August 20, 2017

News


Sunday, September 17, 2006
Time for another technical post.

Earlier this year, I built 2 servers for the house: one as a general file and web server, and the other as a private PBX using Asterisk. I talked about the PBX a bit in a previous post.

On the file server, I decided to put in a RAID 5 system that would give me lots of storage for movies, mp3s, and personal data. Using a RAID 5 system allows me to have lots of space, but more importantly, allows me some redundancy so that I can lose one hard drive due to a failure and not lose any data or uptime.

To build the raid, I wanted to use Seagate hard drives. I have been impressed with them in the past and they now offer an awesome 5 year warranty on their hard-drives which is great. I bought all of the parts off of Newegg (my favorite site to order this kind of stuff - great service). The hard drives I bought were "Seagate Barracuda 7200.9 ST3250824AS 250GB 7200 RPM SATA 3.0Gb/s Hard Drive - OEM". This is a SATA II drive (higher speed) so I needed a SATA II controller that would work well in linux. I decided on the Promise SATA TX4 controller. It does not have a hardware raid: it is a controller only, which is why it is affordable. No worries. The linux software raid works just fine. I also bought a gigabit network card.

Everything has been working great. I used Centos and got all of the hardware set up. I had a small glitch with the Promise SATA driver, but I easily found the bug in the install script and fixed it manually. I set up a software RAID for RAID 5 using the 4 250 GB drives which gave me 750 GB of usable space. I set up mdmonitor to email me on any events I should know about (such as a failed drive) and everything has been great.

A few weeks ago, we had some sort of weird power event. My server locked up for the first time ever. I received an email telling me the array was degraded and it automatically rebuilt across the other drives. Awesome - it did what it was supposed to. I pulled out the failed drive, filed an RMA on seagate's website, and a new drive arrived in a few days. I installed it, and ran "mdadm /dev/md0 --add /dev/sdb". That was it. It detected the new drive, rebuilt the array on all 4 drives (with no downtime), and in about an hour, I was back up to being single fault tolerant. Awesome.

Kudos to Seagate for a really good RMA process. Also to the mdadm developers - the software raid performs very well and is very easy to maintain.

If anyone is setting up a similar software raid and has some questions, let me know and I'll see if I can help.

Labels: , , ,

0 Comments:

Post a Comment

<< News

Archives

Current Posts


Atom news feed RSS news feed

This page is powered by Blogger. Isn't yours?


  Search

Google
Web graffagnino.net
  Friendswood Weather

  Click for Friendswood, Texas Forecast