On Jan 3, 2005, at 12:15 PM, Henry Miller wrote:
This might work, but it isn't best. I can think of the following objections:
First, not all "identical" drives are identical. It isn't uncommon for the factory to give slightly different sector counts for drives of the same model, when something in manufacturing changes. (perhaps a new defective sector algorithm). This would cause the backup to fail on those last sectors of the drive.
If the drives are really identical, odds are they have the same types of defects, and they will fail at about the same time. That is when one disk fails, the other might not be far behind!
You have no protection at all while the copy is in progress. You have overwritten part of the old backup, but not enough to be consistent.
You have made no provision for data loss because of anything other than a failing drive. If your house burns down you can't get your data. (not strictly true, you can recover accidently deleted files so long as you do the undelete before the next time you do the backup)
FreeBSD has a few different RAID options. With the right setup you can achieve disk reliability, and not have to switch cables on reboot. (With your setup you don't have to either, just tell the BIOS to boot from the other drive if the first one cannot boot) This is more work to setup, but more people understand it, so if something happens to you someone else is more likely to figure out what is going on. (this may or may not matter to you) With a good RAID setup, FreeBSD can keep operating even after the disk crashes, while your setup requires manual intervention. If you must have 24x7 operation (web server), then you need RAID. If you don't need 24x7, consider crashed disks an excuse to re-setup your system, in my experience by the time your disk crashes you will have a lot of cruft that you are meaning to remove, so this is a good excuse to re-install.
Thanks for the reply. First off, please reply to the list, so that these emails can be properly archived. This can be accomplished by using reply-all instead of just reply.
You seem to be under the impression that I'm doing this for the sole reason of a disk crash. I'm actually doing it for more than just that reason. For example, if my system gets hacked, most hackers will probably not care about an unmounted hard drive, and screw with the current mounted partitions. Also, these drives wouldn't really be at the same point of this hypothetical drive failure, since one hard drive will only be used roughly once a week, while the other is in a constant state of use. Most of my user-data is destined for a RAID-5 array that's roughly 1.2TB, so that's got it's own backup. This is simply for use in an emergency, so I don't HAVE to rebuild. Quite frankly, I don't have time to sit here and rebuild this system again any time soon. This configuration I'm trying is ideal, with minimal interference. I'm going to be installing removable drive bays so that my roommate is able to simple swap drive positions and reboot the system (it's headless, and he's not very tech savvy in this regard).
Thanks again for the reply.
_______________________________________________________ Eric F Crist "I am so smart, S.M.R.T!" Secure Computing Networks -Homer J Simpson
PGP.sig
Description: This is a digitally signed message part
