[01:03] checking [01:03] imbrandon, jujucharms.com is in ec2 fwiw [01:04] imbrandon, i can hit omg as well [02:13] kk ty, kk ty [02:13] yea i could hit omg the whole time, but i cant use juju [02:13] for status or anything [02:14] but external monitors show fine, and i can hit the website as well , but i cant hit the bootstrap node with juju , wonder but because of dns errors not [02:15] net ones, well least not on the surface [02:15] i'll track it tomarrow if its still happening [02:15] dont feel like messin with it tonight and its not an emrngncy blah blah [02:16] ty for checking tho hazmat [02:54] imbrandon, ping [02:54] imbrandon, fair enough.. not tonight [02:55] imbrandon, i'll be around a bit in the am tomorrow, then off on a plane, but i'd be happy to help debug [03:42] sounds good, yea my brain is about fried as far as anything productive [03:42] for the evening :) [03:43] i've been in my OSX boot playing in photoshop the last hour heh, bought the new cs6 version today [03:44] they drasticly droped the price on the creative suite , so it was a no brainer [03:44] its like a $30 a month subscription now ( optional ) instead of $1200 up front [03:44] and includes ALL the CS on and offline apps [03:45] smart move imho [03:46] if they would just port at leaste the online versions to linux ( the online apps are 98% as powerfull as the offline ones ) [03:46] i would be set [10:54] I'm getting juju FTBFS on current quantal, due to a bunch of failed test cases, is that something anyone would be able to confirm? https://bugs.launchpad.net/ubuntu/+source/juju/+bug/1017113 [10:54] <_mup_> Bug #1017113: FTBFS: Sveral tests fails < https://launchpad.net/bugs/1017113 > [11:50] errr [11:50] bholtsclaw@ares:~/Projects/local/charms/precise/nginx$ bzr push lp:~imbrandon/charms/nginx/trunk [11:50] bzr: ERROR: Permission denied: "~imbrandon/charms/nginx/trunk/": : Cannot create branch at '/~imbrandon/charms/nginx/trunk' [11:50] :( [12:36] imbrandon: you need a series [12:36] ahh crap [12:36] ty [12:45] real fast tho, yea of nothing else at all lets steal that version showing trick, kinda slick AND solves the problem of parsing it possibly later [12:45] k , i'm out for a bit ... SpamapS ^^ [14:23] * imbrandon just put the place holder site/design up on behalf of the php-fig group [14:23] * imbrandon does a little dance , sings a little song [14:23] heh [14:24] http://www.php-fig.org :) they are the group that made/makes like PSR-0 etc [14:24] e.g. the peps in python [14:30] oh [14:30] cool [14:30] wait, PEP? [14:30] I thought PHP had RFC's [14:30] like pep-8 [14:31] rfc's and PSR's now there is only PSR-0 and -1 so far [14:31] thus me getting the chance to do the initial web :) [14:32] its a young group but has some heavy heavy weight behind it like drupal, wordpress zend zf2 symphony welll helll just about any php group or company has a voting member [14:33] fig is "framework interop group" and psr's are php standard resolutions or something, but makes sure all frameworks work togather and do shit the same way [14:33] and even if there is 5 ways to do something there is now a "right" way , or will be :)_ [14:33] heh [14:34] Ok, so PSR's are just about php code [14:34] yea [14:34] where as the PHP RFC's are about php dev [14:34] cool [14:34] LIKE pep :) [14:34] well [14:34] its about interop [14:34] like psr-0 says how autoloaders need to work [14:34] and bare minimum they have to supoort etc [14:35] so i can use bits form zf framework with symphony now, like for real [14:35] and it autolaods [14:35] an extreem case of this will be when i can use a wordpress plugin on drupal [14:35] thats a goal, far far out but that level of crap from the ground up [14:36] and -1 talks about tabs and spances [14:36] and 2 spaces versus 4 etc [14:36] so its a mix [14:37] i think -3 thats in the works is covering Cacheing and standarinzing some Cache interfaces [14:37] i *think* [14:38] but yea in like less than 6 months its got evey major player behind it and like 300+ members [14:47] thats great [14:47] PHP is growing up [14:47] despite all the past predictions that it would eat itself [14:47] (I never subscribed to those predictions btw ;) [14:59] <_mup_> Bug #1017113 was filed: Juju test suite fails sporadically due to low timeouts < https://launchpad.net/bugs/1017113 > [15:00] FRAK [15:01] i just did this whole function in JS [15:01] * imbrandon is working in a php file [15:01] bah, i need a cigarette and a mt dew, heads not right :) [15:03] SpamapS: there we go [15:03] http://cl.ly/HcuA [15:03] fully filled out with temp data and everything [15:04] so .. much.. white(black)space [15:04] till they decide what content they want heh :) or they dont like the look of it [15:04] thats a screenshot [15:04] the black is the website holding the screenshot [15:04] l [15:04] o [15:04] l [15:04] got it [15:04] and on your smaller monitor their wont be so much white [15:04] I'm also just having my first coffee even tho I've been up since 5am [15:04] i'm on a 24inch :) [15:05] heh [15:05] I'm on the 11" [15:05] heh [15:05] thats what s... [15:06] s ? [15:06] res ? [15:06] 1920x1080 x 3 [15:07] no retna for me yet :) [15:09] okies, back in a bit, skipped the shower for breakfast instead this morning, not gonna put it off too long or i'll be all icky ... ewww yea, ok back in a bit [16:00] SpamapS, about 3m40s on an x220 with ssd (samsung 830) with fsync off on zk [16:16] hazmat: yeah I'm about the same using eatmydata [16:17] hazmat: on the MBA [16:17] with "whatever the heck they use for SSD" [16:17] hazmat: hey, 'make check' doesn't seem to work for me [16:18] I have JUJU_TRUNK set... [16:18] make: *** [check] Error 1 [16:39] SpamapS, apple uses samsung and toshiba, pretty slow till this year afaik [16:40] SpamapS, make check needs modified files committed [16:40] hmm.. actually not committed [16:41] make review checks committed changes on a branch [16:41] most of those are ben's additions, i normally just use make coverage [16:42] yea the ssd's untill this last bactch from apple sucked [16:42] they are samsung mostly [16:42] but not they got some intel ones [16:42] but yea before these toy want to buy third party ram/ssd [16:42] ram is that damn hynix [16:42] bleh [16:43] imbrandon, their still using samsung from what i've read, but its the new 830 series controller, which is pretty fast (also what i use for my thinkpad) [16:45] it doesn't do the write deamplification like the sandforce ones, and its not quite as good as the intels as background cleanup, but with trim OS support it works fairly well for me [16:47] imbrandon, did your ec2 issues get resolved? [16:50] yea , well kinda [16:50] its a bug but i figured it out [16:50] now i forgot what it was but i made a note cuz i wanted to report it and let you know [16:51] ahh, the only 2 i've owned have been intels so far [16:51] i used 2 others but quickly took them back for write errors in the first week [16:52] ohh i know what it was now, the nodes got a new juju [16:52] but i had old local [16:52] and it kept looking like a dns error [16:53] but as sonon as i updated my local juju it was all good [16:54] cool [16:54] time to pack up and head to velocity then [16:54] cool cool, dont have too much fun, i plan to fix the docs this afternoon [16:54] so they should be all new and ready for yall [16:55] i wanted ot get my nginx pushed first tho ( i did get the first round pushed, so its atleast on LP now heh ) [16:55] but still is missing a few hooks i need to complete [16:56] SpamapS: and the majority of it is in PHP heh :) [16:56] next one i think i'm gonna use ruby on but not chef/puppet, just ruby [16:57] just to kinda feel out the diffs [16:59] http://bazaar.launchpad.net/~imbrandon/charms/precise/nginx/trunk/view/head:/hooks/install [17:01] LP/logger head makes it all pink :( [17:04] you know what i realized just now, and now i see it i cant beleave no one else has [17:04] why do we do all these config-get's and put them into varables in every hook [17:05] why does juju not just put them in the env when the hook fires [17:05] its pretty much garenteed config-get will be called once, if not many times in a row when either they could config-get once at the top and split the json out [17:06] or none at all and juju could do it for us, repetitive boilerplate is not good, i started to make a "common.php" and was like wtf, why dont juju do this [17:07] SpamapS: ^^^ ( and m_3 / hazmat if yall arent on the road yet ) [17:10] its a little implicit [17:10] well env vars' it happens all the time for websites [17:10] say you get a variable mismatch [17:10] the db conn info in the server env [17:10] where's the error line [17:11] well i'm not saying that the config-get should die, just that it should prepoulate [17:11] the env [17:11] it almost has to anyhow i'm sure it loads it prior to the config-get call [17:11] like whne the hook fires [17:11] anyhow [17:12] imbrandon, worth filing a bug/feature request for [17:12] anyhow, not saying its well thought out either, it just stuck me a few minutes ago :) [17:12] yea, i'll see where it goes ;) [17:12] only takes 5 min to fill out a LP page [17:13] its worth that [17:16] ZOMG !!! the new photoshop cs6 went gimp :( frak [17:16] its all multi windowd, man i knew i should have fired it up fist [17:17] damn the artwork, window chrome , icons , etc alll look gtk3/gimp [17:23] SpamapS: check this out, this is default like 5 seconds after it finsihed installing ( e.g. not a custom theme or something ) [17:23] http://cl.ly/HbID [17:23] watch out tho, there might be alot of black round that one too hahaha j/k [17:24] but tell me that dont look gtk/gimpified [17:24] ps has always done its own unique UI , well untill now i guess [19:28] is there bug open for getting secret key etc from the environment (e.g. AWS_ACCESS_KEY_ID) ? [19:37] lifeless not that i'm aware, but i havent had prblems with that [19:37] just thinking it would make 'get up and go' easier. [19:37] like you set it and then it cant get it ? [19:37] sophisticated installs need partitioned credentials [19:37] ohh [19:37] ok i see what ya mean [19:37] i was thinking something diffrent [19:38] yea, that would make sese and would fall into line with the config-get i was talking about [19:38] all the euca and aws tools, for intsance, read from the environment. (The actual environment, not environments.yaml) [19:38] yea [19:38] I'll check for a bug [19:38] yea a couple two of my charms export those just for that [19:38] so, looks like openstack just does noddy dns names [19:39] rather than the mass assigned reverse-ip-prefix that ec2 does. [19:39] which is why juju + internal openstack is so tedious, I suspect. [19:39] really ? i've been meaning to get the openstack a try on hpcloud the last few days just not made it that far yet [19:39] production instances probably have some workaround [19:40] but the code I'm reading so far depends on e.g. ldap DNS server integration [19:40] kinda was hoping to give it a go and then talk omgubuntu into moving maybe if it was solid [19:40] which is way more complex [19:40] ahh [19:40] i run open directory local on my lan anyhow ( i have a osx box that acts as a pdc and uses open directory ) [19:41] see ./nova/network/ldapdns.py for instance [19:41] funny tho, no windows clients so really i could just use nis or whatevr but bah. i followed the gui setup when i set it up a few months ago [19:42] wasent really intended to be a "real server" when i set it up, was just testing osx server 10.8 and then i came to rely on it more and more, now i got to get a migration plan done at some point [19:42] heh [19:43] thats sad for my home network [19:43] lol [19:43] pretty sure i have it doing primary logins, home directories , perms control on the other 2 linux servers [19:44] and some other minor osx specifc stuff like time machine and iphoto/itunes shares [19:44] all on something that wasent intended to be used in "production" at all [19:44] lol [19:45] ah, they are supported I think [19:45] is kinda neat how much ubuntu picks up on the osx services now tho, used to be a pain, but now banshie/rhythmbox see the itunes , all of them use mdns native [19:45] etc [19:45] I didn't realise that the openstack default is to use different environment variables *for the same values*. [19:45] * lifeless headdesks [19:46] heh [19:46] yea i think it looks in the cli then env, then config home and confg server [19:46] err scratch that last one thats the aws tools [19:47] yup, they are supported. [19:47] cool [19:47] openstack's scripts set EC2_ACCESS_KEY, not AWS_ACCESS_KEY_ID, for instance. [19:47] I'll file a bug on txaws for supporting the openstack variables [19:48] hrm, maybe i'll take a break from the nginx charm a minute and get a os env rolling, then i can use it to finish testing the nginx charm up [19:48] and work some kinks out [19:48] nice, yea i'm sure hazmat will toss it in, seen him do a few patchs quickly on it the last week or so [19:48] but he is in route to a conf today iirc [19:49] and there most of the week next week i *think* [19:49] velocity, not sure its a week long tho [19:49] few day at least i'd imagine [19:50] oh btw, i did get my credentials working on the juju wiki, just needed to log out like ya said [19:50] there we go: https://bugs.launchpad.net/txaws/+bug/1017239 [19:50] <_mup_> Bug #1017239: openstack ec2 credentials not picked up from environment < https://launchpad.net/bugs/1017239 > [19:50] ( that was you that handled the ticket right ? heh sorry if not ) [19:52] my thing with all these cli tools is they all want the same keys but diff env names [19:52] irks me [19:57] mmm, no, I ididn't [19:57] I get cc'd on some rt stuff, part of my job to be looking for trends and issues [19:57] whoop sorry then , mixed the name up, i had just glanced at it [19:58] yea you might have been cc'd i just glanced at the name that handled it but it was so late friday when i got it working i put it off emailing back [19:58] heh [19:58] or it might have just been a similar name :) [19:59] * imbrandon is curious now, goes to look [20:00] ahh yea, you were just cc'd on it for some reason :) [20:01] chris stratford did it, not sure who that is on irc if i even know them /me replies so they will see it monday and close it hopefully , /me should not be so terrible about the responses [20:04] imbrandon: this is what ec2-75-101-245-65.compute-1.amazonaws.com - ec2 returns, vs server-2 in my local openstack install [20:04] you can see the former is trivially bulk-provisionable [20:04] or even just dynamically answerable by a hacked dns server [20:05] hrm [20:05] yea thats nasty [20:06] however, these days we get ip addresses straight back [20:06] so we can use that [20:06] right [20:06] needs a txaws patch tho [20:07] yea dns is a tricky thing to even for experinced programmers / devops / ops , one little variance and can toss a whole range of things out of wak [20:07] so that dont help [20:09] https://bugs.launchpad.net/txaws/+bug/1017245 [20:09] <_mup_> Bug #1017245: ipAddress and privateIpAddress are missing from describeInstances < https://launchpad.net/bugs/1017245 > [20:09] * lifeless is on a yak shaving mission [20:09] haha [20:10] argh [20:10] now [20:10] 2012-06-25 08:10:14,921 ERROR Invalid SSH key [20:10] ^ $curses [20:10] haha [20:10] bad env.y most of the time [20:10] when i get that [20:12] pasted you my env for this [20:12] Am I using the ssh option wrong or something ? [20:12] does it upload it via cloud-init ? [20:15] does the key need to be registered with the cloud [20:15] or is cloud-init used to insert it ? [20:16] yea [20:16] it gets passed as user-data [20:16] to cloud init [20:17] along with some pkgs to be installed like bzr [20:17] and juju zk [20:17] yah [20:17] log shows zk etc installing ok [20:17] and then it also passes a 64base encoded shell script that makes like var/lib/juju [20:18] can you "curl -iS http://instance.data/latest/user-data [20:18] oin maas ? [20:19] EC2 API has it, and i know rackspace and hp both do but not sure if its the compatability layer or OS that also does it [20:19] but that should show you the whole cloud init user-data script including keys [20:19] if they were passed [20:21] or "curl http://169.254.169.254/1.0/user-data" may work too [20:21] well, I can't ssh into the machine [20:21] the other is just an alias on the local nets that implment it [20:21] oh frack [20:21] because the ssh key is whats failing :> [20:21] thats right [20:22] hrm [20:22] otherwise, I'd be like 'woo yeah. lets see' [20:22] this is MAAs right ? [20:22] no [20:22] openstack [20:22] or jusy OS [20:22] seen EC2 error when attempting to delete group juju-devstack-0: Error Message: An unknown error has occurred. Please try your request again. [20:22] ok there is a way you should be able to get at the user data from the outside [20:22] and change it if needed, i just cant rember right off [20:22] * imbrandon things [20:23] So, I want to figure out whats wrong [20:23] I have root on the openstack server [20:23] so I'll fiddle around there [20:23] ohh ok locla too [20:23] yea look in /var/lib/cloud [20:23] it will have where it dumped the keys [20:24] before it added them or should have [20:24] as well as logs [20:24] that juju might not have [20:24] where are the juju logs ? [20:24] /var/lib/juju/charm/somewhere [20:25] if your on the box run ps ax [20:25] it will show where the log is getting piped [20:25] <_mup_> Bug #1017248 was filed: EC2 error when attempting to delete group juju-devstack-0 < https://launchpad.net/bugs/1017248 > [20:25] not the charm stuff; the error coming back from ssh [20:25] yea [20:26] there should be a more general zk log too [20:26] ok, this is weird [20:26] in there [20:26] *I* can ssh to the instance [20:26] juju can't [20:26] ahjh [20:26] from the same shell [20:26] that def means the env.y syntaz it fubar then [20:26] ssh ubuntu@10.0.03 works [20:26] syntax* [20:26] it wont try other keys [20:26] ok, so lets see that url [20:26] curl http://169.254.169.254/1.0/user-data [20:27] let me pastbin you my working one privately [20:27] ', /sbin/start juju-provision-agent] [20:27] ssh_authorized_keys: [ssh-rsa AAAAB3NzaC1yc2EAAAABIwAAAQEA728T0gI08lJFqZQo8lDUgKiE860aWTQz+QSeYAFg2T5TrYbGHKt2GHZy+OHYkAhUiSCjZXogFyh1+TRkQIYCcZTNQdOoMtLVesOk9/jRh6ZIcrQvTzbK2KpLXBMhNX9J+HZ5MiAYTZRX9uJSmvDAxrsof2qcVyYBs67hPdE3s5I0Zg5uNm93M9/ciEr+UWTWiIxounHhiEbdW1LIszBlAtvLpsw9bgtB6rRjygiSvoiXMTt00YhWip9PpxBBa6OqtETF/Qu+Uf+guujTnwO9Ue77kNDoocMrZfDsBxlSG6gsByGO/ue7YlRI1w96W68xaGLFl5cgt60SUK1BIVJW9w== [20:27] so you can compare [20:27] /home/robertc/.ssh/id_rsa] [20:27] thats three lines [20:27] but it looks quoted [20:27] that looks right [20:27] hrm [20:27] wow [20:28] I mean, it worked right? I can ssh in... [20:28] ok one sec, let me still get mine, i bet its that damned authorized-keys vs authorized-key-path i came accross the other day [20:28] some things use one and some the other i am willing to bet [20:29] imbrandon: I don't understand though: how does this affect 'juju status' - the ssh environment is *working* [20:29] I just did 'ssh ubuntu@10.0.0.3' and et voila [20:29] right, that was their issue too, liek two days ago, i just unfortunately did not pay enought attn [20:30] hmm, status -v [20:30] -> its connecting to 10.0.0.2. *bong* [20:30] nice [20:30] fixed, I think [20:30] lets try that again [20:30] \o/ [20:30] ok. [20:31] time to do patches, bugs, writeup and then I can do what I actually wanted to be testing. [20:31] but first, a quick break [20:31] http://paste.ubuntu.com/1058118/ [20:32] thats how my keys are layed out fwiw, just for ref later [20:32] if you add more [20:32] obviously thats only the middle of the file, but you get the context [20:34] i *think* you can have it read a file as well like authorized-key-path: ~/.ssh/authorized_keys [20:34] but i have not actually tried that yet [20:37] btw for anyone arround , or that reads this in the backlog, i started a full mirror on a box at hpcloud i set aside just for that reason, apt-mirror should be done here in less than an hour and i'll have it sync very very regularly ( like every hour or so if not more ) as a "local mirror" on the hp cloud [20:38] figured i needed to play upstream and dust off apt-mirror for a little workout and fix some of the miror bugs in bts anyhow and that gives me a good excuse [20:39] point is i'll post the priv IP for it etc etc somewhere we can all refrence it like the mailing list [20:40] to use, and should no problem with trust as i'll put a real ssl cert on it ( got one i'm not useing ) and the keys are all canonicals etc cuz i'm not repackin stuff, real mirror [20:40] i'll make sure it runs at minimum the 3 months HP gave us, if its not too much i may keep it past then too [20:41] or see if they will just sponsor that one node etc if i put the ops time in keeeping it up [20:41] we'll hit that bridge when we need to but for now should greatly speed up apt-get and such [20:42] hrm, and actually it just hit me that apt-mirror would make an excelent charm too [20:43] juju deploy apt-mirror; wait 2 hours and you have a private mirror :) [20:43] could even make it smart and use ec2 on aws and such :) [20:43] err s3 [20:43] well [20:44] otoh yes, otoh we have canonical run mirrors on the major clouds, in each reach,that have free traffic for instances... [20:58] ok, it was fixed in txaws rev 134 [20:58] which we're not running yet [21:01] SpamapS: what do you think the chances of updating txaws in precise are? to get ip address support in Instances (rev 134) [21:04] if we could isolate it to fixes only i'm sure we could make a case and get it tested good enough [21:04] i would think [21:04] ajmitch: ping pong :) here is good too [21:04] :) [21:04] not sure if i'm even in uwire right now [21:05] lol [21:05] so what are you having issues with? it's good timing as I had the fabric script open in front of me right now [21:06] like was having problems groking even a working file at all [21:06] imbrandon: this is what I need: http://bazaar.launchpad.net/~txaws-dev/txaws/trunk/revision/134 [21:06] * ajmitch uses it in a really basic way, checking out the branches [21:06] like hellp world stuff, i'm only ever done capfiles like that [21:06] yea thats what i have the charm doing now [21:06] its checking out the git repo [21:06] and then dumping a config [21:06] into place [21:07] and as well as a cron than looks for git updates to the prod branch [21:07] but i wanna change that to a web post hook on next iteration [21:07] no need to be looking 24/7 running crons like that every 10 min [21:08] webhook that takes a postrecieve from a git hook commit or bitbucket or github webhook etc etc [21:08] and pulls and swaps dir [21:09] ok, what do you have so far? dump in a pastebin or something [21:09] but yea, i like dident even get hello world functioning ,i'm guessing because i was trying to do it liek a cam file and use config/deploy etc [21:09] yea , let me find which branch i was on [21:09] give me like 2 min [21:10] lifeless: yea i'm thinking that LOOKS like it could be sane enough to land in an -update but i often over look other parts so SpamapS could give ya a MUCH better idea [21:11] but we already had juju its self in -updates [21:11] so the team isnt shy of it [21:11] fabric is really quite simplistic, it's just "run this stuff on these hosts" [21:11] doesn't do any of the provisioning magic of juju :) [21:11] yea i was hopign it was a python version of capistrano [21:12] since all thse damn hooks are in python it will make since to use fab, and the ones i write in ruby or php i'll use cap :) [21:12] or say screw it and convince them to build it into juju :) [21:12] * imbrandon looks for the code [21:13] you can probably do similar things with it [21:13]