This post originated from an RSS feed registered with Ruby Buzz
by Samuel Kvarnbrink.
Original Post: Why I'll never trust a software RAID set again
Feed Title: monkey in the middle
Feed URL: http://www.contegix.com/rss/feed.xml
Feed Description: Ruby, Rails, Macs, PostgreSQL and life in general.
I'm writing this as I'm standing in the server room at work, waiting for some stuff to compile. I've just managed to kill a software RAID 1 set, which (oh irony of ironies) I had set up to be safer than plain disks. I don't know whether to laugh or cry, but since I only stored various junk on it I'm leaning towards the former.
The weird thing is that I still don't know what went wrong! The RAID set, which consisted of two identical disks, was ridiculously slow even from the beginning, but had started to cause weird freezes and OS X was barfing RAID-related error messages into the system log. That's not especially unnatural if one of the disks is faulty, and Disk Utility reported that the set was in degraded mode (i.e. only running from one disk). But when I tried to get more detailed info from diskutil, in order to find out which drive that was failing, everything looked just fine! Both disks were OK, and every attempt I did at forcibly repairing the set failed.
I eventually ended up wiping the whole thing, swearing never to trust a software based RAID again, and the disks are now running - flawlessly - as separate units. The only thing I lost was an SVN repo I was setting up, along with a few archived user accounts (that are backed up on tape as well), so it really didn't matter. A bit frustrating though, and definately not what I needed on the first day after my holidays :)