have juju inject keys into every instance's startup... environments.yaml there's a per-environment key:value for authorized-keys (http://paste.ubuntu.com/1056340/ indent is important in yaml heredocs) [18:03] as long as I can sudo, my big problem is I have 5 nodes stuck in comissionign and only 1 that made it to ready during the Ubuntu MaaS setup [18:04] so this is prior to juju being even installed [18:04] ah, dunno the maas ramifications sorry [18:08] EvilMog: maas is just a fancy frontend to pxe [18:09] yeah, but the image it pushes never fulyl syncs its nodes [18:09] syncs to its master [18:09] and sicne the image doesn't have a username and password attached to it I can't go in and fix it [18:09] EvilMog: right, but you can modify the pxe preseed file to install pubkeys if you wish [18:09] ahh [18:09] or do any other kind of preconfiguraiotn [18:10] ntp, packages, netowkr config, etc [18:11] m_3: I think I'll give puppet a try, just looking for something other than chef [18:11] capify :) [18:11] * imbrandon hugs ruby [18:11] they serve different purposes, eh [18:11] I'm lukewarm about ruby hehe [18:11] well kinda, so do puppet and juju in that sense :) [18:12] language is cool, working in the environment is painful [18:12] i personally like to use juju for orchastration and fabric or capistrano for code deployment and config mgmt [18:12] togather [18:12] kinda, but I'd say puppet and juju are much closer. cap is just a app deploy tool pretty dialed in for ruby apps [18:13] imbrandon: I'd say that's typical [18:13] nah, i use it for php apps mostly [18:13] ther are alot of plugins for many langs, even drupal and wordpress specific ones [18:13] etc [18:13] my condolences ^_^ [18:13] ( in cap ) [18:13] haha , i actually _like_ php :) [18:14] color me perplexed :P [18:14] good php is a beuitiful thing, there is alot of bad out there, but there is bad js and python too [18:14] :) [18:14] php is easy, good php is hard, but good php is rockin, and it powers 77.9 % of the net :) [18:14] php drives me mad, it's like they just took a bunch of ideas and threw them in a bag and shook them up [18:15] no consistency in language design [18:15] docs are great though [18:15] sure but same thing can be said about perl too [18:15] :) [18:15] perl? what's that? [18:15] ./ducks [18:15] and perl is the foundations of ubuntu and debian :) [18:15] as much as people want to think python is [18:15] hehe [18:15] i hear ya [18:15] perl is an old gray beard for sure...not sure I'd say it's as bad as php though [18:16] nope, your right, its worse :) [18:16] * m_3 goes to watch tv now that the discussion has turned to religion :) [18:16] hehe [18:16] it just dont get as much press since its not powering 77% or more of the net [18:16] I see perl as much closer to ruby than php [18:16] or ruby close to perl, rather [18:16] php was built on perl :P all the way till version 3 [18:16] jej [18:17] heh* [18:17] version 3 introduced the compiled code / zend engine , before that php was a ton of perl functions [18:18] ( and a bit of C, nothing like today tho, its come a LONG way, even just in the last 3 years with 5.3+ ) [18:19] it got its big boy pants arround 5 or 5.1 when they started doing releases right, and ppl like me could stand behind it with a streight face :) [18:19] heh [18:20] ruby is awesome for one thing, and one thing only, but its a HUGE thing [18:20] ruby EVERYTHING is an object , EVERYTHING [18:21] and THAT is awesome, it has a TON of implecations most dont even consider [18:22] like ... "hi".repeat(4) class String def repeat self * N end ; puts "hi".repeat(4) >> "hihihihi" [18:22] thats psudo code, but very very close to exact [18:22] :) [18:23] imbrandon: in python as well [18:23] everything is an object in python [18:23] nah, i mean EVERYTHING is an object, not so in python [18:23] alot is [18:23] and its very OO [18:23] what's not an object in python [18:23] but not EVERY SINGLE THING [18:23] a string [18:24] a literal in ruby is a object [18:24] no extra code [18:24] etc, right off the bat [18:25] like [18:25] 5.*(10) [18:25] == 50 [18:25] the litteral 5 has a method [18:26] jkyle: check this [18:26] http://paste.ubuntu.com/1056377/ [18:26] strings are objects in python [18:27] i'm not saying that cant be done in python but its a diff beast [18:27] i was pretty sure python litterals and strings and such wernt in the same way [18:27] * imbrandon will dig into it tho [18:28] I like that functions are first class objects in python [18:28] python has the whitespace downfall tho :) [18:28] I like the whitespace :) [18:28] yea same in ruby [18:28] imbrandon: no, in ruby they are not [18:28] sure they are, and you even can mask them as builtins [18:28] ir they seem like it when used [18:29] imbrandon: no, methods in ruby cannot be assigned or passed as normal variables [18:29] sure they can, rbuy def has lambdas and the like [18:29] you have to use a proxy object, this is the purpose of Proc [18:30] def soemting do |func| [18:30] func blah blah [18:31] imbrandon: it's a side effect of ruby's being designed around messaging [18:32] def weightedknn(data, vec1, k = 5, weightf = :gaussian) ... weight = self.send(weightf) [18:32] imbrandon: pastie an example of directly assigning a function object to a variable [18:33] just use .send [18:33] send is aproc object [18:34] like I said, you can't assign a function to a variable [18:34] its just diuff synctax not any less able to pass functions [18:34] no, it's the definition of a first class object. [18:34] it must support the assignment operator [18:34] the variable its self is an object with functions too, its just syntax sugar [18:34] at that point [18:35] no, it's not syntactic sugar. it's a core language feature >.< it has significant implications in how the language can be used and how binding works [18:36] ohh crap, 30 minutes untill my talk, gotta start up VM;s and such, i'd love to finish this later with ya tho, but i do disagree that they arent first class, kinda have to be as objects [18:36] x`http://stackoverflow.com/questions/2602340/methods-in-ruby-objects-or-not [18:36] give me an hour to give this user days cli thing :) [18:37] * imbrandon will open the tab for later [18:37] that's not 100% accurate as they _are_ objects, just not first class objects [18:38] that's what's implied by "not objects like strings and numbers" [18:38] thats not entirely true either, look at :name [18:38] etv [18:38] but i really need to get started, sorry :( [18:38] heg [18:38] heh [18:39] otherwise i'll be mobbed in #ubuntu-classroom lol [18:40] m_3 is a nother ruby person i think, i dont claim to know ruby well, but i do like it quite a bit more than python ( but thats mostly due to my hatred of whitespace delemited things like yaml heh ) [18:41] but yea they are very very similar , in utility and placed to use them wisely [18:45] imbrandon: btw, you can monkey patch python like your example too. though not builtins, this is by design. effecting core classes at the global scope is the cause of many problmes with ruby [18:46] yea class String undef_method :blah [18:46] is fun :) [18:47] i'm thinking more like a = "blah" [1,2].each do |func| puts func.*a [18:47] kinda bastardized but yea [18:50] btw on the s3 thing, s3ql works great [18:50] for that [18:59] oh wow, i fail at reading the clock [18:59] i have one more hour :) [18:59] heh [19:00] jkyle: yea i used s3ql the frist time the other day as a single machine s3 replacement, seems to work ok for limited use things like that [19:00] its designed for like use ona dev machine to practice on the api etc [19:00] but works well for non-critical data like that [19:00] or seems to, not used it a ton yet [19:53] imbrandon: ping [19:53] yup [19:53] just about ready [19:54] running about 30 secoinds late [21:05] m_3: heh , left a lil toss out to ya at the very end of my cli session ( after explaining aliases a lil bit ) [21:05] 15:59:27 <+imbrandon> OK well with that, i'll leave you with this last thought [21:05] :) [21:05] 15:59:33 <+imbrandon> alias pushit='git push && afplay ~/Music/saltnpepa-pushit.mp3' [21:05] 15:59:50 <+imbrandon> ^ m_3 that ones just for you brother :) [21:06] * imbrandon goes to work on the nginx charm a little more, spent the first half of today on it, and will spend the last half too, it _IS_ getting to the review queue today ... $sometimes [21:06] s/$sometimes/$sometime [21:18] jkyle: i dug a little more into what you were saying about the ruby stuff, and yea your are exactly right , but really i dont see that as a drawback really as the alternatives or "work arounds" how ever ya wannna look at them are solid and such, e.g. it dont seem like a phpish cludge^Hhack [21:19] seems more like a codeing style type thing, like the diffrence between Pascal and Bordland Turbo Pascal :) hahahaha [21:22] i mean hell , lamda functions are awesome and a staple in JS, server and client, and as close as syntax wise JS and PHP are ( scarry really ) php also has the same lambda and closure function ability ( its very awesome actually , dont know why its not used more ) but rarely used , esp in "production" code with the exception of Symphony2 and one other small framework i cant think of the name right off [21:22] but definatly not the norm [21:43] 2012-06-23 20:06:52,707 ERROR Cannot connect to environment: DNS lookup failed: address 'ec2.eu-west-1.amazonaws.com' not found: [Errno -2] Name or service not known. [21:43] fun [21:43] hazmat: can you reach your aws stuff ? [21:43] ( or anyone around )