No, this isn't about an insecticide spray, but about a Redundant Array of Independent Disks. This web site is hosted on a FreeBSD server that has four 120 gigabyte hard drives in it. The data is stored in a RAID 5 array, which means that if any single drive fails for some reason, the server can still continue operation while I quickly replace the failed disk. All without any data loss. At least that's the general idea behind RAID level 5.

A little less than a year ago, I wrote about the file/web server this web site runs on. Until this week, it has been using that same vinum software to spread the data across the four disks as a RAID 5 array. Unfortunately, as I wrote about here, problems with vinum under FreeBSD version 5 forced me to run FreeBSD 5.1 even though I wanted to instead use the newer FreeBSD 5.2.1 release.

On Monday, FedEx delivered my new 3Ware 7506-LP parallel ATA hardware RAID PCI card and a new 200 GB drive. That evening, after cleaning up the data on my RAID array (a 333 gigabyte array) until I had perhaps 80-90 gigabytes of data, I copied the data across my local network to the new 200 gigabyte drive. That took a while.

Once the data was backed up, I shut down this server and installed the new hardware RAID card, booted, and created a new hardware RAID 5 array. Then I installed FreeBSD 5.2.1-RELEASE. So far, so good. I was really excited that I'd be able to have a reliable system that wouldn't force me to run an older FreeBSD release.

Over the next few days, I copied the backed-up data to the new RAID 5 array and brought my web site and services online. (Not all of them are back up just yet, in particular, my Jabber instant messaging server isn't running yet.)

Imagine my chagrin when this morning my server froze solid, without any output on the console to let me know what was wrong. Eeek! I also learned that a similar problem at work that we've been having with a database server freezing up randomly may be due to problems with the 3ware RAID hardware under FreeBSD 5.2.1-RELEASE (which is what we have been running on the database server). Well, now we have two similar situations, two different FreeBSD 5.2.1 installations on different hardware, the only thing in common being that both use 3ware cards to do RAID 5 (though they even use different cards).

My coworker, Cassidy, will be working on downgrading the database server at work temporarily to FreeBSD 4.10-RELEASE to see if the random freezing stops under the older "STABLE" FreeBSD line. Since my personal web and file server isn't as critical, I'll just hope for the best and hope that the problem is fixed in the forthcoming FreeBSD 5.3 release (which it is intended will become the new "STABLE" FreeBSD release line).