We all know that "stuff happens." This week our co-lo provider pulled a major screw-up and had a complete loss of electricity that included the UPS. In fact, the UPS appears to have surged and took out three of my servers. The entire data center went dark and came back up with a snap, crackle and pop.

Using backups created by Virtuozzo I was able to restore VE's elsewhere while I reformat and provision three servers. It is yet unknown if the drives that were in RAID 0 configuration are actually defective now, or are just throwing off bad sectors that repartitioning will correct. Our 3Ware Escalade 8006-2LP controller just keeps spitting out media errors. All that I can think of is starting fresh and hoping the drives aren't gonners.

http://65.44.220.10/3ware-8006-2lp.gif

BTW. This brings up a point. If one Google's the error code it will tell you that this is common with the Escalade controller, and CentOS v.X. The solution is usally to add noapic to the boot loader line. This didn't work in this case.

But this brought to my mind an old thought of being able to fully load a Virtuozzo server, complete with all current patches and templates, to a thumb drive. I don't know if my Super Micro servers are able boot from a USB, but even if I had to put a CD in and have it load the USB drivers.... this would sure be better than hours of loading software. All of the software downloads are on the network path, so they don't rely on Parellels to be available. Nonetheless, they still have to be installed.

Is there an imaging product that will allow me to do this? Names? URL's?

It sure would be nice to do an entire drive backup for restoral after something like this instead of this -- without a high load on the OS.

Ouch. We definitely feel the pain you and your Co-Lo provider are feeling, but at least you have some backups.

On the RAID 0, there's a good chance the data is corrupted, although there may be a chance one of the recovery companies could help out. Keep in mind that the data recovery companies are proud of their services and it'll cost a bit. I highly recommend staying away from RAID 0 in an environment where data integrity is a requirement.

Theoretically, you can do a bit-level full drive backup using dd(disk druid) in Linux to anything you can mount as a block device(even files). You can also use rsync or even tar+gzip/bzip2 to create archives. Live databases are only a slight bit harder, sometimes requiring that you take them out of production. MySQL provides a few tools to do this manually.

You may want to focus on the specific data which can't be quickly provisioned(e.g. php, html, js, config files, databases) instead of the whole drive as you may have issues with drivers, libraries, and binaries.

Offsite backups are always recommended. Everything from a simple NAS where you script your own backup process to something like the CDP which does just about everything for you.