[01:25] <oneman> hello
[01:26] <oneman> I really want to use a high-cpu-medium instance on EC2, but It won't let me choose that with the maverick 10.10 ami
[01:26] <erichammond> oneman: You have to pick the 32-bit AMI for c1.medium, not the 64-bit AMI
[01:27] <oneman> Whats the reason behind that?
[01:27] <erichammond> c1.medium is 32-bit.
[01:27] <oneman> I have used 64bit on c1.medium
[01:27] <oneman> let me double check
[01:27] <erichammond> oneman: No you haven't :)
[01:27] <erichammond> c1.medium is a nice machine, though.  Good performance as long as you don't need lots of memory.
[01:27] <oneman> indeed
[01:27] <oneman> oops ;P
[01:28] <oneman> I'm actually trying to do something new and crazy
[01:28] <oneman> Run jackd on the cloud
[01:28] <oneman> 1024 sample period is fine tho
[01:28] <oneman> so its not super tight..
[01:29] <erichammond> oneman: I just read about JACK yesterday.  My son is interested in connecting his MIDI keyboard to Ubuntu and I have no idea how to set all that up.
[01:29] <oneman> start jack, plug in the keyboard, and then start a program that can make use of the keyboard
[01:30] <oneman> I've only used a generic akai drum machine pad thing... with hydrogen drum machine and jack and ardour
[01:30] <erichammond> oneman: I'm lost at the "plug in" point.  Not sure if I have to get a special audio card with MIDI support that works with Ubuntu.
[01:30] <oneman> qjackctl
[01:30] <oneman> is it not usb?
[01:31] <erichammond> oneman: If that works, I could buy a MIDI to USB cable.  Or perhaps we should get a more modern keyboard.
[01:31] <oneman> Heres the basic info, you use qjackctl gui program to launch/start/stop jackd, you use ardour for recording/playback, and then whatever other programs to work with the midi keyboard / drum machine etc
[01:32] <oneman> yeah, midi to usb cable sounds good to me
[01:32] <oneman> anything recent will be midi via usb tho
[01:32] <oneman> let me ask you another thing about the cloud, I could have shot myself in the foot here but luckily I had only messed with the instance for about an hour
[01:32] <erichammond> Thanks.  I took us offtopic, but it's a slow channel.
[01:33] <oneman> So, I loaded up that AMI, it had EBS as / , so I expected if I terminated the instance that I would still have my ebs volume
[01:33] <oneman> but it got deleted when I terminated!
[01:33] <oneman> How do I clone it or something ?
[01:33] <erichammond> terminate: you lose any volumes that EC2 created for you automatically on startup
[01:33] <erichammond> stop: all the volumes are kept.
[01:34] <oneman> But if the volume was not created by ec2 and I terminate it still sticks right?
[01:34] <erichammond> right
[01:34] <oneman> could I snapshot the volume, then restore it to another volume in order to "copy" it?
[01:35] <erichammond> There is an override option that goes either way.
[01:35] <oneman> where is this over ride :!
[01:35] <erichammond> Check out #2 in http://alestic.com/2010/01/ec2-instance-locking
[01:37] <erichammond> Yes, you can snapshot the root EBS volume, register it as an AMI, and start a new instance from that.
[01:37] <erichammond> It is recommended that you initiate the snapshot while the instance is stopped to reduce risk of file system inconsistencies.
[01:37] <oneman> ok great
[01:37] <oneman> thats really what I want
[01:38] <erichammond> There is an API / command line that does all of this for you (including stopping/starting the instance)
[01:38] <erichammond> er, does all of this through registering the AMI.  It does not start a new instance.
[01:38] <oneman> glad I figured this one out before setting up my environment
[01:38] <oneman> hehe
[01:39] <erichammond> I still recommend starting from a fresh public AMI and running an automated script to set up your software and configuration whenever this approach might work.
[01:39] <erichammond> This lets you keep up with the latest AMIs and makes it easier to switch to new OS releases as they come out.
[01:40] <erichammond> In place upgrades are supposed to work, but I almost always have issues.
[01:40] <erichammond> I also recommend keeping your data on a separate EBS volume.
[01:41] <oneman> lemme ask you this, I set myself up a development server at home, also 10.10, and except for the programs I compiled myself (which are the ones I'm writing anyway), I installed a load of dependancies (luckily all came from standard ubuntu repos as well), whats the easiest way to just clone this list of packages, so I don't have to just try to compile and see what I'm missing 10 times over  ( and also have an exact knowled
[01:41] <oneman> ge of my actual dependencies )
[01:42] <oneman> All my usage of the cloud so far has been through engineyard's sort of web interface to ec2
[01:43] <oneman> so thats why I'm kind of half way on my cloud knowledge
[01:45] <erichammond> oneman: I think I've used this in the past: dpkg --get-selections
[01:45] <oneman> is it feasable to mount /home as your seperate ebs volume
[01:45] <erichammond> then on the target machine : dpkg --set-selections < OUTPUTFROMABOVE ; apt-get dselect-upgrade
[01:45] <oneman> or do you reccomend a /data or /app
[01:46] <erichammond> oneman: Yes.  I mount an EBS volume on, say, /vol then mount --bind directories from there over top of /home, /etc/lib/mysql, /etc/log/mysql, etc.
[01:47] <oneman> ok
[01:47] <erichammond> See: http://ec2ebs-mysql.notlong.com
[01:48] <oneman> I'm a postgresql user ;p
[01:48] <erichammond> oneman: See that article for the principles involved.
[01:48] <erichammond> in moving data around on EBS and mounting it back.
[01:48] <oneman> ok thanks alot
[01:49] <oneman> this is all exactly the information I needed
[02:01] <oneman> 2 upgraded, 1267 newly installed, 0 to remove and 0 not upgraded.
[02:01] <oneman> Need to get 919MB of archives.
[02:01] <oneman> After this operation, 2,980MB of additional disk space will be used.
[02:01] <oneman> exciting!
[02:01] <oneman> haha
[03:39] <mwhudson> it seems that starting ami-ec1aea85 (today's 64 bit instance store in us-east-1) on an m1.xlarge instance doesn't work very well
[03:40] <mwhudson> [1122470.505234] Kernel panic - not syncing: Attempted to kill init!
[03:40] <mwhudson> in the console outptu
[03:40] <erichammond> mwhudson: Please submit a bug report in launchpad.net
[03:40] <mwhudson> erichammond: which project?
[03:44] <mwhudson> ah ubuntu-on-ec2 i guess
[03:45] <mwhudson> ah bah, the second one i tried worked fine
[03:45] <mwhudson> yay reliability :(
[03:45] <flaccid> you could still raise the bug as intermittant and show the whole console output
[03:46] <mwhudson> yeah
[06:14] <oneman> ok so I use a public ami, then I snapshot the ebs vol, create a new volume from that, attach that to the original instance (when it was stopped) and it won't boot
[06:14] <oneman> what noobie mistake am I making
[06:39] <oneman> I set it to /dev/sda instead of /dev/sda1
[06:40] <oneman> hard to believe it let me shoot myself in the foot like that ;p
[08:05] <ubuntucloud845> hello
[08:05] <ubuntucloud845> can i auto back up file server to another server
[08:05] <ubuntucloud845> same ubunto 10.10
[11:47] <superxgl> hi all, i used hybirdfox, and my image is based on i386, but why hybirdfox shows the architecture is X86_64??
[12:07] <flaccid> superxgl: check the api
[12:07] <TeTeT> superxgl: maybe it has been uploaded as x86_64, check the output of euca-describe-images for that image
[12:10] <superxgl> IMAGE   eri-89321725    centos-ramdisk-buket/initrd.img-2.6.24-19-xen.manifest.xml      admin   available       public          x86_64  ramdisk                instance-store
[12:10] <superxgl> IMAGE   emi-CD7E14B8    centos-image-buket/centos.5-3.x86.img.manifest.xml     admin    available       public          x86_64  machine eki-27E215E0    eri-89321725    instance-store
[12:10] <superxgl> IMAGE   eki-27E215E0    centos-kernel-buket/vmlinuz-2.6.24-19-xen.manifest.xml admin    available       public          x86_64  kernel                  instance-store
[12:11] <superxgl> oh, what happened?
[12:11] <superxgl> i used the eucalyptus's image
[12:11] <TeTeT> superxgl: it's been stored as x86_64 as the output says
[12:12] <superxgl> why? but my image is i386
[12:12] <TeTeT> but should be not a problem, I'd continue testing if I where you
[12:12] <flaccid> thats what it was registered as
[12:12] <TeTeT> superxgl: when you upload the image you can specify which architecture it is
[12:12] <superxgl> hmm..
[12:12] <superxgl> i have to do this ?
[12:13] <superxgl> i follwed the guideline and it did not say that
[12:13] <TeTeT> superxgl: if it comes from the image store and you think it's an error, please file a bug in Launchpad against the image store
[12:13] <TeTeT> superxgl: which guideline?
[12:14] <superxgl> http://open.eucalyptus.com/wiki/EucalyptusInstallationSource_v2.0
[12:15] <superxgl> but when i loged in the instance ,  "uname -a" shows that it is i386
[12:16] <flaccid> sure thats because the kernel is i386
[12:16] <flaccid> the image was registered as x86_64 in the api
[12:16] <superxgl> hmm...
[12:17] <superxgl> flaccid: so where to check the api?
[12:19] <superxgl> TeTeT:  when i upload the image how to specify the architecture ? sry, im new to this
[12:19] <flaccid> superxgl: http://open.eucalyptus.com/wiki/Euca2oolsUsingOverview_v1.3 http://open.eucalyptus.com/wiki/Euca2oolsImageManagement_v1.3
[12:22] <superxgl> im checking now
[12:42] <superxgl> oh, i found the problem as "euca-bundle-image --help" shows :
[12:42] <superxgl> -r, --arch                      Target architecture for the image ('x86_64' or 'i386' default: 'x86_64').
[12:43] <flaccid> right
[12:43] <flaccid> and keep in mind euca-register
[12:43] <superxgl> i will ..tnx very much for all of ur help :)
[12:43] <flaccid> np
[13:07] <kim0> Hi folks, any help answering questions on the forums is appreciated http://ubuntuforums.org/forumdisplay.php?f=392
[13:26] <willem> I want to create a tar.gz of some directories in an S3 bucket. I have got that mounted using s3fs. And it seems to go fine, however after some time it fails with the following message: gzip: stdout: No space left on device
[13:45] <kim0> willem: seems like your output directory has run out of space
[13:46] <willem> @kim8 it is an s3 bucket, which I do have write access to...
[13:47] <willem> I am just thinking... I might be going over the maximum file size....
[13:49] <kim0> willem: why don't you create the tgz file locally .. then upload that
[13:49] <willem> Because all the data is already on the S3, lots of small files...
[13:50] <willem> That is why I want a tar. To be able to download and verify the download easily.
[13:51] <kim0> so, you have many little files on S3, you want to compress those files into a tar file that is written to local disk ?
[13:54] <kim0> willem: ^
[13:59] <kim0> is walrus throughput as slow as 5MB/s ?
[14:05] <willem> No, I want to compress (or at least tar them) into one file. So that other people can download it from s3.
[14:31] <kim0> willem: I would suggest compressing them into a local .tar.gz file first, then uploading that to S3 as a second step
[14:32] <willem> kim0: ok, currently trying zip using a split at 4gig. If that doesn't work, I will try it your way :)
[14:32] <willem> thanks for the help!
[16:58] <elasticdog> possibly a stupid question, but can you use `euca_conf` to set arbitrary variable values?
[16:58] <elasticdog> it seems like there are flags for setting a few specific things, and --name to read any variable's value, but how would you set something like MAX_CORES...just add it by hand?