Import an HVM AMI

tgerla edited this page Nov 19, 2012 · 2 revisions
Clone this wiki locally

You can directly import an HVM AMI from Amazon into Eucalyptus with a few modifications.

Note: These steps are very similar to the section section of and should perhaps be merged.


  • An HVM AMI running in Amazon
  • A scratch instance running in Amazon
  • A scratch instance running in Eucalyptus


In EC2:

  1. Run a cc2.8xlarge instance of the HVM AMI. This example uses ami-6e649707, a SLES11 SP1 image.

    $ ec2-run-instances -k keypair ami-6e649707 -z us-east-1c -t cc2.8xlarge2
  2. Wait for it to go to "running" state, and then immediately stop the instance:

    $ ec2-stop-instances i-87f05bf8
  3. Take a snapshot of the instance's backing volume:

    $ ec2-create-snapshot vol-4f4fce33
  4. Create a volume from the snapshot:

    $ ec2-create-volume --snapshot snap-c8d4c681 -z us-east-1c
  5. Attach the volume to your "scratch" instance:

    $ ec2-attach-volume -i i-39f85346 vol-1f61e063 -d /dev/sdf

    Log into the instance and run fdisk -l to determine correct attachment device. In this example it was /dev/xvdf, not /dev/sdf--the device name depends on the Linux kernel.

  6. Download the volume contents to local disk:

    $ ssh -i keypair.priv root@aws-scratch-instance "gzip - < /dev/xvdf" | zcat > ami-hvm-10gb-suse.img

In Eucalyptus:

  1. Create a 10GB volume to hold the AMI contents:

    # euca-create-volume -s 10 -z cluster01
  2. Attach it to your scratch instance:

    # euca-attach-volume -i i-scratch vol-xxxxxx -d /dev/vdc

    Log into the scratch instance and run fdisk -l to determine the correct device to which the volume was attached to.

  3. Copy the volume contents to the instance:

    # gzip - < ami-hvm-10gb-suse.img | ssh -i tgerla-euca.priv root@scratch-instance "zcat > /dev/vdc"
  4. Log into the scratch instance and run these commands (assuming volume device is /dev/vdc):

    # yum install parted.x86_64
    # partprobe /dev/vdc
    # mount /dev/vdc1 /mnt
    # mount -o bind /dev /mnt/dev
    # mount -o bind /proc /mnt/proc
    # chroot /mnt
    # passwd <-- (set a password for 'root' here)
    # mkinitrd
    # exit
    # e2label /dev/vdc1 rootdisk

    NOTE: These commands will be different for other Linux distributions. In particular, regeneration of the initial ramdisk (mkinitrd) will be different.

  5. Edit /mnt/boot/grub/menu.lst and ensure that the selected kernel line says: "root=LABEL=rootdisk" not "root=/dev/sdc1" or similar. This is important so that grub can find the right root device.

  6. Unmount the partition:

    # umount /mnt/dev
    # umount /mnt/proc
    # umount /mnt
  7. Detach the volume:

    # euca-detach-volume vol-xxxxxx
  8. Snapshot and register the new EMI:

    # euca-create-snapshot vol-XXXXXX
    # euca-register -n "Imported HVM image" --root-device-name /dev/vda -b /dev/vda=snap-XXXXXX