=== niemeyer_ is now known as niemeyer [01:33] SpamapS: around ? [01:33] SpamapS: can you perhaps lurk in #maas, plenty of cross over i suspect, and we get lots of juju q's in there. [05:15] I have officially figured out the killer Juju app, although I doubt I can get the package maintainers to make one [05:15] John the ripper + MPICH2 [05:15] build a charm for that and you'd have an instant password cracking cluster, and every infosec junky would have it in their private cloud [05:54] EvilMog: I'll take that challenge [05:54] that would be awesome [06:01] EvilMog: they're both packaged, so thats at least easy [06:01] ypi [06:01] you'll want to use the latest jtr source [06:02] the packaged one doesn't have mpich support [06:02] need to edit the makefile uncomment a few lines to add it in [06:08] actually I'll be honest I'd even be happy with an auto mpich setup and pre-setup the hosts files and node information for mpich2 [06:08] EvilMog: why isn't the latest one packaged? [06:08] because it changes so often [06:08] ah so you're just one of those devs who thinks "latest" == "best" ;) [06:08] what you'd want is the latest code-train with unofficial jumbo-5 patch [06:09] theres 2 releases, theres the stable but missing tons of features, then theres the jumbo patch which supports MPI and way more password formats [06:10] I'm just a heavy john user, and virtually every cluster I've built has used the latest code train with most recent jumbo release [06:10] this also greatly improves crack performance [06:10] EvilMog: it just baffles my mind that an OS released < 2 months ago would have something so out of date [06:11] the debian base package uses the stable tree [06:11] EvilMog: sounds to me like john+mpi is a fork [06:11] its not [06:11] its an official build [06:11] theres 2 releases stable and unstable [06:11] the debian base package uses stable [06:12] I only see "official free" and "community enhanced" [06:12] same thing [06:12] official free = stable [06:12] which is whats packaged [06:12] community enhanced is where the good stuff is [06:13] including all the performance enhancements [06:14] Ok https://launchpad.net/~pi-rho/+archive/security seems to have jumbo-5 packaged [06:14] nice [06:14] it never used to [06:14] just have to copy that source package and verify the sources [06:14] so the only thing you neeed to do then is verify they compiled it with mpich support [06:14] http://openwall.info/wiki/john/parallelization [06:15] theres 2 lines that need to be uncommented in the make file [06:15] https://launchpadlibrarian.net/107245212/buildlog_ubuntu-precise-amd64.john_1.7.9-jumbo-5-1ubuntu1~ppa2~p_BUILDING.txt.gz [06:15] CC = mpicc -DHAVE_MPI -DJOHN_MPI_BARRIER -DJOHN_MPI_ABORT [06:15] MPIOBJ = john-mpi.o [06:15] no mention of mpich [06:16] morning [06:17] DEB_MAKE_EXTRA_ARGS := SYSTEMWIDEFLAGS="-DJOHN_SYSTEMWIDE=1 -DJOHN_SYSTEMWIDE_EXEC=/usr/lib/john" [06:17] EvilMog: looks like it will need some work [06:17] imbrandon: alo [06:17] yeah, I'm working on a build script [06:17] thinking wget the source, apt-get install the pre-reqs [06:17] imbrandon: You'll be excited to know I'm uploading PHP 5.4.4 to quantal right now [06:18] find and edit makefile [06:18] sweet [06:18] then make and install package [06:18] so if the mpich2 environment is pre-setup I can just call the build script as a post install [06:18] gonna try for backport ? [06:18] ahhaha [06:19] imbrandon: you already know what that will be the problem [06:19] yup [06:19] EvilMog: *edit* the makefile? [06:19] EvilMog: no wait to configure it w/ cmdline args? [06:19] -CFLAGS = -c -Wall -O2 -fomit-frame-pointer -I/usr/local/include $(OMPFLAGS) [06:19] +CFLAGS = -c -Wall -O2 -fomit-frame-pointer -fPIC -I/usr/local/include $(OMPFLAGS) $(SYSTEMWIDEFLAGS) [06:20] thats the diff the package has [06:20] SpamapS: is there a little SpamapS yet ? hehe [06:21] imbrandon: 2 weeks old now [06:21] oh wow, i'm so behind [06:21] Congrats man [06:21] imbrandon: and this would be the 3rd child process I've forked [06:21] that works too [06:21] :) [06:22] I'm just a very hacky analyst who does things the hard way out of ignorance [06:26] http://download.openwall.net/pub/projects/john/contrib/mpi/John_the_Ripper_on_a_Ubuntu_10.04_MPI_Cluster.pdf [06:27] EvilMog: the GPG key that openwall uses is pretty crusty [06:27] if that helps [06:27] more than likely [06:27] pub 1024R/295029F1 1999-09-13 [06:28] ick [06:28] yeah its like they don't care [06:28] 1999 [06:28] you'd think the authors of hash cracking software would give a damn :P [06:29] EvilMog: actually I'd think the uathors of hash cracking software would run a single evil mirror just to see if anybody checked [06:29] sadly though I have to sneak off to work, I'll be back on in about 8 hours [06:29] you know that would be entertaining [06:30] I have a reasonably well filled out web of trust here.. and I can't find any Debian developers who have signed it yet [06:30] time for fooood, my tummy is growling [06:30] or even any transitive [06:36] Wow finally I found one like 3 levels deep thats in my keys === zyga-afk is now known as zyga [06:56] mmm coco-puffs [06:59] EvilMog: looks to have built fine [07:52] alright.. skeleton john the ripper charm done [07:52] complete w/ mpich2 scale-out (hopefully) [07:53] If I can add gluster/ceph .. I am a god ;) === fenris is now known as Guest13007 [08:51] EvilMog: lp:~clint-fewbar/charms/precise/john/trunk .. needs to setup the SSH keys between hosts, but mpi kinda sorta works [08:51] thinking this should just be a generic 'mpich2' charm [08:51] * SpamapS goes to bed [12:41] jml: the "ubuntu" charm I just recommended to you on askubuntu had some problems yesterday, I don't know if we fixed it [12:42] jcastro: what was wrong with it ? [12:48] jamespage, the Juju on LXC issue still giving issues [12:49] in one day it created good GB of logs [12:50] just trying to find documentation on how to set up environments.yaml for ec2 [12:50] looks like it's not in the package docs or on the wiki. trying code tree instead. [12:54] jml: https://juju.ubuntu.com/docs/getting-started.html#configuring-your-environment-using-ec2 [12:55] jml: https://juju.ubuntu.com/docs/getting-started.html#introduction [12:55] imbrandon: it didn't launch [12:55] ta [12:55] jcastro: how would I have found that from https://juju.ubuntu.com? [12:55] jcastro: ahh [12:55] on the Documentation link [12:55] but you're right, the webpage sucks [12:56] we're supposed to have a new one [12:56] jml: there is a Documentation link on the right, but its not clear, i'm gonna try and fix that today [12:56] I wonder if we can get rid of all that marketing crap on it and put the useful information [12:56] * jcastro will ask [12:57] what are control-bucket and admin-secret? [12:57] look at the note underneath [12:58] Note If you already have an AWS account, you can determine your access key by visiting http://aws.amazon.com/account, clicking "Security Credentials" and then clicking "Access Credentials". You'll be taken to a table that lists your access keys and has a "show" link for each access key that will reveal the associated secret key. [12:58] oh [12:58] I see what you mean [12:58] those are generated for you [12:58] its a s3 bucket name [12:58] if its not made it will make it for you [12:59] so I don't have to specify those? [12:59] yes you have to specify those but the are filled in if you followed the doc [13:00] imbrandon: oh right. I followed another, different doc first that had me specify a local environment. [13:00] fun times, eh? [13:00] hehe [13:00] right :) [13:01] ugh, this doc needs some work [13:01] basicly they can be anything as long as they are unique to the env [13:01] this LXC section sucks, I'll work on it today [13:01] jcastro: yes , yes it does, i've been doing bitesize chunks as i had time [13:01] heh [13:02] but the juju.ubuntu.com wiki needs fixed as well [13:02] can you get me access to edit that ? [13:02] how do I set a default environment? [13:03] jcastro: also, I get this on 'juju deploy ubuntu': Error processing 'cs:precise/ubuntu': entry not found [13:03] jml: add default: env-name to the top [13:05] imbrandon: thanks. [13:06] jml: try "juju deploy cs:~charmers/precise/ubuntu" [13:06] thanks SpammapS I'll give it a shot in the morning shortly [13:07] imbrandon: same problem, except cs:~charmers/precise/ubuntu [13:08] hrm [13:08] yes, we noticed yesterday it was broken [13:08] oh, that's what broken meant. [13:08] jml: no unfortunately the wiki is acl'ed to ~charmers because we enable html to put whiz bang things on it [13:09] jml: you should be able to bzr branch the ubuntu charm and deploy it manually [13:09] bzr branch lp:charms/precise/ubuntu [13:09] jcastro: i'm in ~charmers and cant edit [13:09] then "juju deploy --repository . local:ubuntu" should work [13:09] imbrandon: huh, really? [13:09] yea i tried to fix it up some over the weekend and it said i had no rights [13:10] jcastro: thanks. [13:10] imbrandon: k, I'll ask to fix that [13:10] jml: when clint wakes up I'll ask him what's up with that [13:10] jcastro: rockin let me know, cuz i have time to do some fixin on it today if it get worked otu [13:10] out* [13:11] imbrandon: lp:juju/docs could always use love in the meantime. :) [13:11] yup already have that open :) [13:11] been working on the navigation [13:11] err secondary nav [13:17] jcastro: we should just have sphinx generate the whole juju.ubuntu.com page , its controled by the same acl ... woudlent take that much to add the front page stuff in [13:17] hmmm.. [13:17] err i guess its not [13:18] morning hazmat [13:19] hazmat: what ya think ? wouldent be too much work [13:20] pkgme.net is doing it that way i see [13:20] :) [13:21] heh [13:21] * imbrandon waves to jml :) [13:21] jcastro: have updated the question with the qualifiers to my success. [13:21] I think I'm going to write something to do this for me :( [13:22] jml: wait for smoser or someone else on the server team to respond [13:22] you can't be the first guy to want to do this [13:22] jml: i hadent looked at your ask question yet but there are helper tools in juju-jitsu and charm-tools [13:23] imbrandon: thanks. I'm not going to look those up right now, I'm afraid. [13:23] jcastro: good call. [13:26] what are we responding to, jml [13:26] jcastro, [13:26] smoser: http://askubuntu.com/questions/153029/tool-to-fire-up-ec2-instance-and-ssh-in [13:27] i swear, this seems like a trap [13:28] no, a proper trap would include getting you to maintain something new [13:28] imo ubuntu-ec2-run does the correct thing and lets you create security groups and pass them through. [13:30] jml, see kirkland's cloud-sandbox in bikeshed [13:31] hazmat: ping, any idea why the copyright at the bottom of juju.u.c/docs wont show when generated automaticly , but when i run "make html" local it shows hust fine , there is a huge diffrence in the sphinx version , maybe thats it ? [13:31] s/hust/just [13:31] but having a tool create a security group and upload ssh keys for you seems overkill to me. you might as well just do those things (euca-create-group, euca-import-keypair) and then use them. having each tool do that for itself doesn't make sense. [13:35] smoser: I guess I don't see why I should have to think about those things at all when a tool could just create a disposable security group and keypair and then clean up after itself. [13:38] imbrandon, dunno offhand [13:38] kk [13:39] imbrandon, i imagine some delta between the build steps.. but i didn't setup the j.u.c/docs build [13:39] i'll just pull the if/else out for now, was going to move the js to the bottom anyhow [13:39] imbrandon, i did setup this one.. http://jujucharms.com/docs/ and the copyright shows [13:39] rockin, kk yea likely a delta in the sphinx version then [13:40] imbrandon, yeah.. 0.6.4 vs 1.1 [13:40] and thats old :) [13:40] heh [13:40] well, you're certainly welcome to write such a tool. patches to ubuntu-ec2-run would be welcomed, although i think it would actually better fit in the bikeshed tool. the reason being that ubuntu-ec2-run is a single "run something" command rather than a long lived environment command (which would be required to properly manage security groups and ssh keys. === niemeyer_ is now known as niemeyer [13:42] hazmat: btw i did split out the theme into its own [13:42] lp project under .... one sec [13:43] https://launchpad.net/ubuntu-community-webthemes/light-sphinx-theme [13:43] smoser, a flag to just enable web and ssh seems like a nice option without mucking with security groups, ditto for lp id or manual specify ssh key for a key to use instead of keypairs.. [13:43] its not about long lived, its about convience. [13:43] patches welcome i know ;-) [13:43] well, it is long lived. [13:44] if you're making temporary things, then you have to clean up those things. [13:44] and that means you have to be around when the instance terminates to cleanup [13:44] smoser, you mean the sec group as a transient thing to clean.. fair enough [13:44] Hi, um, I'm trying to deploy a blank unit for charm development, but I'm getting 'not found' on both cs:precise/ubuntu and cs:oneiric/ubuntu. Any idea what I'm missing? [13:45] imbrandon, cool [13:45] hazmat, or keypairs if you were uploading them (import-keypair). but thats not necessary, really. === zyga is now known as zyga-afk [13:45] (and you're right that 'ssh-add -L' probably is going to get you the right thing by default) [13:48] so i guess the security group is the only thing that would need to be temporary, and fwiw, the 'default' security group can be modified to include 'ssh' [13:48] at which point, you [13:50] there's also some prior art in lp:lp-dev-utils which creates security groups and keypairs, and cleans them up on following runs. [13:50] you'd have nothing to clean up, and ubuntu-ec2-run "just works" . the one command extra that jml had to type would be "euca-authorize default -p 22". [13:51] in the end, i use the 'default' group [13:51] so tha tisn't an issue for me, and i have wrappers that specify '--key brickies' (which i agree would be a nice add to ubuntu-ec2-run) [13:52] why they default to the 'default' security group, but provide no way to default to keypairs is strange. [14:05] mgz, ping [14:05] pong [14:05] mgz, would you mind using the lbox command to submit the ostack branch? [14:05] mgz, it should be a simple thing of just installing lbox , and lbox propose from within the branch [14:05] I'm on it, battling through go being weird about things [14:06] I believe I have it defeated now though [14:06] mgz, k, ping me if you need any help wrt [14:06] mgz, slay the dragon :-) [14:06] hm, don't have exp/html [14:08] hosted on google code after it got removed for 1.0 apparently === hamptonpaulk_ is now known as hamptonpaulk [14:15] mgz, huh.. are you installing the package or from bzr? [14:15] it should be staticallly linked in the lbox bin [14:16] I'm trying to build the tools myself [14:16] which has been... interesting [14:24] okay, done. amusing turns out you need a trunk copy of the go source anyway. [14:26] ...and the lbox webbrowser launching script is broken? [14:27] ...ah, nope, just couldn't cope with giving up the term to lynx. [14:29] ...unknown files doesn't mean the branch in unclean [14:30] hazmat: done, not sure what exactly, but something. it posted a comment as me at least. [14:31] mgz did ask you for google login info? [14:31] nope. [14:31] mgz first time round i'd run with -v.. it is reentrant/idempotent incidentally [14:32] mgz.. oh.. sorry it needs a flag to enable the reitveld integration.. [14:32] lbox propose -cr -v === zyga-afk is now known as zyga [14:32] ah, now I see that the subcommands have --help too === salgado is now known as salgado-afk [14:37] ...and it failed in a fun manner [14:37] 2012/06/19 15:35:55 error: Failed to send patch set to codereview: Issue creation errors: {'subject': [u'Ensure this value has at most 100 characters (it has 204).']} [14:40] how do you spell a string slice in go... [14:51] dammit, there are two rietveld.go files, edited the wrong one [14:56] hazmat: really done now. === bcsaller1 is now known as bcsaller [14:59] how close is juju to being ready for production use? [15:00] mgz thanks [15:02] zirpu: define "production" [15:03] used to build a set of instances for running a service. mostly i want to setup a set of mongodb shards and keep it running. [15:03] zirpu: juju runs this -> http://www.omgubuntu.co.uk/ [15:04] I guess my suggestion would be to try it in a test environment, let it run a bit, and if you are satisfied with it...roll it out [15:05] i'll troll the bug list too. that should help me avoid gotchas etc. [15:07] s/troll/trawl/ oops. :-\ [15:07] troll is valid, it might just be misconstrued [15:15] zirpu: we've been developing a set of practices for long-running juju environments... writing them up is stil tbd for this cycle tho [15:17] zirpu: most is prettymuch what you'd expect... freeze your charms, freeze juju.. the rest really depends on the number of people who're maintaining and need access/control === salgado-afk is now known as salgado === zyga is now known as zyga-afk [16:28] negronjl: m_3: hazmat: we should talk velocity/oscon demo [16:28] yup [16:29] jcastro: now? irc meeting in #ubuntu-meeting atm that I could hang "over" [16:29] I think today sometime would be awesome [16:31] jcastro: see what's good for negronjl... I'm just charmpiloting the rest of the day [16:31] * jcastro nods [16:32] whoa, holy docs queue batman [16:40] http://puppetlabs.com/blog/module-of-the-week-puppetlabs-openstack/ [16:40] jcastro: I was running off of charm-tools' review-queue [16:40] now I see the docs queue [16:41] I don't think juan has fixed the bug in the cli tool [16:41] we only found it yesterday [16:41] * m_3 face palm [16:43] jcastro, m_3: I'm here [16:44] jcastro: what bug in which tool ... I can fix it but I've been working on the maas demo [16:44] your CLI review checker [16:44] negronjl: no biggie... jorge was mentioning the docs queue and I realized I was looking at the charm review-queue which didn't show them [16:44] jcastro: what's the issue ( or bug number ) [16:44] I can fix it [16:45] https://bugs.launchpad.net/charmworld/+bug/1002977 [16:45] m_3: thx [16:45] <_mup_> Bug #1002977: Review queue should list lp:/juju/docs < https://launchpad.net/bugs/1002977 > [16:45] negronjl, i put some notes in the bug about what i had to change [16:45] negronjl: when's a good time to hang on node/mongo stuff? [16:45] kapil left notes there [16:45] negronjl, its slow.. [16:45] for interactive usage [16:46] hazmat: m_3 is going to fix the cli tool [16:47] I see the comment on the fix [16:48] m_3: i can hangout [16:48] jcastro: ? [16:49] count me in! [16:49] nodejs ? /me wants on [16:50] I'll start the hangout [16:50] spinning up the other machine (cam's still borked on this one) [16:56] imbrandon: invited, come on in [16:57] had my headset off [17:03] bah [17:03] my connection keeps resetting [17:04] i'm just gonna not hang this time :( but yea m_3 hit me up on the node monitoring stuff after [17:04] if ya like [17:05] imbrandon: cool [17:21] negronjl: martin's in china so I guess that means "sure, keep it." [17:22] jcastro: lol [18:04] hi - is anyone else having trouble finding cs:precise/nova-cloud-controller in the ubuntu store? [18:06] m_3, you going to strange loop? [18:07] hazmat: not planned [18:07] hazmat, i submitted a talk for strangeloop but it was not accepted [18:09] jimbaker, bummer [18:11] dang [18:11] that sounded like a shoe in === koolhead17 is now known as koolhead17|afk [18:13] jcastro: we just need to keep trying... juju'll gradually get more notice [18:13] jcastro: gluecon rejected my proposed talks even though the conference is right on target [18:13] yes, we will go until mims is like superdiamond on all airlines [18:13] jcastro: the cool part was that there was enought mention of juju during the conference that I'll think we'll get in next year [18:14] lol [18:14] that pattern's not suprising with new stuff [18:15] unless it's huge commercial things with big marketing budgets (i.e., not juju) [18:15] japage: http://jujucharms.com/tools/store-missing [18:15] japage: it is, indeed, missing [18:15] thanks for the link SpamapS! [18:16] japage: we are working on getting more visibility into the charm store import process [18:16] when I say working, I mean, we are aware of the problem [18:16] I don't think anybody is actually working on it directly. :-P [18:16] i've asked relevant folks about it [18:16] japage: the charm store isn't really useful for production anyway.. since you can't fix any charm bugs [18:16] i'll see if i can get a more proactive tooling around the issue [18:17] SpamapS, you can publish a fix with a new version at the store [18:17] what we really need isn't switch-charm' but 'fork-charm' [18:17] hazmat: *I* can, but Users* cannot [18:17] switch-charm still seems like the right alt solution [18:17] SpamapS : hazmat : no worries, i was giving it a test run doing evaluations of all of the different cloud stuffs. [18:17] SpamapS, what's fork-charm? [18:17] hazmat: production doesn't really involve letting third parties control your destiny at 3am :) [18:18] SpamapS, their not upgrading your systems at 3am ;-) [18:18] whats charmload expecting anyhow, i couldent get it to populate mongod [18:18] a deployed charm is cached within the env, the store isn't consulted again unless your upgrading. [18:18] hazmat: fork-charm would download the charm, unpack it, bump the revisioand switch the charm [18:18] imbrandon, charmload? [18:19] from charmstore [18:19] hazmat: In the case where your site is broken at 3am, and the fault is found to be the charm, you must be able to fix the charm, and upgrade then [18:19] charmload and charmd [18:19] SpamapS: isn't that what destroy-environment is for? :) [18:19] hazmat: fault isn't always found at the time changes are made :) [18:19] SpamapS : what kind of refresh rate does that tool tend to see? [18:19] japage: 15min [18:19] I think [18:19] cool [18:20] juju is a lot of magic [18:20] nah, just smoke, mirrors, and python [18:20] the charms are the magic! [18:20] lots of python [18:21] thanks guys, super appreciate the help [18:22] hazmat: there is like -0- documents and the only info is "run charmload " after 3 hours nothing in the db still [18:22] and charmd seems to run a store, but empty without i'm guessing charmload [18:23] /charm-info?... etc all "work" but are empty [18:23] imbrandon, oh.. this is the store impl [18:24] something definitel wrong with this xen box + tmux + irssi .. keyboard/buffer/something lags out after a few weeks [18:24] screen ftw :) [18:24] imbrandon, i set it up once.. just to experiment, it was pretty hands off afaicr [18:24] * hazmat digs it up again [18:25] except it picked the same db collection that i was using for the charm browser [18:25] yea like it shows a ton of stuff trying to get from getBranchTips, but never fills in the mongodb [18:25] imbrandon, it populates the juju db / charms collection in mongodb [18:25] do i need to have created that db first ? i wouldent think so [18:26] so, when something is MIA in the charms store, is there another way to go get that code and package it similarly for use? or should i just be patient? [18:26] perhaps we should be running our own parallel charm store to see if it has the same problems and results as the production one :) [18:26] SpamapS: heh i was just trying that [18:26] japage: bzr branch lp:charms/precise/foo [18:26] or was getting there [18:26] heh [18:27] SpamapS: uuuh, why does this should a 0 line delta? :P https://code.launchpad.net/~charmers/charms/precise/sbuild/trunk/+merge/110934 [18:27] awesome [18:27] comparing the two trees clearly shows my changes. :P [18:27] japage: put that in a dir named 'precise' under another dir which you can name anything (mine is /home/clint/charms) and then set JUJU_REPOSITORY=/home/clint/charms ... it will now be local:charms/precise/foo (assuming metadata.yaml has name: foo) [18:28] imbrandon, charmload localhost:27017 is all i do [18:28] it starts loading and spewing stuff on the console [18:28] SpamapS: you rock. thanks! [18:28] kees: well, thats backwards for one (lp:charms/sbuild into your branch) .. but.. perhaps they already got merged? [18:28] hrm i dident add the port , let me try that, but yea i get a ton on console [18:29] imbrandon, if its writing the console its writing to mongo [18:29] its got a hardcoded db name of 'juju' [18:29] http 27018 on the web shows nothing in the db tho [18:29] one sec [18:30] imbrandon, the data is there, try the cli [18:31] http://15.185.100.228:28017/ [18:32] SpamapS: ah, you're right. okay. fixed: https://code.launchpad.net/~kees/charms/precise/sbuild/trunk/+merge/111080 [18:33] imbrandon, http://15.185.100.228:28017/listDatabases?text=1 [18:33] http://paste.ubuntu.com/1049617/ [18:33] imbrandon, use the mongo cli [18:34] or a real front end, the data is there [18:34] k [18:34] hazmat: this is why i said it isnt [18:35] http://15.185.100.228:9090/charm-info?charms=cs:~charmers/precise/memcached [18:35] but i know it loaded that key already [18:36] imbrandon, ic.. it failed processing it because you have the ssh accepting all clients and it doesn't like the output [18:36] er. all server fingerprints [18:37] erm? [18:37] imbrandon, ie you have something in your ~/.ssh/config like UserKnownHostsFile=/dev/null [18:37] yes [18:37] it doesn't like that [18:37] because it pollutes the cli output its parsing [18:37] heh wow, ok [18:37] ahh [18:37] * hazmat just figured that one out as well [18:37]