The importance of raid - and monitoring

[Red Squirrel]

Just had a drive fail in one of my arrays, got an email alert on my phone and also noticed it on my alarm display which I happened to have open. I don't have a fancy setup where a failed drive has a LED indicator so I document all my drives in a spreadsheet. checked the serial of the failed drive with smartctl, located and pulled it out and popped a spare in. Now testing the spare before I add it to the array to let it rebuild.

No data loss, no need to pull backups, just keep going. Raid FTW.

In this case it's mdadm raid 5, this is an older array with 1TB drives. My other arrays that use bigger drives are all raid 10.

Funny thing is I was just thinking the other day how I have not had any disk failures in a long time and that I may be due... I jinxed it apparently.

 

[master_shake_]

Raid 5 har har

What is this 1990?

Raid 6 is where it's at

 

[Red Squirrel]

I've considered raid 6 actually, but now I tend to do raid 10. I'd have to actually do some testing to see how the performance is though vs raid 5/6. Raid 6 is more resiliant in that you can lose 2 drives. Raid 10 you CAN lose 2, but it depends which ones.

This particular array is just for backups, (and I have cold backups on top of it) so I would not bother doing raid 6 or even 10.

 

[XavierMace]

Yep. I've had a total of two drives go out on my RaidZ2 setup. No issues. just replace the drive.

I also recommend fancy setups with LED's. It makes life so much easier to click an button and have a light tell you what drive to pull.

 

[Red Squirrel]

Yeah one of these days I want to figure out how to have the fail led actually work. I'm not sure if that's doable without using enterprise gear with it's own hardware raid though. Basically the raid needs to be able to actually initiate the LED to turn on.