| flaccid | where is that webpage that shows the latest published kernel images ? | 00:24 |
|---|---|---|
| SpamapS | erichammond: good call.. almost done 23 minutes into the build on a c1.medium. :) | 00:29 |
| erichammond | SpamapS: :) | 00:30 |
| erichammond | flaccid: By "latest published" do you want to see temporary testing kernels or do you mean the latest officially supported kernels? Also, do you mean kernel images themselves or the AKI/ARI ids for EC2? | 00:31 |
| erichammond | flaccid: If you want the AKI/ARI ids for the officially supported kernels on EC2, I think you can just look at the AKI/ARI ids associated with the latest officially supported AMIs on EC2. | 00:33 |
| flaccid | erichammond: i think all of the above please :) i need to bookmark these URLs | 00:33 |
| flaccid | i recall a manifest or something gets published | 00:33 |
| erichammond | flaccid: You might want to browse http://uec-images.ubuntu.com/query/ | 00:35 |
| flaccid | ah yes /query thats what i missed, thanks eric! | 00:35 |
| erichammond | flaccid: For image files, you should be able to build the URL paths from the info in those files. | 00:35 |
| flaccid | http://uec-images.ubuntu.com/query/lucid/server/released.txt awesome | 00:36 |
| flaccid | thanks again! | 00:36 |
| erichammond | thanks to smoser | 00:36 |
| smoser | flaccid, one thing to note, is that there really aren't going to be newer versions of aki and aris | 04:52 |
| smoser | other than on lucid | 04:52 |
| smoser | on maverick, we're using the pv-grub kernels from amazon and loading the kernel from inside the images. | 04:52 |
| flaccid | smoser: really happy to hear that! | 09:41 |
| progre55 | hi guys. I have rebundled an aws ami into an uec image, and uploaded it into a bucket and registered. | 14:28 |
| progre55 | now I can see it in my "images", but how do I start it? | 14:29 |
| progre55 | it says "state - available, actions - disable" | 14:29 |
| === mathiaz_ is now known as mathiaz | ||
| progre55 | ah nvm =) | 14:33 |
| progre55 | when trying to run an instance, I get "ImageVerify: Unable to determine required kernel image." Any suggestions, please? | 14:46 |
| marrusl | Hello. I'm running the standard image store Lucid 64 bit EMI. On "cluster1" it works correctly and I can ssh to it, on a second cluster, "cluster2" (connected via openvpn to the CLC) I get errors in the console. For example: | 14:51 |
| marrusl | Caught exception reading instance data: http://169.254.169.254/2009-04-04/user-data | 14:51 |
| marrusl | And it never boots far enough to ssh to it. | 14:51 |
| smoser | progre55, most likely you want to launch it with one of the ubuntu kernels. | 14:52 |
| smoser | you *can* register it with a default kernel, then you dont have to specify one at launch time. | 14:52 |
| progre55 | smoser: how do you register it? | 14:53 |
| progre55 | while registering the image? | 14:53 |
| smoser | marrusl, you're seeing a metadata service bug. the metadata service is the thing that provides that web api/interface to the client. it is a bug or misconfiguration in eucalyptus if you can't get to it. | 14:54 |
| smoser | if you download and register a newer image (http://uec-images.ubuntu.com/releases) you will get better information on what is going on in the console. | 14:54 |
| smoser | how did you register it ? | 14:54 |
| smoser | progre55, ? | 14:55 |
| smoser | you must have bundled and uploaded ? | 14:55 |
| progre55 | yep | 14:55 |
| smoser | when you bundle you can specify --kernel/--ramdisk | 14:55 |
| progre55 | and registered it on the console | 14:55 |
| smoser | when you register, you can also add those flags. | 14:55 |
| smoser | i have little to no experience with the console | 14:56 |
| smoser | but i would expect it would let you specify them somehow. | 14:56 |
| progre55 | oh I see.. and can I use one of those standard-ubuntu-image kernel/ramdisk's? | 14:56 |
| smoser | yeah, you'll want to use those. | 14:56 |
| marrusl | thanks smoser. I will bring down a newer image. I swear I had this one at least booting once before though! :) | 14:56 |
| smoser | you can look at a ubuntu registered image, and use that, or get that information from http://uec-images.ubuntu.com/query | 14:56 |
| smoser | (traverse and find the build that you based it on and use the aki/ari that are listed there) | 14:57 |
| smoser | note that lucid images do not use an ari | 14:57 |
| smoser | marrusl, it is possible. | 14:57 |
| smoser | unfortunately, the metadata service is not 100% | 14:57 |
| smoser | and i have seen people have issues with multiple clusters (i think). | 14:57 |
| smoser | i have little to no experience with multiple clusters, personally. | 14:58 |
| progre55 | smoser: according to the description, neither euca-register nor euca-upload-bundle have --kernel or --ramdisk options.. | 14:58 |
| marrusl | smoser, yup. it doesn't seem many people have done it. out of those I talk to at least. fun. | 14:59 |
| === dendrobates is now known as dendro-afk | ||
| smoser | progre55, euca-bundle-image does. | 14:59 |
| smoser | sorry.. its bundle time or register time, not upload time. | 15:00 |
| progre55 | oh yeah =) | 15:00 |
| progre55 | thanks | 15:00 |
| smoser | the ec2 tools do have it for register. the euca2ools lag that a bit. | 15:00 |
| progre55 | yeah, this is my first day with uec =) | 15:01 |
| === dendro-afk is now known as dendrobates | ||
| smoser | progre55, oh. you're using uec ? | 15:16 |
| smoser | i thought ec2. | 15:16 |
| smoser | if you're using uec, i suggest using "uec-publish-image" | 15:16 |
| smoser | which wraps the upload and bundle and such. | 15:16 |
| smoser | its usage should be self explanatory | 15:16 |
| smoser | marrusl, you can use uec-publish-tarball with the thing you download from uec-images.ubuntu.com | 15:16 |
| smoser | they're just convenience wrappers | 15:17 |
| marrusl | Nice, smoser. I will start. Although I'm glad I learned it the hard way the first few times. :-) | 15:17 |
| smoser | the hard way is quite painful | 15:18 |
| smoser | :) | 15:18 |
| progre55 | well, I'm using euca2ools | 15:19 |
| marrusl | indeed. eventually it makes sense though. | 15:19 |
| marrusl | still annoying to type. | 15:19 |
| smoser | progre55, yeah, the euca2ools are fine. you can use the uec-publish-image, it just wraps them. | 15:22 |
| smoser | uec-publish-image x86_64 my-rebundled-image.img my-bucket --kernel aki-abcdefg --ramdisk ari-hijklmn | 15:23 |
| smoser | and it does what it needs to do | 15:23 |
| progre55 | smoser: thanks | 15:24 |
| marrusl | smoser, I won't know if that helped for a while. it's pretty slow bringing down an image over VPN the first time. :) | 15:25 |
| marrusl | walrus is not local. | 15:25 |
| progre55 | hmm.. the instance is starting up as "terminated" | 15:33 |
| smoser | progre55, hm... | 15:36 |
| smoser | you have to make sure that the instance size will hold the root filesystem | 15:36 |
| progre55 | smoser: yeah, the image is 10Gb, and the first time I ran it as m1.small, but it didnt start saying the disk is too small | 15:37 |
| smoser | yeah. what is your host ? | 15:37 |
| progre55 | then it didnt say anything when I started it as m1.large | 15:37 |
| smoser | 10.04 ? | 15:37 |
| progre55 | yeah | 15:37 |
| progre55 | it's a rebundled AMI | 15:37 |
| smoser | i think that euca 2.0 gives a better error message | 15:37 |
| progre55 | from amazon | 15:37 |
| smoser | but its a pita | 15:37 |
| smoser | you can resize the filesystem | 15:37 |
| smoser | uec-resize-image will help | 15:37 |
| smoser | if you have space to spare | 15:38 |
| progre55 | aha, thanks, let me try that | 15:38 |
| progre55 | do I resize after I bundle it? | 15:38 |
| smoser | before bundle | 15:38 |
| smoser | it operates on a partition image | 15:38 |
| progre55 | oh I see, let me try that | 15:39 |
| marrusl | smoser, well it is a new error now, though sort of the same... | 15:43 |
| marrusl | waiting for metadata service at http://169.254.169.254/2009-04-04/meta-data/instance-id | 15:43 |
| marrusl | 14:41:31 [ 1/100]: http error [500] | 15:43 |
| marrusl | (and then the 2nd line keeps repeating and counting up) | 15:43 |
| marrusl | looks a lot like this thread actually: http://groups.google.com/group/ec2ubuntu/browse_thread/thread/17101fb30179c8f6?pli=1 | 15:54 |
| progre55 | how long does it take for an image to run? cause I ran it, and it's been "pending" for a long time now.. | 15:54 |
| smoser | marrusl, your'e on lucid (10.04) ? | 16:00 |
| smoser | https://bugs.launchpad.net/ubuntu/+source/eucalyptus/+bug/637659 is 10.10 with 500 | 16:00 |
| uvirtbot | Launchpad bug 637659 in eucalyptus "metadata service returns 500 error" [High,Confirmed] | 16:00 |
| marrusl | progre55, if it's the first run of a particular EMI or if has to recache it later. it can take a little while. | 16:00 |
| smoser | marrusl, you should open a bug, please. use 'ubuntu-bug eucalyptus-common' | 16:00 |
| smoser | progre55, it does take a long time on the first time to a node controller | 16:01 |
| smoser | you can ssh to the nc, and you should see something happening | 16:01 |
| smoser | sudo euca_conf --list-nodes | 16:01 |
| smoser | might tell you which node its running on | 16:01 |
| smoser | (that has to be run on the CC i think) | 16:01 |
| marrusl | smoser, thanks. will do. | 16:02 |
| progre55 | aha, thanks | 16:04 |
| progre55 | ah, I guess I've ran out of space on the node =) | 16:07 |
| progre55 | how can I stop the instance run process? | 16:07 |
| progre55 | the thing is, for now I have only one node, set up on a 4G flash drive, as I dont need much hd space, but processor power only ) | 16:08 |
| progre55 | and the flash drive only had 3.24G available after the node installation, and my image was 4G | 16:09 |
| progre55 | when I ssh'd into it, it even said "-bash: cannot create temp file for here-document: No space left on device" =) | 16:10 |
| progre55 | so I guess I need to stop the instance run now.. | 16:11 |
| smoser | progre55, i dont know if you can easily do that if you cant get to the node | 16:21 |
| smoser | if you could, just start killing 'dd' pids or whatever looks like its doing someting | 16:21 |
| smoser | or sudo restart eucalyptus-nc CLEAN=1 | 16:21 |
| progre55 | well, I have terminated it | 16:21 |
| smoser | would probably do it. | 16:21 |
| smoser | ok. | 16:21 |
| progre55 | using euca-terminate-instances | 16:22 |
| smoser | if you want to test to see if things are generally running, i have smaller images available that you can start with | 16:22 |
| smoser | http://smoser.brickies.net/ubuntu/ttylinux-uec/ | 16:22 |
| progre55 | cool, thanks | 16:22 |
| smoser | you can use those with 'uec-publish-tarball' to bundle/upload/register. | 16:24 |
| progre55 | how can I browse the walrus contents and my buckets? | 16:34 |
| smoser | hm... | 16:37 |
| smoser | progre55, you can patch s3cmd | 16:37 |
| smoser | there are other s3 tools that can be pointed at euca i think | 16:37 |
| smoser | (i'm interested in if you find something easy) | 16:38 |
| smoser | i just dont know much there. | 16:38 |
| progre55 | but how do YOU do it? ) | 16:40 |
| progre55 | there is JetS3t, let me have a look at it | 16:42 |
| smoser | i dont put anything in buckets that isn't an ami | 16:45 |
| smoser | so i use euca-describe-images | 16:45 |
| smoser | :) | 16:45 |
| progre55 | =) | 16:48 |
| progre55 | I see | 16:48 |
| progre55 | so, again, my instance was terminated when I just ran it | 16:49 |
| progre55 | is there any ways I can see the errors or causes brought to its termination? | 16:49 |
| smoser | it just went straight to terminated ? | 16:55 |
| smoser | there are loads of logs, the problem is there are too many usually. | 16:55 |
| smoser | i might look on the node controller | 16:55 |
| smoser | and look in libvirt to see if there is any informatoin there. | 16:56 |
| smoser | /var/log/libvirt | 16:56 |
| smoser | and /var/log/eucalyptus | 16:56 |
| progre55 | let me check it out.. | 16:58 |
| smoser | the libvirt/<instance-id>.log might have something if you got that far. | 16:59 |
| smoser | you may be running out of space | 16:59 |
| smoser | the NC needs space for at least 2 copies of the partition image | 17:00 |
| progre55 | well, libvirt doesnt have anything | 17:02 |
| progre55 | but /var/log/eucalyptus/axis2c.log has a buch of "[Tue Sep 14 18:01:36 2010] [error] error.c(94) OXS ERROR [x509.c:287 in openssl_x509_get_subject_key_identifier] oxs defualt error , The extenension index of NID_subject_key_identifier is not valid" | 17:02 |
| progre55 | oh and in /var/log/eucalyptus/nc.log I have this "insufficient disk capacity remaining (2042MB) in VM Type of instance i-3A780611 for component disk" | 17:04 |
| progre55 | so it IS the disc space issue =) | 17:04 |
| === niemeyer__ is now known as niemeyer | ||
| === dendrobates is now known as dendro-afk | ||
| === dendro-afk is now known as dendrobates | ||
| rberger_ | Does cookbook_file use the checksum attribute like remote_file or does it do that automatically? I'm cleaning up my remote_file calls and wondering if I should delete the checksum attribute? | 23:50 |
| rberger_ | Opps sorry wrong irc | 23:50 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!