On Thu, Oct 22, 2009 at 21:04, adfas asd <chimera_god@xxxxxxxxx> wrote: > How could this possibly have happened? The whole idea of RAID is so something like this won't happen. Actually no. No it is not at all. The point of raid it is to provide larger and faster block devices than otherwise physically exist, and in some cases allow you to change out failed drives without making the array unavailable for use. I believe you are confusing raid with backup. You won't do that again. If your data isn't completely disposable, I'd recommending breaking that raid10 array into two raid0's, on two separate machines, and rsyncing between them. Each individual raid0 array should actually be even faster this way, and you'll definitely have more read IO available using two machines. > I was using JFS. > > I've lost confidence now in mdadm. I have too much data to back up practically, and am now at a loss. mdadm works fine. But it is not to be used as bubble gum, or a web browser, or a text editor, or backup. Reply with an mdadm -E /dev/sd[abcd]. Once the array assembles, your problem is with JFS. The machine was malfunctioning when you pulled power. It is entirely possible that its malfunction caused it to corrupt the filesystem beyond recovery. mdadm's job (and the job of any raid solution software or hardware) is to immediately and irrevocably accept the io from the filesystem driver, and pass it on to the disks. The easiest and cleanest solution is to dd the first and last 100mb or so of the disks with /dev/zero. Remake the array. Use something sensible like ext4 this time, and restore your backups on to it. If you don't have any backups to restore, you will next time. -- To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
