[00:24] AHHHHHHH [00:24] marcoceppi jcastro SpamapS , it has finally clicked [00:25] wow, i went about this totally wrong ( IMHO not really, still gonna be a great ) [00:25] imbrandon: yes.. Tyler Durdon *doesn't exist*.. well done ;) [00:25] so its too late to shift gears for the charm i got for the contest, like i said its gonna be good but [00:25] still [00:25] lol [00:25] SpamapS: hahah [00:26] SpamapS: yea, some reason and i dunno why it did not click before now [00:26] but i've been looking at juju as a gimped replacemtnt to my processes i'm used to [00:26] and its not [00:27] its the glue to bring them all in as one, not replace them [00:27] and THAT just clicked [00:27] imbrandon: *ding ding* :) [00:27] yea i've been like totally doing this wrong in my head [00:27] lol [00:27] imbrandon: I think thats in large part our fault, because we haven't written more charms using existing config management. [00:28] i was sitting head watching a old hangout with jono and sabdfl [00:28] and it just hit me when he said somehting [00:28] yea [00:28] it seems like they are all kinda reinventing the wheel [00:28] so it kinda feels that way [00:28] but no i mean i kinda came in the stuff heaqd first too and bass ackwards [00:29] to be honest untill OMG burned down and i got with jorge on that for the first time in 2 years , i had be steering away from juju as a fad [00:29] so its my faiult too [00:29] :) [00:30] my marcoceppi and you proved it was worth looking into imho, not that you had to nor was even trying i bet [00:30] but you konw [00:30] anyhow i feel like shit tons better now [00:31] i like had this guilt on WTF am i doing this stuff over, and over [00:31] like it aws a not invented here syndrome [00:31] but its now [00:31] not* [00:31] its just marketed wrong to the old timers :) [00:32] ok back to your regualrly scheduled shows, i'll probably have alot more relevant questions form here on i hope [00:32] lol [00:33] wow and i REALLY need to give in and get a new keyboard tomarrow [00:33] this typo shit ios for the birds [00:34] oops /sorry ubotu if your looking ( and anyone else, slip-up ) [00:35] SpamapS: i ahvent quite figured out how to NOT deploy to aws yet though, whats the quick and dirty to deploy local, or got a doc page , scanned the wiki but not long [00:35] like i got it working going to a aws now, etc :) kinda working backwards [00:36] lol [00:36] imbrandon: https://juju.ubuntu.com/docs/provider-configuration-local.html [00:36] rockin , ty [00:37] enjoy the weekend ? i think i spent mine infront of a console ( not that i havent done that 1000000000 times over the years , goes with the teratory hehe ) [00:37] Its pouring outside [00:37] cold [00:37] ouch [00:37] console is keeping me warm :) [00:37] what part of the globe ? [00:37] LA [00:37] ahhh [00:37] as in, Los Angeles [00:38] right right [00:38] i assumed [00:38] its one of our 9 rainy days a year [00:38] LOL [00:38] yellow rain ? [00:38] wife hates it because now I have an excuse to sit on my ass and play with code :) [00:38] LOL [00:39] when i was married *shhh not divorcded due to any of the geek stuff* hehe anyhow we had a deal, like before dinner was my geek time [00:39] after was our time ;) [00:39] no matter what day of the week [00:39] so it worked out pretty well [00:40] yeah thats mostly how it works here too [00:41] and thankfully , i;ve been fortunate enough to been high enough on the food chain not to pull pager duty since i was marrie ( and after ) [00:41] I kind of miss some aspects of pager duty [00:41] except for the rare occasion where the world really does blow up [00:42] yea i like some things about it, esp if it was for some of the companies i worked for later ( as in nowish ) but early on when i had to do it it sucked [00:42] Interesting.. this supply_drop thing is a bit like juju.. just defining puppet for each server, not having a centralized puppet master http://www.braintreepayments.com/devblog/decentralize-your-devops-with-masterless-puppet-and-supply-drop [00:42] and not the act but the job part ,had it been another firm prob would have been ok [00:42] * SpamapS hopes one day that juju becomes decentralized [00:43] ZK is cool but it feels like we could just move all that to each node's agent and use 0mq to pass data between nodes. [00:43] yea thats what i was wanting to hear earlier [00:43] when asking about it [00:43] if you got a sec look at the "brew" distro model [00:43] IMHO that would be PERFECT for charms [00:43] well I'm not talking about the repositories [00:43] infact it kinda is charms [00:43] only for osx [00:43] at a lower level, how its architecture actually works [00:43] all like apt-fast [00:43] I don't think we actually need machine 0 [00:43] or something [00:44] yea, the first itteration of things your never completely happy with [00:44] imbrandon: no, brew is just packaging. [00:44] no matter what it is [00:44] well kinda [00:44] it was [00:44] its grown to more [00:44] brew doesn't define how two services talk over the network does it? [00:44] it has taps now and recipies and binary and build distro [00:45] sure but it can via recepies [00:45] and some of the nodejs recepies do [00:45] and now that its distrubted taps i think more will pop up like that [00:46] i mean as for as technical a recipe and a install hook are about the same [00:46] would not take musch for the two to become feature parity [00:46] both ways [00:47] recipies are just make files btw, but they dont have to be "make" infact most are rake *.rb files [00:47] but php or sh or whatever like charms [00:47] so really you can make them do whatever [00:47] imbrandon: interesting.. so there's a uniform way to say "when you ask for X, make it talk to Y on another box this way" ? [00:47] its geared more to the desktop [00:47] but really the ideas are very very similar [00:48] imbrandon: because thats the thing I haven't seen any other system do yet. [00:48] many enable it [00:48] SpamapS: well almost , i dont think there is a formal api for te network part BUT the groundwork is ther and you CAN do it [00:48] but none of them make it a 1st class citizen [00:48] i just dont think its part of the apis [00:48] right [00:48] yeah you can do anything you do with juju with chef, puppet, cfengine, or your own shell scripts... [00:49] but they don't encapsulate you, so it makes portability a real problem [00:49] right , but i think brew is the closest to actually being the same . and half ass the same market, welll no but yes [00:49] * SpamapS looking at brew recipes now [00:49] well i notice brew more and more growing into that thinking though [00:50] and it would not take much for it to [00:50] imbrandon: I'm skeptical because I doubt OS X would have something for defining two way config channels, but perhaps the same result can come from some other means of encapsulation. [00:50] get all the way there VS say a ant script or chef [00:50] why do you say that ? [00:50] osx server does [00:50] like now [00:51] its alot more limited but it does do it in a limited scope and officially [00:51] imbrandon: did you see what I did with the peer relationship with omg? thats very easy in puppet or chef.. but there are so many ways to do it, you can't really *share* them outside an environment. [00:51] right [00:52] imbrandon: os x server has relations? [00:52] yea, see in that sense i can see brew adboping the stuff to do that where as chef wont, but osx server.app does now commercially [00:52] or rather, something like it? [00:52] yup [00:52] show me [00:52] yea forget what its called exactly [00:52] but like my mini here [00:52] I'm always looking for vindication of the juju idea... sometimes I wonder if we're not a little crazy with it :) [00:52] i install server.app and make it a server [00:52] then install any number of servers on the lan [00:53] it will auto pick them up and ask abourt config things like ldap slaves and myswl slave and mx host [00:53] and distcc [00:53] all predefined OR automagic [00:53] and will do it over wan too if predefined [00:54] nah [00:54] apple does it ( the juju idea ) [00:54] the thing about apple is [00:54] imbrandon: that rendevouz IIRC [00:54] its not just limited to their servers [00:54] its limited to a very small subset of theior apps on their servers [00:54] nah [00:54] much more than bonjour [00:54] not just discovery [00:55] i mean full on config [00:55] and preconfig [00:55] like the hooks [00:55] I want to see where the docs are for exposing and sharing this config data. It sounds fascinating. [00:55] like i can fire up server.app and tell it that the next 3 distcc nodes use blah.config [00:55] and then those when/if join will use it [00:55] etc [00:55] sure let me look the proper name up [00:56] imbrandon: I'm digging too. [00:56] it builds a ton on bonjour and ldap [00:56] but there is some propitary jiunk in there [00:56] junk* [00:56] i only have 10.8 installed for server.app and its very very very beta right now [00:57] so i only fire it up when needed [00:57] lol [00:58] can't find anything... apple has hidden this well ;) [00:58] one sec [00:58] and hell yea they hid all iinda shit well goto opensource.apple.com sometimer [00:58] heh [00:58] let me take a screenshot [00:58] one sec [00:58] its not the full story but will get ya looking [01:00] http://dl.dropbox.com/u/461492/10.8server.app.png [01:00] ok there we go [01:01] cant get it to load the profile manager there as i said 10.8 is still supper buggy as shit [01:01] but the profuile manage is not like ldap profile its app and server config profile [01:01] and more than like AD [01:01] cuz you dont have to join [01:02] and there is subgroups [01:02] etc [01:02] and the prepare there in the back [01:02] with the (1) alert indicator [01:02] on the apple configuratorr [01:02] thats the "make a charm" tool [01:03] but see those apps listed on the left , those are the only apps you can interact with it like that [01:03] its very limited [01:03] nmot open like juju or brew [01:04] and the supervise in the back is more like the AD user and machine policys [01:04] and the assign in the back is like assign corp assets to a employee thingy [01:04] but you can remote wipe and lock down if stolen etc [01:05] anyhow kinda pisses me off though 10.4, 10.5 and 10.6 server was the shit,but and 10.7 and 10.8 are really just an "app" [01:06] liek you have all the underlying stuff alreaddy insatlled and you get a 50mb gui app from the app store for 10.7 and 10.8 "server" [01:06] like atleast 10.6 even though apache and such is on 10.6 not ALL of the server apps are there [01:07] like BIND and iCal servers etc [01:07] but they are now, and you buy a management app [01:07] lol [01:08] but yea, they intelegently knwo about each other , and more than anycast dns ;) [01:08] i actually forgot the port but there is a dedicated mgmt port [01:08] on the lan they all talk over [01:08] as well as udp [01:09] ahh and see like pushing out the time machine configs [01:09] forgot about that too [01:10] and these arent AD boxes, like i said other macbook pros and minis on my lan [01:10] etc [01:10] not only just find the time machine, thats bonjour , but config the encryption passwd and times to run [01:11] etc [01:11] i thouhgt it was tied to my apple ID [01:11] but then i used like the new ability to install osx legally in ESX [01:12] and it did it to with no apple id [01:12] it may still be the icloud ports iirc now [01:12] but yea not account [01:12] and really even if it did thats ok [01:12] needs glue somewhere [01:13] might as well be my apple id, a.k.a launchpad id [01:14] SpamapS: you run apple hardware right, i'll make ya a deal, i'll try full ubnutu on this mini to try the 4x video and you can use my apple developer account to get a copy of server and try it out [01:14] i'm legally entitled to put it on 100 machines [01:15] if they are associated with me, and you work on the same OS i do so i;d say thats associated :) [01:15] just signup for an apple id if you dont have one ( same as itunes etc SSO ) and i can provision it to your account i am pretty sure for you to download [01:16] as well as IOS early etc [01:16] perk of my last job i got ios and mac os paid developers for the app stores :) [01:16] developer access* [01:17] anyhow, then you can see first hand, and as long as its on apple hardware it should install in vmware no problem [01:17] it checks a tpm chip or something for the hardware [01:17] but them moves right along as long as its 10.7 or 10.8 [01:18] oh on a ubuntun host [01:18] i ment to add [01:19] gah, spending way too much time on this, back to my charm [01:19] lol [01:20] gotta make the new hotness good, not the shiney current stuff [01:21] SpamapS: what about like say vsphere too, relation is kinda there , abeit via the central console [01:21] bah [01:22] back to juju [01:26] <_mup_> juju/unit-stop r425 committed by kapil.thangavelu@canonical.com [01:26] <_mup_> merge trunk [01:37] goal next week: clean up OMG PHP and JS code;Polish qickdrop drupal charm; and get juju tools working on OSX!! [01:37] gonna be a busy week, lol === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away [07:56] morning all === TheMue_ is now known as TheMue === almaisan-away is now known as al-maisan [08:41] goood morning juju beees [08:46] SpamapS: you know if nick is gonna finish his drupal charm or [08:46] would it be worth to try to tag him and snag it and merge what i have [08:46] mine is a toaly seperate name, but its not plain drupal either, and thats not used [08:46] its pressflow [08:46] hrm [08:47] yea i'm gonna keep em seperate and if nick dont finish his i'll juat fixish it too, let em co-exist [08:49] SpamapS: btw your sugestion for drush from the archive on his bug really isnt feasable, as its to old to work with the current drupal, but there is an alternative pear chan that is https [08:49] so that will work out [09:07] bbcmicrocomputer, looking at your hive charm now [09:12] jamespage: ok, thanks [09:13] jamespage: I did have a request for the Hive package you own [09:13] jamespage: can we get the HWI web interface included in that? [09:13] bbcmicrocomputer, I should think so yes [09:13] I had that on my list [09:13] jamespage: ah ok, great! [09:14] could you raised a bug at http://launchpad.net/hadoop-ubuntu-packages pls - that way I won't forget [09:14] jamespage: sure, np [09:14] bbcmicrocomputer, of course you could take a punt at it yourself [09:14] the packaging branches are under http://launchpad.net/~hadoop-ubuntu [09:15] bbcmicrocomputer, https://wiki.ubuntu.com/ServerTeam/Hadoop [09:16] bbcmicrocomputer, charm rocks BTW [09:16] how easy does this make stuff.... :-) [09:16] jamespage: very :) [09:17] jamespage: yes, it would be a good one to fix so I might give it a go when not busy [09:17] jamespage: thanks! [09:18] bbcmicrocomputer, nice README as well [09:18] jamespage: never used Hive before, it's actually not a bad solution [09:18] we should raise bugs on the mysql charm to get those features added as config [09:18] jamespage: all based on your work ;) [09:18] lol [09:18] jamespage: sure, will raise the MySQL bugs [09:19] I need todo the same for the haproxy one [09:19] I wanted to use it to front a cluster of hbase REST/avro/thrift servers [09:19] but its httpchk is a bit to in-depth ATM [09:19] jamespage: that would be cool [09:31] bbcmicrocomputer, its a bit http centric ATM [09:42] Do settings applied to a relationship with 'relation-set' persist for other new nodes that join the relationship? [09:44] <_mup_> Bug #965101 was filed: The MAAS provider does not raise an exception when get_machines(...) does not find the requested machines. < https://launchpad.net/bugs/965101 > [09:45] e.g. if the MySQL charm sets DB credentials when it creates a DB with 'relation-set', and then a new unit is deployed for the service, are those credentials supposed to be available to the new unit, or is the MySQL charm supposed to set them in its 'db-relation-joined' hook again? === asavu_ is now known as asavu [09:55] bbcmicrocomputer, I think so yes [09:55] i.e. they should already be avaliable - its set at the service level not the service unit level [09:56] might be wrong [09:56] jamespage: ok thanks [10:05] bbcmicrocomputer, one though on the hive charm - it would be nice to be able to switch in a metastore without having to re-create everything [10:05] i.e. re-use the same mysql database [10:05] is that possible? [10:05] I think the charm supports it [10:05] mysql that is [10:06] jamespage: that's what I'm not sure of [10:06] bbcmicrocomputer, I think the openstack charms do this - you might need to use the db-admin interface I think [10:07] jamespage: I thought when I added in another metastore the DB credentials on the relationship weren't given to the new unit, but that sounds wrong from a design point of view, so it might be me [10:07] jamespage: I'll have a look === al-maisan is now known as almaisan-away [10:33] <_mup_> Bug #965138 was filed: Juju concatenates the full API endpoint URL when it gets absolute paths from the MAAS api. < https://launchpad.net/bugs/965138 > [12:02] dear juju-users and devs, have you had a look at cdist already? [http://www.nico.schottelius.org/software/cdist/] it does a similar job on a pretty straight forward level === almaisan-away is now known as al-maisan [12:31] ok so inside the /hook/somehook script how would i refrer to something in /files/blah/file , e.g. whats the PWD , the script dir or ??? relitive work ? read(../files/blah/file) [12:35] juju make me some breakfast [12:43] telmich, i've looked at cdist previously, its targeted at something a bit different than juju [12:43] telmich, cdist is a configuration management tool, juju is service orchestration and deployment. [12:44] hazmat: hrm, it depends on how far you define service orchestration: cdist aims to do that job as well, the cut there is at the deployment stage [12:44] telmich, orchestration sits above config management, and within juju your free to use any tool you'd like to effect config management on the system, including cdist [12:44] telmich, so how does cdist connect two different services? [12:45] i.e. cdist is being used for setting up webservers with rails running [12:45] running on two different machines [12:45] I understand, that's a bit a different approach [12:45] how does juju do that? [12:46] well and the cut really isnt there either, we used its to deply too on OMG lately [12:46] juju that is [12:46] btw, the reason why I was mentioning cdist in here is because I thought it may be valuable as a supporting tool [12:46] given its not ideal at the moment it can certanly do it [12:47] telmich: yea possibly , like phone and ant kinda , same kinda suport roll [12:47] telmich, juju is designed for running interconnected services over a cluster of machines from the ground up.. the juju docs go into more details https://juju.ubuntu.com/docs/about.html [12:47] phing* [12:47] telmich, i agree cdist is a useful tool for effecting changes on a machine, in much the same way that other configuration management tools like puppet are [12:47] at that level juju is agnostic to what a charm author chooses to use [12:48] yea but i see what he is getting at kinda, it could be used by juju to push code to all the inder connected nodes of a type that needed them [12:48] but yea [12:48] it could be anything ant or phone or cdimg [12:48] or whatever [12:48] phing* [12:48] hazmat: ackI believe that higher level tools (like juju) could make use of CM tools like cdist, not sure about that though [12:48] if i keep typing phone its gonna become one:) [12:48] imbrandon: exactly [12:49] telmich: sure thats alrady part of the work flow then [12:49] telmich: we're just not tied to one tool though [12:49] the dev can decide [12:49] or devop [12:50] for inclusion into a 'blessed'/vetted distribution of charms.. there are some policies in place.. but juju also incorporates the notion of ppa without such policies to allow for charm authors to experiment and use whatever tool their comfortable with, and still allow their charm to be consumed by other users. [12:50] fits alot better into existing systems that way, the majority of the users IMHO wont be starting from scratch, but be iterating and working this in [12:50] and alrady have tools [12:50] telmich, indeed there are some charms that use puppet to install their configuration [12:51] hazmat: alright :-) [12:51] and i'm sure there will be some that use cdist [13:10] i still want juju to make me breakfast [13:23] Is there an example of an OpenStack stanza so juju can deploy to an OpenStack Compute cloud? [13:58] <_mup_> juju/repository-broken-charms r478 committed by kapil.thangavelu@canonical.com [13:58] <_mup_> merge trunk === hspencer is now known as hspencer[afk] [14:10] jcastro: m_3: you guys available this morning for a quick chat about Hadoopworld / Stratus? [14:17] arosales, I am! [14:18] jcastro: right on, thanks. Talk to you in a bit then === al-maisan is now known as almaisan-away [14:59] imbrandon: sorry I disappeared on you... Sundays and all. ;) [14:59] imbrandon: anyway, re the apple stuff.. I'll keep looking into it. Interop or at least some idea-theft would be good. :) [15:00] no worries [15:00] heh yea [15:00] imbrandon: and re drush not working with current drupal.. WTF drupal! isn't drush supposed to be the way to get drupal? And they break the backend that quickly? [15:00] amatures immatate, geniouses steal ? [15:00] lol [15:01] well its not [15:01] its more the fact that drush was borken with newer php [15:01] then drupal [15:01] newer php, like, 5.3.6 ? [15:01] drush is still very very young like in that sense, they iterate very very fast [15:01] yea, lol [15:02] So yet another thing that should not be in the archive. :-P [15:02] actually i think its a mix between 5.3.6 and 5.3.8 [15:02] * SpamapS thinks we've been rushing things into the archive [15:02] 3.8 is whats on mamp [15:02] and it breaks someone if both are used on same deal, or something, anyhow i would need to go look up the bugs again [15:03] well there was no way to know drush shouldne t have been [15:03] it was rock solid for monthsn [15:03] even a year or so [15:03] then suddenly like a flurry of releases came [15:03] all fixing and breaking major things [15:03] <_mup_> juju/repository-broken-charms r479 committed by kapil.thangavelu@canonical.com [15:03] <_mup_> add metadata error [15:03] like in 3 months [15:04] i think they had a change of leadership or something, it smelled of that [15:04] its just now kinda slowing back to a managheable pasce [15:04] pace* [15:04] imbrandon: well either way, why can't older drush's download newer drupals? [15:05] i dont rember the exacltym but instal from the archive and run "sruah status" i know its imediate obvious [15:05] drush status [15:05] like it break spectacular [15:05] but what i was saying is they mirror on github that has https [15:06] sooo it sould not be a big deal [15:06] just to grab from there instead [15:06] but yea i dont think any pear or pecl modules should be in the archive IMHO , not just drush [15:06] SpamapS: is there documentation for deploying _on_ openstack using Juju? [15:08] and i know stuff is crazy as heck right now , but its a standing offer with the access to my apple dev account software ( legitly ) whenever , month or 6 months from now [15:08] just to inspect the competition :) [15:08] marcoceppi: there's an askubuntu question/answer about it [15:08] Huh, must have missed it [15:09] marcoceppi: basically you need to set the ec2-uri, s3-uri, default-image-id, and probably default-instance-type ... [15:09] i need more points over there, i should spend a evenincg doing ask [15:09] that is.. until they break those default-* things and then I don't know how its going to work. [15:09] SpamapS: cool, I just found it: http://askubuntu.com/questions/94150/how-do-i-use-openstack-and-keystone-with-juju [15:09] (I say they because I am against removing them... but others are against keeping them... hence.. "they") [15:10] is there a way to keep them as a "module" per se , like add on-ish [15:10] not really a module [15:10] imbrandon: yeah github https would probably be the preferred option [15:10] yea that was the jest of the whole thing [15:11] imbrandon: I believe they'll be kept until a better solution is found [15:11] was github https [15:11] there is actually a ton of official mirrors on github for most things i found [15:11] How hard would it be to add another provider to the setup? Would it be pertinent to add prominent cloud services as different types (instead of ec2, have a generic eucalyptus type, hp-cloud type, etc) [15:11] if ya dig a bit [15:12] marcoceppi: i was wondering about zend's phpcloud.com myself, as i actually curreent use them [15:12] and they are openstack / php fabric [15:27] marcoceppi: like flavors of the ec2 provider? Not a bad idea (note that hp's cloud does not expose the EC2 API of openstack) [15:27] or at least, so I've been told [15:28] Ah, that's what I was trying to play around with [15:29] I have to think an OSAPI provider will appear ... one of the nova founders approached me about wanting to write one. [15:34] kinda like google storage uses the openstack s3 stuff but goggle app engine dont [15:35] in any way iirc [15:35] its funny you hit a google storage bugket and it even says in the xml its amazon, i guess not to break the tools or something is all i can figure [15:36] http://g.pixeldrop.net/ <-- thats a google storage bucket [15:37] but xml namespace to aws still [15:37] so i guess its only a namespace [15:40] imbrandon: yeah its just for api equality [15:40] <_mup_> juju/relation-id-option r500 committed by jim.baker@canonical.com [15:40] <_mup_> Initial commit [15:41] SpamapS: they seem to provide endpoints. I'm going to try to bootstrap an env on their hp cloud [15:41] it would be nice if app engine was ec2 compatable and vice versa [15:41] * marcoceppi will then self-document [15:41] imbrandon: that wouldn't make any sense at all though [15:42] imbrandon: since appengine is a PaaS not a VM provider [15:42] unless they changed something recently [15:42] yea but like phpcloud.com is a paas and does part of the api [15:42] nah [15:42] * SpamapS goes afk [15:55] <_mup_> juju/trunk r495 committed by kapil.thangavelu@canonical.com [15:55] <_mup_> merge repository-broken-charms repository.find should never break on charm error [r=fwereade] === jcastro_ is now known as jcastro [16:38] SpamapS: marcoceppi: m_3: ok so ... starting tomorrow we should review everything and start looking at entries? [16:38] I was thinking review and get them all ready, and promulgate all at the same time? [16:38] or do you guys want to just do them one by one? [16:38] jcastro: is there a list of what needs to be reviewed? [16:39] https://bugs.launchpad.net/charms/+bugs?field.tag=new-charm [16:39] any new, confirmed, or fix commited [16:39] jcastro: sounds good [16:40] m_3: clint and I sort of counted yesterday and there's like 15 things we can approve and promulgate. [16:43] I meant counted on friday of course. :) [16:43] <_mup_> juju/status-changes r486 committed by kapil.thangavelu@canonical.com [16:43] <_mup_> merge trunk [16:51] jcastro: I'm in [16:52] I've got local provider working on my laptop now [16:52] oh nice! [16:52] 12.04 *bows* [17:06] jcastro, been idlin' here for awhile :-0 [17:07] balloons: ok do you have links of stuff for people to try [17:07] I'm not sure what daviey and crew have in mind just yet [17:07] I haven't seen there work [17:08] ok [17:13] <_mup_> juju/relation-id-option r501 committed by jim.baker@canonical.com [17:13] <_mup_> AMP doesn't like None for string values [17:17] <_mup_> Bug #965507 was filed: Juju uses http to contact uec-images.ubuntu.com < https://launchpad.net/bugs/965507 > [17:18] <_mup_> juju/relation-id-option r502 committed by jim.baker@canonical.com [17:18] <_mup_> Merged upstream [17:21] <_mup_> juju/trunk r496 committed by kapil.thangavelu@canonical.com [17:21] <_mup_> merge status-changes simplify status output and support services with multiple relations under the same name [r=fwereade] [17:22] <_mup_> juju/force-upgrade r467 committed by kapil.thangavelu@canonical.com [17:22] <_mup_> merge trunk [17:24] Hi Folks, just a heads up status output has changed. [17:24] per discussion on the list last week [17:27] is it in ppa yet? [17:28] nvm, /me scans up a few lines [17:33] hazmat: thanks [17:34] marcoceppi, not yet.. [17:34] marcoceppi, ppas get built nightly [17:34] right, cool [17:34] m_3, i'll update the charmrunner service watch this evening [17:34] negronjl lynxman: ^^ (maybe neeed to update status parsers) [17:34] m_3, there will be some more status output changes when subordinates land latter this week fwiw [17:35] hazmat: gotcha... thanks for the heads-up [17:38] hazmat: won't that just be adding keys? not so much modifying the existing structure? [17:42] marcoceppi, it is adding keys, but the subordinate service representation in status is different, the subordinate units are reported under their principal units [17:42] ah, cool. I look forward to that landing [17:45] jcastro: gah i cant push, did it get added to a team ? [17:45] push to where? [17:45] bzr push ~imbrandon/charms/oneiric/quickdrop/trunk/ [17:45] says dont exist sooooooo [17:46] and i've pushed like lots already [17:48] lp:~imbrandon/charms/oneiric/quickdrop/trunk [17:49] marcoceppi: ahh i am a idiot sometimes, in a damn hurry :) [17:49] lol [17:49] ty [17:50] mt dew refill, brb === al-maisan is now known as almaisan-away [18:12] hallo. I used to have juju kind of working on canonistack. Then canonistack didn't work for me at all--when I did a euca-start-instances call (that is, nothing to do with juju) I would eventually get an instance that showed the status as "error" when you ean euca-described-instances. IS told me that was fixed, and indeed, that works now. However, juju still gives me the same error. I run juju bootstrap, and eventually ( [18:12] pretty soon) I get a system in an error status. [18:12] $ euca-describe-instances [18:12] RESERVATION r-0yhnhx0e gary_project juju-canonistack, juju-canonistack-0 [18:12] INSTANCE i-00001ce3 ami-00000079 server-7395 server-7395 error 0 m1.large 2012-03-26T18:08:04Z nova monitoring-disabled instance-store [18:13] any ideas on what's going wrong here, or any thing I should ask IS for? [18:13] best: any way I can fix this? :-) [18:14] fwiw, I'm on very newest juju from ppa [18:15] and happened in previous version (earlier today) as well [18:18] hmm [18:18] gary_poster, can you pastebin the error [18:19] someone know if the variable JUJU_RELATION_NAME is a good way to get a unique relation name that I can use in a path? [18:19] hazmat, one of the frustrations is that the above is all I get. From juju's perspective, it is simply unable to connect to the zookeeper. Unless I'm missing a log somewhere? [18:20] * gary_poster hunts for a moment [18:20] gary_poster, no logs on the client side.. juju -v status gives verbose output [18:20] * hazmat spins up an openstack instance [18:21] avoine: that should work yes. [18:21] great thanks [18:21] avoine: though you also need to take into account that there may be two remote-endpoints for the relation, so $JUJU_REMOTE_UNIT is probably also needed. [18:22] oh yeah [18:22] I'll concatenate those two to be sure [18:22] hazmat, doesn't seem informative to me, but dunno: https://pastebin.canonical.com/63059/ [18:25] gary_poster, your trying to connect from outside of the network [18:25] gary_poster, you'll need to associate an address by hand to the bootstrap node for that to work [18:26] or you need to be located somewhere where the public address of the instance ( server-7396) is resolvable [18:26] gary_poster, i do euca-allocate-address, euca-associate-address -i bootstrap_node_isntance_id public_addr [18:27] hazmat, the .canonistack trick won't help with that? I think it will--that's what RT 50773 is about--and the problem I see here is that the instance doesn't even start. that's still in line with your diagnosis? [18:27] gary_poster, hmm.. actually that appears to be broken [18:27] associating ips to the isntances [18:28] once it starts it is supposed to be assigned a .canonistack name [18:28] which is supposed to be resolvable now, if you have the right ssh config stanza [18:28] (that's what I was trying to test) [18:28] but the machine doesn't even start [18:28] gary_poster, yeah.. it doesn't start for me either [18:28] at least, that's the way it seems to me [18:29] ok, at least that's two of us :-) [18:29] gary_poster, that seems to be a stack instance issue [18:29] * hazmat tries launching one manually [18:30] manually works for me, if you mean euca-run-instances [18:30] gary_poster, which image are you using? [18:30] the output is slightly different [18:30] hazmat, ami-79 [18:30] beta [18:30] in both cases [18:31] gary_poster, cool, i'm using 78 [18:33] hazmat, fwiw, I noticed a slight diff in the euca-describe-instances of the juju vs. manual instances but I didn't know what it means. https://pastebin.canonical.com/63062/ shows manual, and then juju. If you compare the two, the manual instance shows the status as "pending gary" while the juju instance shows "pending". I have no idea what this means or if it is relevant, but it is the only difference I saw other than one [18:33] working and the other not. [18:33] gary_poster, its the key pair name [18:34] oh...the juju doesn't need a key pair? [18:34] juju doesn't launch with the api key name, it uses cloud-init to seed the key [18:34] ah ok [18:34] so maybe irrelevant, unless cloud-init is the borked part [18:34] gary_poster, still that's all post instance running, these instances are in an 'error' state [18:35] which would suggest a problem pre launch [18:35] agreed [18:35] even cloud-init failures leave instances running [18:35] I see [18:36] I tried to geta vanguard to look at this earlier to look for a traceback but they were swamped [18:36] the poor error reporting in canonistack, from the user's perspective at least, is annoying [18:36] gary_poster, definitely [18:37] gary_poster, for fun try terminating an instance, and checking status [18:37] it just shows as running still before it pops off [18:37] heh [18:37] I didn't see that [18:37] I saw a bunch of errors [18:37] but never running [18:38] gary_poster, i'm trying to narrow down the param that causes the instance state 'error' [18:38] tbd [18:39] hazmat, oh ok cool [18:40] ah.. its the group [18:40] huh [18:41] hazmat, for fun I asked the vanguard again for a traceback. it may turn out to be irrelevant by the time they get this, but I figured it was wirth another try [18:41] gary_poster, juju associates two security groups to a machine, an environment group, and a per machine group [18:41] right, I saw that on ec2 [18:41] seemed nice [18:41] gary_poster, if try on the cli against openstack to specify both groups when doing euca-run-instances.. the instance will fail [18:42] trying again at the moment with just one [18:42] huh, ok [18:42] also fails with one instance [18:42] er. one group [18:42] heh [18:42] previously openstack would basically ignore security groups, i think its got them implemented, but its not clear it has self-referential groups working [18:43] the environment group opens traffic up internal traffic between machines in the environment [18:43] yup.. its the self-referential group thats killin git [18:43] the per machine group is fine [18:43] ah [18:44] gary_poster, now the question is who to forward the issue to.. [18:44] heh [18:44] I've heard that IS updates canonistack with newest openstack every Monday [18:45] so a bug against openstack might be the right idea [18:45] depending on turn-around time, a workaround might be convenient too [18:46] and btw hazmat, duped that "euca-run-instances -k $NOVA_USERNAME -t m1.large ami-00000079 -g juju-canonistack" leads to an error state for me as well, if you wanted any confirmation. [18:48] <_mup_> Bug #965591 was filed: failure to launch instance with self-referential security group < https://launchpad.net/bugs/965591 > [18:50] cool [18:50] #is did not find a traceback :-/ [18:51] gary_poster, well it all depends on where they looked.. [18:51] heh [18:51] * gary_poster has jokes, squelches them [18:51] gary_poster, i'd have to guess there is one somewhere, just not in the nova api endpoint, it definitely hits the queue, its probably in the network component [18:51] hm, ok [18:52] or directly on the instance node [18:52] * hazmat keeps missing lunch [18:52] i aim to resolve that, bbiab [18:52] :-) k, bye, thanks [18:54] gary_poster, i'm not sure there is a good workaround, juju needs that environment group, as minimum to tag its instances in the env, we could not define its access rulz for ostack and see if that works, but if they do implement security access for internal network usage, it will still be hosed [18:57] hazmat, yeah I can see the broad shape of the problem. [18:57] why did it work before? [18:57] no access rules meant free-for-all? [18:59] gary_poster, yup.. it didn't really process them for 11.10.. it could be a regression for the current release though, not sure [18:59] gotcha [18:59] m_3, do we have any automated tests against ostack? [19:00] thanks hazmat. #is is still looking for logs, which I'll add in some hopefully safe-ish manner to the bug report if they find some and they look potentially informative [19:01] though it sounds like the description you have may be sufficient for the sorts of openstack devs ho would actually work on it [19:01] who [19:10] hazmat: not yet [19:11] hazmat: that's what I'm currently working on.. [19:11] hazmat: only openstack installation I have access to is totally barfing atm... images coming up 'error' [19:23] m_3, i just filed a bug report on it in nova & juju after gary_poster reported the same [19:24] hazmat: thanks [19:24] m_3, bug 965591. hazmat, m_3, doesn't tell me a lot, but looks like https://pastebin.canonical.com/63067/ is relevant traceback. deej is confirming on #is [19:24] <_mup_> Bug #965591: failure to launch instance with self-referential security group < https://launchpad.net/bugs/965591 > [19:26] gary_poster, that looksl like it [19:26] gary_poster: cool [19:27] m_3: oh hey, how'd the summit thing go? [19:28] jcastro: not too much new... talked to cjohnston about a new theme branch for the site [19:29] jcastro: we've gotta move the environment to a different account for financial reasons [19:29] jcastro: so that'll happen prob later this week [19:30] ive been helping him learn the ropes of proper css and js with it and the lenaro [19:30] subthemes [19:30] jcastro: once that's done, then I'm back on hold until we have a theme... prob nother couple of weeks it sounds like [19:30] imbrandon: cool... thanks! [19:31] m_3: nah, we'll have it in a week or less, week is giving some breathing room [19:31] nice [19:31] he just inst confident in timeline :) [19:32] but yea i;d say realistic about 3 more days for the django and another 1 for wordpress and 2 more for drupal [19:32] is the timeline [19:32] lenaro is in with django as a child theme [19:33] e.g we're jst doing one template system for all the community themes, thats why he isnt confident beacuse of the big changes [19:33] but i've been thought it ,we'll crank her out [19:33] through* [19:34] but if you look in the -refresh branches of the themes you can see the progresss [19:34] they are about 80% [19:35] infact here is the base we finished last night , for the django, now just got to sub template it for summit and lenaro [19:35] http://theme.chrisjohnston.org/ [19:35] imbrandon: cool... I'll try to spend a little more time externalizing the theme from the charm config instead of the settings.py [19:35] might have a few more MPs [19:36] m_3: the appflower one looks cool too [19:36] but it's configured to pull from summit production atm [19:36] kk, yea we err i restructure all the themes to be like www --> base(top) -->django(top) --wp(top) -->drupal(top) and the [19:37] base is pure html [19:37] and then we make the django and wp tempalate just by subst [19:37] the mp in place [19:37] and last will be the drupal one [19:37] but i also made bash scripts to [19:37] generatte the django and stuff from the html again [19:38] like thay are a psudo template wirth ##TITLE## [19:38] imbrandon: hmmm... I'd have to see that to figure out what the charm should be doing to install the theme [19:38] the charm probably should just include the base_website.html and the /media directory [19:39] ideally, I'd just have a django branch and a theme branch [19:39] and then makes its own inheriting the base_website.html [19:39] yup [19:39] exactly [19:39] the django folders are not django sites [19:39] right, then I can add vhost aliases for media [19:39] they are just the assets and .html templates [19:39] yup [19:39] cool [19:40] i pretty much made him do it, he about shed a tear but then loved it when it was clear what was gonna happen [19:40] heheh [19:40] will make future updates aklot alot easier and using thing s like less or scss too much simplete [19:41] simpler* [19:41] plus there is actualy build scripts to minify the assets and such and create the templates from a common base [19:41] so they all match ubuntu guidlines with no extra work too === hspencer is now known as hspencer[afk] === jml_ is now known as jml [20:35] jcastro, sweet 2 other juju talks at europython were submitted [20:35] nice! [20:54] jcastro: we about ready to close down the contest? [20:55] jcastro: maybe in 3 hours? That would be 00:00 UTC [20:55] sure [20:56] jcastro: I'm going to try and do 2 hours of review tomorrow and Wednesday. [20:57] negronjl: you going to be around and able to do reviews this week? [20:57] between you, mims, and jamespage, we should be ok [20:57] we have a ton.. [20:57] indeedy we do! [20:57] imbrandon: did you submit your charm yet? [20:58] negronjl: nijaba: if you guys have time to review one or two this week it would really help out [20:58] jcastro: Instructions for deployment and accessing subway added (merge proposed) [20:58] lynxman: you too! [20:59] SpamapS: is there anyone from the launchpad team that we added to charmers? Maybe one of them could lend a hand for one or two? [20:59] jcastro: What do these Juju mugs look like? [20:59] :D [20:59] it's the ubuntu travel mug from the store [21:00] yea [21:00] i've been iterating on it [21:00] but i submitted it last night late [21:00] SpamapS: ^^ [21:01] jcastro: Uhh you didnt say that... ok you can send me one of those :P [21:01] :D [21:01] I thought it was going to be a ceramic mug :D [21:01] SpamapS: i still need to update the nginx configs a little , but its 95% complete and submitted [21:01] well 92 :) [21:01] I think hosted IRC will be a popular charm [21:01] but working on it now [21:01] jcastro: subway [21:01] ? [21:01] yeah [21:02] yea i love that thing [21:02] i found it like 2 days before dudes post on planet [21:02] imbrandon: as long as its in and tagged as new-charm, we'll allow refinements through this week as we review them. [21:02] does juju upgrade work with subway? like, does doing that just update the subway code from git? [21:02] SpamapS: rockin, ok yea it is [21:02] jcastro: Not yet [21:02] i showed jcastro awhile ago [21:02] jcastro: this will be added later when I bug SpamapS some more :D [21:02] i had trubble attaching the banch so its on via a comment [21:03] SpamapS: ^^ [21:03] imbrandon: ok thats fine [21:03] something with chrome on osx was stopping me, gonna do it from insde my ubuntu vm later on next push [21:03] i'd say another hour or two [21:03] not all week and i'm good [21:04] but thats good i got leway :_ [21:04] heh [21:05] also once i have the next push done i'm gonna convert to full git pushes , talked to jcastro about it already as a good thing so it will work you guys on that workflow anyhow [21:05] so the first frwe times i'll likely ping someeone and let em know a pull req is on gh waiting [21:05] but try not to be annoying about it :) [21:05] but i'll* [21:09] imbrandon: *boo* git [21:10] but i'll*/msg SpamapS ~imbrandon/charms/oneiric/quickdrop/trunk , gonna update the normal drupal one too even if it dont make the contest as it seems abandon and jcastro said it looked fair game [21:10] yea [21:10] "git: because VCS *SHOULD* be hard." [21:10] i love my git [21:10] :-P [21:10] hahah [21:10] i actually find git easier than bzr , like for reall, not trolly [21:10] Yeah I keep hearing that from people [21:10] but as a 5+ year bzr user.. [21:11] but its because i learned it first, i think the same thing about hg too [21:11] and hg is supose to be easy [21:11] and before that, a user of cvs, svn, perforce, and even a tiny bit of vss ... I despise git's obtuseness. [21:11] yea i've only used bzr when i had to, even when i was very ver actve most of the time i use git on anolith and sync :) [21:11] or use git-bzr [21:11] :) [21:11] hehe [21:12] I'm sure as I get pushed along toward git since bzr has been mostly marginalized.. I'll figure out that git is great. But right now its like.. a million times harder for me to use. [21:12] yea cvs -> svn more svn , a little svn, then git, then svn then bzr then git/bzr then git git git git git hg curse git git [21:12] not bzr a little again :) [21:13] now* [21:13] SpamapS: I am very interested in adding a upgrade hook to Subway but not sure how best to implement that part of the hook [21:13] bkerensa: one line "install" :) [21:13] heh [21:13] imbrandon: ? [21:14] bkerensa: the best way to do it is to simply make sure all your hooks are idempotent, and then just have your upgrade-charm hook call them all in the right order.. which is stop, install, config-changed, start [21:14] as in my updgrade just calls instll and i check in install with conditions for if/else things [21:14] ahh [21:14] good idea [21:14] imbrandon: which is ur charm? [21:14] bkerensa: for subway, using git.. you should just check to see if the dir is already there.. if it is.. cd into it and 'git pull' [21:14] its not toally done but its ~imbrandon/charms/oneiric/quickdrop/trunk [21:15] imbrandon: yeah the most basic thing to do is just to call install [21:15] yea, jst look for a -z /dir/.git [21:15] or something [21:15] imbrandon: adding the other bits in there will force you to write better other bits [21:15] yea i was trying to make the deadline so somethings arent ideal but work clean [21:16] and i'll fixem up as i go i figurred release early release often :) [21:16] i actually had to make a install stamp file i check for with the config, i dont like that at all [21:16] but its doing the job for the moment [21:18] SpamapS: infact if its not cheating blah blah, can you eyeball what i got pushed, its missing some essential bits to function fully but i got those here and am polishing them up [21:18] not a full revirew etc etc just an eyeball or two for a sec [21:18] fi you have time and dont mind [21:19] mostly whats missing is the nginx bits and the apt-get line [21:19] imbrandon: I'll be doing some reviews later today, will look at yours then. [21:19] kk sounds good [21:19] i shoudl have those bit up by then too then [21:20] bkerensa: i'm still very knew to this charm thing too but if you need a hand i can try ( not a newb coder etc just charms :P ) [21:20] if ya cactch a snag from something from mine into youer or whatever [21:23] i need to adapt my ide to like the charm layout [21:23] IDE?!?!?! [21:23] Y U USE IDE? [21:23] :D [21:24] nano for the win [21:25] editor wars, FINALLY SOMETHING FRESH AND NEW ON THE INTERNET [21:25] lol [21:26] well kinda, bkerensa been using bbedit for the charm so far, zend studio seemd a bit overkill :P [21:26] but the dir layout in bbedit is kinda finly [21:26] SpamapS: indeed [21:26] funky* [21:27] bah i dont pay attn to em anymore, if someone is harping on vim or emacs i use pico ( aliased to nano ) in their view :) [21:27] i find its normaly the ones that purport aptitude over apt-get in online tuts and lbog post comments [21:27] e.g. same group :) [21:29] on a serious note though it did take years , for real for my muscle memory to not type pico, and tupe nano, i has a bash alias untill 3005ish or later [21:29] 2005* [21:30] actually back then i think i just made symline to nano --> pico as i dident know about aliases [21:30] heh [21:30] ( 90's ) [21:31] symlink [21:31] dude i'm getting that damn keybopard today [21:31] apple store dont close till 9pm i'll make it [21:31] this thing is gonna be the death of me [21:32] ( its one of the small bluetooth ones , my usb wired aluminum one died to coke-a-cola a week or two ago ) [21:33] i dont have big hands and i feel like i got huge fat fingers with this thing [21:34] SpamapS: what can we garentee is on the system, ubuntu-minimal or ? [21:34] like accross clouds [21:35] imbrandon: yeah minimal [21:36] k [21:36] imbrandon: eventually we'll have all of 'cloud-image' [21:36] imbrandon: or rather, '^cloud-image' since its a task, not a package [21:37] ahhh does that add php so i dont have to bootstrap it with bash ? hehe [21:37] lol [21:37] should toss php-cli on there python and perl are :) [21:38] no [21:39] imbrandon: somehow I think an idea like that may be given a label the PHP community would not appreciate ;) [21:39] I'm not saying people will call it *crack* [21:39] or *b.s.* [21:39] but something that you might not like ;) [21:40] hahah yea i think mine is a rare case [21:40] where i have phing and drush to do alot fo te heavy lifting for install and config [21:40] and both are php tools [21:40] writeen in php [21:40] :) [21:40] written [21:40] bleh [21:40] but most wont be , even like wordpress [21:40] unless its intentional [21:41] i think a zend framework one would be the only other case likely [21:41] but that wont come till partials are here [21:42] ls [21:42] . [21:42] .. [21:42] build.xml [21:43] pr0n.jpg [21:43] jorge-jono-and-camel.gif [21:43] $_ [22:00] bkerensa: merged your change. Note that you should probably re-branch the charm from lp:charms/subway so that you get all the latest stuff ~charmers adds .. your branch is missing one thing already [22:00] SpamapS: ahh did you promulgate it properly? [22:00] :D [22:02] bkerensa: promulgate is only needed to create the official pointer to the official branch. After that we just merge and push. :) [22:07] SpamapS: is there no way to sync my branch with the upstream lp:charm/subway? [22:07] so that my branch stays frosty? [22:10] bkerensa: sure, you first should do 'bzr missing lp:charms/subway' and make sure that your branch doesn't have anything not in the official one (ignore any that are only in lp:charms/subway) [22:10] bkerensa: if you don't have any local changes, then simply remove your branch, and bzr branch lp:charms/subway [22:11] bkerensa: if you do have extra changes, move it out of the way, and do 'bzr branch lp:charms/subway', and then cd subway, then 'bzr merge ../path-to-your-old-branch' [22:12] bkerensa: then commit, push, and propose for merging. :) [22:12] and yes I know that git makes this *infinitely* simpler with rebase [22:12] :P [22:12] I prefer bzr :) the git people were not so nice to me :P [22:12] this is actually the one time where git wins in my head. [22:12] I submit patch to them and they make me jump through hula-hoops [22:14] <_mup_> juju/relation-id-option r503 committed by jim.baker@canonical.com [22:14] <_mup_> Allow for context to access invoker so it can get cached children [22:16] SpamapS, we could probably install the cloud-task with minimal changes into the local provider [22:17] the switch out to the ubuntu-cloud template is more extensive, since it also involves dropping libvirt [22:17] SpamapS: Ok well I req'ed to join ~charmers because I wanna contribute more :) I am really getting to like juju [22:22] <_mup_> juju/relation-id-option r504 committed by jim.baker@canonical.com [22:22] <_mup_> Test invoker is accessible from context [22:25] <_mup_> juju/relation-id-option r505 committed by jim.baker@canonical.com [22:25] <_mup_> Merged upstream relation-ids-command [22:25] bkerensa: *awesome* [22:26] hazmat: thats a good idea actually [22:33] <_mup_> juju/relation-id-option r506 committed by jim.baker@canonical.com [22:33] <_mup_> Mark conflicts resolved [22:38] <_mup_> juju/relation-id-option r507 committed by jim.baker@canonical.com [22:38] <_mup_> Update tests re relation_get's support of relation_id [22:52] <_mup_> juju/relation-id-option r508 committed by jim.baker@canonical.com [22:52] <_mup_> Allow Invoker to not take a context to fix failing test === hspencer[afk] is now known as hspencer [23:07] <_mup_> juju/relation-id-option r509 committed by jim.baker@canonical.com [23:07] <_mup_> Test relation-ids command through the invoker [23:51] hello [23:52] I'm sure it's been reported by now, but I'm unable to get to the juju charm browser site [23:52] http://charms.kapilt.com/ [23:52] 502 Bad Gateway [23:58] hazmat: ^^ [23:59] oh dear [23:59] hmm. mongodb went down