With computers, bleep happens. (Part 2) A.K.A When the RAID fails.
Back to the computer Saga.
Once I got my laptop working, it was time to get things up and running on my new Tower, Kusanagi_Makoto. This singular wonder, which I have also come to call Minus Morgol in the back of my head, had a bit of a RAID problem. To be more accurate, it had a RAID 5 array that ceased to function.
A little bit of background for those not technically adept.
A RAID array is a simulated drive which combines multiple other drives (or partitions) in order to enhance either reliability, performance, or both.
RAID 0 is a striping array, where you write too and read from different drives in blocks of data called ’stripes’, of a size specified when you build the array. The delay in almost every computer operation comes down to waiting on the hard drive, so by splitting the data between two drives, you can basically cut the wait time for a particular large chunk of data in half, thus speeding up your computer. Of course, if either drive fails, you loose your data.
RAID 1 is a mirroring array, where you write and read data to two different drives, so that if one fails you can replace that drive, and still have a copy of all of your data.
RAID 0+1 is where you stripe two drives, and create an exact copy of those two striped drives. Only one drive can fail.
RAID 1+0 is where you mirror two drives, but stripe those mirrors between two different drives each. Only one drive can fail.
To be honest, I’m not entirely sure of the difference on these two setups myself any more.
RAID 4 is similar to a RAID 0 set up. The main difference is that a third, smaller, drive is added. This drive contains “parity” information which can be used to rebuild the data on one drive from the data on the other. The parity data can also be rebuilt from the data on the two other drives, so that if any drive fails, it can be replaced with an identical one, the array rebuilt, and then computer use can continue.
RAID 5 separates works like RAID 4, except that it has 3 identical drives, and it spreads the parity data across each of the drives. Like RAID 4, you have to rebuild the array with a new drive, which can take quite some time. And like any RAID, while that drive is down, you have a much higher chance of a new drive failing due to the array trying to write to a non-existent drive.
This is the kind of RAID I have.
RAID 6 separates the data out across 4 drives, and you can have up to 2 failures. This is in case another drive fails between when you get your first drive.
Now back to the story of my particular RAID
Until recently I had been operating in rather cluttered, cramped conditions. My new tower was actually restricting my leg space, and I tended to bump into it when mounting or dismounting my chair. It seems that on two separate occasions when I bumped my tower, I caused one of the data cables on my SATA (Serial Advanced Technology Attachment) Hard drives. The first time it “degraded” (the state in which I need to replace a drive) the second time it became an “error” drive, and my computer ceased to function.
I thought this might be the case, so I disconnected and reconnected each of the drive cables in turn. When I turned my computer back on, joy of joys, all the drives were there! However I had a further problem. Only 2 of the drives were associated as a part of a degraded RAID 5 array (that I could continue to use) while the 3rd was off in it’s own little Error RAID 5, likely due to the fact that it had disconnected first, and thus wasn’t in sync with the other drives. So I had to delete that 1 disk ERROR RAID, and repair my 2 disk DEGRADED array with it, causing me a case of nerves as the drive rebuilt it’s self.
So now I’m back with a RAID 5 array that works, and happily typing away on my computer.
However immediately there after, I began the next part of the repair process, preventing it from happening again. I began a full scale assault on the clutter that was my room mate, and in the course of half of a week, had my room presentable by my old standards. I continued on, and further on, and am continuing to clean in order to make sure I have a work, living, and sleeping space I can live with. For a few nights I had trouble adjusting, as my room seemed as clean and foreign to me as a nice hotel room.
As a part of this, I moved my old tower, Da-Vinci (an old, and somewhat decrepit tower I have come to think of as Minus Tirith in the back of my head at times) to the basement to act as a sometimes file server. I fully intend to experiment with installing Linux on this old machine over the next few months, perhaps turning it into a full fledged server. I won’t be able to run any long term net server applications over it, but perhaps I can get away with setting it up for FTP. I’ll have to check the terms of service for my Cable provider.
And now, back to cleaning.
March 28th, 2007 at 2:47 pm
I don’t understand what 1+0 and 0+1 do. I don’t mean what the difference between those two are, I mean I don’t get them at all
March 28th, 2007 at 2:59 pm
Ah. Well I’m not sure I understand the difference in the end result, but the main difference is in the order in which the drives are grouped.
In 0+1 you create 2 striped drive from two other drives each. In other words you take 4 drives, and make 2 larger drives from those 4, using striping. You then take those two larger drives, and have them mirror each-other. So that when you write to the new 4 disk mega-drive, you write the same data to each cluster of 2 striped drives. These striped drives then alternate which disk they write the data too.
Let’s say that I have 4 disk drives. Let’s call them A, B, C, and D.
I put A&B together as a striped drive, and call it x and C&D together as a striped drive and call it y.
Now I make sure that y and y mirror each other.
1+0 is the opposite order.
You take A&B and mirror them, calling that mirror drive x. You take C&D and Mirror them as well, calling that mirror drive y.
Then you stripe x and y.
Does that help at all?
March 28th, 2007 at 3:06 pm
OoooOOOooo.