As the title says, I regularly download my Rackspace images for backup of my VPS, however I'm wondering if it's possible to run them locally on KVM (for example) so I can run tests on my application without affection production.
Almost a year to the day after the original question, I think I have a good procedure set up for downloading Rackspace images and running them locally, under Xen running on Ubuntu Precise.
I'll write down the steps here so there's an easy recipe for you to follow if you want to try it out. The steps work for linux/Ubuntu domU's at least, for others YMMV.
(These instructions are also posted on our blog at http://www.everichon.com/blog/2013/05/29/running-rackspace-images-locally/)
Install Ubuntu precise (12.04 LTS), and follow the steps mentioned in http://blog.heidt.biz/blog/2012/09/17/setting-up-ubuntu-12-dot-04-precise-pangolin-as-xen-dom0/ to set up your server as a Xen Dom0. At least make sure to have a running 12.04 with xen and xe-tools installaed, and a LVM volume group to use for virtual disks for your DomU's.
Go to your servers page on the rackspace cloud (
manage.rackspacecloud.com -> Hosting -> Cloud servers) and select the one you want to run locally. Go to the
Imagestab for that server. Create an on-demand image. (Depending on the time the server was first created the on-demand image will either go directly to cloud files or be stored with the server. If the image is stored with the server, you have to move it to cloud files after creation.)
Check the name of the image in cloud files. Go to
Hosting -> Cloud Filesand open the container
cloudservers. There should be the file there named like
myservername_YYYYMMDD_XXXXX_cloudserverZZZZZ.tar.gz.0, and that's the file you want.
Download the image to your server. The easiest way to do this on Ubuntu 12.04 is to use
swift. To install, just
sudo apt-get install swift. Then run it like
swift -A https://auth.api.rackspacecloud.com/v1.0 -U username -K API_key download cloudservers myservername_YYYYMMDD_XXXXX_cloudserverZZZZZ.tar.gz.0
When the download is done, unpack the downloaded file (
tar zxf myservername_YYYYMMDD_XXXXX_cloudserverZZZZZ.tar.gz.0). When it's done you'll have gotten the following files:
image -> image.vhd -> manifest.ovf -> snap.vhd
image.vhdis the one we're interested in, as it contains your root partition from the cloudserver.
To make it easier to work with the image file, we'll convert it to a raw disk image. This is done using
qemu-img(if it's not available on your system, do
sudo apt-get install qemu-utils). To convert the image.vhd to something that's easier to work with, do
qemu-img convert -p image.vhd disk.img
Check the disk image using fdisk. (NOTE: I've only been working with 20 GB large images from rackspace, and my notes are about them. If you use another size, you'll have to adjust some steps accordingly!)
fdisk -l disk.imgshould give something like the following:
Disk sda1.img: 20.4 GB, 20400758784 bytes 255 heads, 63 sectors/track, 2480 cylinders, total 39845232 sectors Units = sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disk identifier: 0x00000000 Device Boot Start End Blocks Id System disk.img1 * 2048 39845887 19921920 83 Linux
Note the start offset for the first partition, and the sector size.
(Optional) Check that the disk image is ok. For this, we mount it loopback and simply see that it works.
mkdir tmp mount -o loop,offset=1048576 disk.img tmp/ ls -la tmp (The offset above is start for the first partition * sector size)
This gives you an indication that the disk is ok. Then unmount the image.
Create logical and swap volumes for your new DomU.
lvcreate -L20G -n myservername-disk vg0and
lvcreate -L2G -Cy -n myservername-swap vg0. Also initialize the swap disk,
Copy your disk image to the new logical volume.
dd if=disk.img of=/dev/vg0/myservername-disk bs=4096 skip=256. The
skiparguments are there to get the correct offset for your partition. I could go with
bs=512 skip=2048, of course, but making the block size larger makes copying much faster. (And 512*2048 = 4096*256, so it does not matter for the offset.)
Check the filesystem on your logical volume (and fix any errors), resize the fs to fit the lv, and check again.
e2fsck -f /dev/vg0/myservername-disk resize2fs -p /dev/vg0/myservername-disk e2fsck -f /dev/vg0/myservername-disk
Check the virtual disk settings in your virtual machine. This will be needed for your DomU config file. Mount the logical volume.
mount /dev/vg0/myservername-disk tmp. Then check the fstab file, and (if it exists) the grub
more tmp/etc/fstab more tmp/boot/grub/menu.lst
Note the disk names for root and swap!
Update the network settings for your DomU. With the virtual disk still mounted, edit the
interfacesfile to something that suits your local environment.
vi tmp/etc/network/interfaces. I've usually changed mine from something like:
# The loopback network interface auto lo iface lo inet loopback auto eth0 iface eth0 inet static address 184.106.XXX.YYY netmask 255.255.XXX.YYY gateway 184.106.XXX.YYY dns-nameservers 173.203.XXX.YYY 173.203.XXX.YYY auto eth1 iface eth1 inet static address 10.179.XXX.YYY netmask 255.255.XXX.YYY up route add -net 10.176.XXX.YYY0 netmask 255.240.XXX.YYY gw 10.179.XXX.YYY down route del -net 10.176.XXX.YYY netmask 255.240.XXX.YYY gw 10.179.XXX.YYY up route add -net 10.191.XXX.YYY netmask 255.255.XXX.YYY gw 10.179.XXX.YYY down route del -net 10.191.XXX.YYY netmask 255.255.XXX.YYY gw 10.179.XXX.YYY
(for rackspace's environment) to:
# The loopback network interface auto lo iface lo inet loopback # The primary network interface auto eth0 iface eth0 inet dhcp
for just using dhcp and one virtual network interface and my local server. After these changes, we should be done. Unmount the virtual disk,
Create a xen config file for your imported DomU,
vi /etc/xen/myservername.cfg, and put in something like this:
name = "myservername" bootloader = '/usr/lib/xen-default/bin/pygrub' #uncomment the following if you want to boot into single user mode #extra = " single" vcpus = '2' memory = '512' # # Disk device(s). (Use the correct mappings from step 12.) # disk = [ 'phy:/dev/vg0/myservername-disk,xvda1,w', 'phy:/dev/vg0/myservername-swap,xvda2,w', ] # # Networking # vif = [ 'bridge=xenbr0' ] # # Behaviour # on_poweroff = 'destroy' on_reboot = 'restart' on_crash = 'restart'
Again, note that you should use the same device names for the virtual devices as the ones you discovered in step 12.
Now everything should be ready! To verify, create the domU and attach a console by running
xm create -c /etc/xen/myservername.cfg
Here is good link on "Run Rackspace Cloud Server locally on Hyper-V"