It wasn't me. You can't prove anything.


2010-09-24

RAID sucks

We finally replaced and re-RAIDed the drive that was reporting massive errors today. I had to warn every one. I found yesterday the distribution list I've been using for engineering does not work. I copied and pasted the email addresses in to my email this time. At least that worked.

I attempted to match the drive partitions by hand. Of course, I screwed it up. Matt, my RAID surveyor of the day, showed me a sftab command that did the trick. Need to be careful with that one, but it works like a champ.

The first round of messing with the RAID stuff didn't work. The second, fifth, ninth either. Then, something like the twelfth worked. We swapped drives in and out of the system about a dozen times. We played with grub. I hate playing with grub. It scares me. Particularly, it scares me on a system that is not just a throw away system.

It appears to be working now. We did have to go back and make sure grub was booting off the correct manner. The drive usage report doesn't match up to my expectations. The massive writes make sense, but the two drives do not have identical writes and reads after you adjust for the sync of the two drives.

Next time. ... I have no idea. Maybe making the boot partition a real non-RAID partition. I'm lost. What have I learned? Don't trust RAID mirrors to be restoreable. RAID is a scam.

No comments: