Blog (workarounds: avoid ', use '&apos' instead)

  Company Web Site

Rackspace competes with Amazon, Microsoft, Google, and others to provide various cloud services.  Rackspace is especialy known for providing virtual Linux servers using their variant of the OpenStack framework.

Server Image Backups

  Elsewhere [ edit ]

The semantics around what Rackspace claims are server images are a little murky.  Documentation and forum posts suggest server images are a way to save templates of server for replicating additional new server creation.  However at its heart, server image backups are byte for byte server disk state snapshots, perfectly suited for backing up entire server state for later recovery.

Tips include:

Offsite file backups
updated 2015-12, rickatech


Rackspace allows making regular server image backups that they keep in the cloud.  However any resilient backup strategy includes distributing achives amongst different locations to insure they are available during outages, service interruptions, and after hosting service is discontinued for some reason.  To this end Rackspace provides external file transfer access via various mechanisms including an OpenStack Object Storage interface well suited for for transporting massive files, something basic web browser downloading can't easily perform.

Uploading files to Rackspace is essentially free while normal bandwidth charges apply to download files.  Often server image files will present as a series of compressed tar.gz files that need to be recombined to recover the native VHD storage file.  The VHD file can be converted to a runnable instance or mountable dataset with other tools.

It is usually sufficient to secure a reliable way to access archive files locally, and below are some instructions showing how to do this for Linux boxen.  With Rackspace, your server backup images will remain in their cloud even after a cloud server instance is deleted/retired.  If there is a reason to dynamically recover a live instance of your cloud server, simply create a new server, but specify a backup image for it to spawn from.  Although it is possible to resurrect a live server originally hosted at Rackspace from backup files locally with Xen, Open Stack, VMware, or some other virtualization setup, the low level disk partitioning and network interfaces will require manual tweaking that is quite finicky and tiresome to adapt, typically making such efforts non-trivial, unreliable and hence impractical.  The best strategy then is to make sure all critical files, data, and configuration settings are backup up regularly so that simple access to to backup files is sufficient for your most egregious recovery needs.  Also, don't underestimate the power of distributed configuration management tools like git to completely obviate need to backup important files that can be easily managed thusly.

Update 2015-12

Rackspace has changed the backup file sets slightly.  It's still important to use a OpenStack file transfer utility like Cyberduck, but now you need to create a 'container' export of the server files first from their Cloud Image site  There is now a single zero length 'magic file' that is part of a file set with almost identical filenames that all match the Export ID.  For some reason copying the empty file pulls together the file fragments during download, producing a single monolithic massive file with Export ID as it's filename.

  - 1st pass
    pop up will provide Export ID
  - 2nd pass
    use status/check
    provide the Export ID ... it will say processing,
    when it shows success then try see files available for download  


Record hard disk and network settings - example.

[ use Cyberduck or other agent to transfer large files ]

$ ls -lh

  5.0G Oct 18  2014 c6x0ipxx_20141018.tar.gz.0
  2.3G Oct 18  2014 c6x0ipxx_20141018.tar.gz.1
  217B Oct 18  2014 c6x0ipxx_20141018.yml

$ cat c6x0ipxx_20141018.tar.gz.* > c6x0ipxx_20141018.tar.gz

$ ls -lh

  7.3G May 12 10:22 c6x0ipxx_20141018.tar.gz
  5.0G Oct 18  2014 c6x0ipxx_20141018.tar.gz.0
  2.3G Oct 18  2014 c6x0ipxx_20141018.tar.gz.1
  217B Oct 18  2014 c6x0ipxx_20141018.yml

$ tar xvf c6x0ipxx_20141018.tar.gz 

  x ./
  x ./image/
  x ./image/manifest.ovf
  x ./image/image.vhd
  x ./image/snap.vhd
  x ./manifest.ovf

$ ls -lh image

  9.4G Oct 18  2014 image.vhd
    8B Oct 18  2014 manifest.ovf
   24K Oct 18  2014 snap.vhd 

The image.vhd file can be converted to an image.img file using tools like qemu-img.  See Elsewhere for tips on how to mount converted image files.

edit | log