On Mon, 11 Jan 2016 12:29:46 +1300, Ian Stewart wrote:
xI wondered if this meta-data made the disk unreadable. E.g. If you had one volume of a RAID5 set then you cant recover your data from it, but a volume of RAID 1 should be OK to recover data.
Yes, it has worked fine so far with RAID-1. That’s why I’ve only used RAID-1. ;)
I also wondered if device names would cause conflicts.
No, because the RAID devices are named /dev/md0, /dev/md1 etc. The original disk names /dev/sda etc remain unaffected.
For scenarios 3 and 4, I was thinking along the lines that your desktop computer at work has RAID1 based USB drives for all your ~/home/ data. You just pull out the USB stick when you go home at the end of the day, and plug it in again in the morning. If the office burns down at night and you loose your computer, then your data is all on the USB stick and available to be read by another computer. This RAID1 approach may not be the best way to get yourself a backup, but would be one of the quickest / easiest to manage.
Good point. But you can’t be sure that syncing has finished when you want to unplug. I have a custom script that uses rsync to back up between volumes. I can easily run that several dozen times a day. It’s fast.
Maybe at a wlug meeting we could get Linux RAID1 up and running on a laptop with some USB sticks and subject it to some testing and see if it handles the abuse and how auto-magic it is, etc.
Certainly can do. Keep the volumes to a modest size, so it won’t take too long. :)