[01:03] <crazed> hmm
[01:03] <crazed> where does write-mime-multipart come from
[01:03] <crazed> don't see it in the cloud-init repo, cloud-utils or cloud-init package
[01:03] <crazed> on 10.04 or 10.10
[01:03] <crazed> though i do see it on the filesystem of an ec2 image
[01:48] <smoser> on 10.04 and 10.10 it might not be packaged at all
[01:48] <smoser> in 11.04 it is in cloud-utils
[01:49] <smoser> if you see it in the filesystem of the ec2-image, dpkg -S /path/to/file
[01:49] <crazed> ah -s!
[01:49] <crazed> didn't think of that
[01:50] <smoser> it is cloud-init if it is there in 10.10
[01:50] <smoser> and you probaly dont want to isntall cloud-init on your laptop
[01:50] <smoser> (or other non-cloud instance)
[01:50] <smoser> which is why it moved to cloud-utils
[01:50] <smoser> its stand alone, though, you can  jsut grab it from cloud-utils bzr
[01:50] <smoser> i have to go
[01:51] <smoser> later
[01:53] <crazed> oh good call
[13:15] <TeTeT> what would be the best way to give cloud credentials to an instance?
[13:23] <semiosis> TeTeT: credentials for what, exactly?
[13:23] <TeTeT> semiosis: credentials for S3, so the instance can store data there
[13:26] <semiosis> TeTeT: from what i've seen it depends on the tool you're using.  for example, euca2ools would read /etc/euca2ools/eucarc, but another tool (or suite) may look elsewhere
[13:26] <TeTeT> semiosis: yeah, but how to transfer the eucarc and cert files to the instance? Using user data?
[13:26] <semiosis> TeTeT: i use puppet but you could very easily do it with cloud-config
[13:27] <semiosis> TeTeT: ...and a little shell script
[13:27] <TeTeT> semiosis: right, I will consider cloud-config. I want to leave out puppet for now
[13:27] <TeTeT> even though it's very useful for configuring instances
[13:29] <semiosis> TeTeT: couple things you may be interested in: storing the file or script in s3 & just pulling it with cloud-config (using a presigned S3 url), and also using IAM to create an access key with limited rights to put on the instance
[13:30] <TeTeT> semiosis: I don't know about presigned s3 urls and IAM, can you clarify or point me to some docs?
[13:33] <TeTeT> semiosis: nevermind, googled http://docs.amazonwebservices.com/AWSJavaSDK/latest/javadoc/com/amazonaws/services/s3/model/GeneratePresignedUrlRequest.html
[13:34] <semiosis> right on
[13:40] <TeTeT> semiosis: I fear both are not available on UEC yet :(
[13:40] <TeTeT> semiosis: but thanks for your help, learned something new!
[13:40] <semiosis> yw
[14:52] <TeTeT> hggdh: hi, a first result from the runs of my scheduler script, bug 728018. Does it look familiar to you? e.g. is this a usual failure or something extraordinary? in the past I thought my script for capturing the ip was bad, but now it's using boto directly
[14:54] <hggdh> TeTeT, looking
[14:56] <hggdh> TeTeT, I do not remember this error on 10.04 (when I tested it)
[14:56] <hggdh> TeTeT, so I would say you hit a new one...
[14:59] <TeTeT> hggdh: ok, wonder if it's also in 10.10 or 11.04. Unfortunately I cannot upgrade the training cloud, as I lose the ability to reset it then and IS has to intervene
[15:01] <TeTeT> hggdh: i thought it was quite rare, after launching hundreds of instances, but just a few mins ago it failed after 29 instances, so seems to be quite random, though rare
[15:02] <hggdh> TeTeT, I ran a lot of thousands of instances on 10.10, and did not see it. I am not sure we had a  similar setup, though
[15:03] <TeTeT> hggdh: hmm, ok
[15:03] <hggdh> TeTeT, but it is something to try now, on 11.04
[15:10] <TeTeT> hggdh: you've got a testing cloud with 11.04?
[15:11] <hggdh> TeTeT, yes. but euca is right now broken on DHCP
[15:11] <TeTeT> hggdh: oh, ok, so no point for me to upgrade my cloud at home yet and try to test
[15:11] <hggdh> TeTeT, no, not right now
[15:12] <hggdh> TeTeT, I can ping you when euca is workable on 11.04, if you wish
[15:14] <TeTeT> hggdh: yes, would be nice. FYI I've added a basic hadoop workload to the scheduler script, now on to others
[15:16] <hggdh> cool!
[15:18] <TeTeT> hggdh: I also moved the branch on lp to lp:~canonical-pse/+junk/uec-scheduler if you want to peek at it. It's now python/boto and cloud-init shell scripts
[15:20] <hggdh> TeTeT, pulling it now ;-)
[15:21] <TeTeT> hggdh: he he, don't be afraid of bad code ;)
[15:29] <hggdh> TeTeT, I am not, I write them all the time ;-)
[15:29] <TeTeT> hggdh: I'm just writing down some setup script
[15:29] <TeTeT> hggdh: errr, doc
[15:36] <TeTeT> hggdh: pushed, though it's untested :) Have to run an errand now but will be back later
[17:54] <vadi2> Is there a micro instance available for ec2 asia?
[21:01] <jamespage> smoser: around?
[21:01] <smoser> yes sir
[21:01] <jamespage> hey - time for a bit of help with cloud-init?
[21:02] <smoser> gladly
[21:03] <jamespage> OK; so I've been working on expanding the test cases for ec2 testing.
[21:03] <jamespage> And I've started to look at the User Data tests.....
[21:04] <jamespage> originally I had put a test execution into the cloud-config part; however this gets executed before all of the user data scripts
[21:04] <jamespage> so that did not work great when I wanted to test that they all ran OK.
[21:05] <jamespage> question is: can I imply an order in which the user data scripts/data/parts will be executed (I want to run the tests at the end)?
[21:06] <smoser> where di you put it in the cloud-config part ?
[21:06] <smoser> in runcmd ?
[21:07] <jamespage> yes
[21:08] <smoser> runcmd is run in the same way as user-data scripts
[21:09] <smoser> it gets put into a directory and run with run-parts
[21:09] <smoser> so you can manipulate the ffilename of the other user-scripts to cause it to run last (make them named earlier in order)
[21:10] <jamespage> ah - right - I see - let me try that out
[21:10] <smoser> *or* you could run those scripts as a upstart job that runs after stopped-cloud-init-final.. or something.
[21:10] <smoser> the way i ran tests inside the instance was to ssy after the instance was up
[21:11] <smoser> as , if you're testing user-data, you can't really be sure that cloud-nit is going to run the stuff that you put in there.
[21:17] <jamespage> smoser: thanks for you help - that worked just great!
[21:18] <smoser> read ^
[21:18] <smoser> i suggest though not running those things via cloud-init
[21:19] <jamespage> That would be another way around it; would pickup a break in cloud-init better.
[21:19] <jamespage> At the moment my test would time out after 5 minutes of waiting for the test results to appear
[23:16] <lotreck> does anyone have experience using a 10gbit ethernet card on ubuntu lucid ? I just want to know if there are drivers on ubuntu for it . I'm looking at the Intel X520 line.
[23:16] <lotreck> I'd hate to purchase new dell gear and some 10 GB nics only to find out that they don't work together.
[23:17] <jeremydei> hm, this is weird, i just noticed that maverick on ec2 doesn't have /proc/xen/*
[23:17] <jeremydei> anyone know what's up with that?  im particularly interested in the xen clock stuff .. and how it appears to be mia
[23:21] <jeremydei> smoser, am i crazy or is the xen independent wallclock not in maverick ec2?  should be in /proc/sys/xen/independent_wallclock
[23:27] <flaccid> lotreck: this is ubuntu-cloud. you might want to check driver support for the chipset.