Wednesday, April 02, 2008

Dell PowerEdge R200 - Simulating hardware failure

Warning, I'm not advocating that you *you* do this!
Determined to have a free (as in beer) operating system on the new servers I tried to put Ubuntu on them, but I couldn't get dell's open manage to work with the storage controller :-( so I've gone to CentOS 5.1 instead. Its more of a pain to get going with but probably a better bet in the long run.

Anyhow OpenManage installed fine, but being an untrusting guy I couldn't ship them off to the colo without knowing that I would spot a drive failure, so we simulated a failure by pulling out the cable. What do you know? It worked as advertised, and after pluging it back in again and rebooting (the R200's we've got don't have hot-plug drives) it rebuilt the drive quite happily.


Time to do the other R200 and the PE2950III, the 2950 has 8 hot plug 2.5 inch bays, we've only got 3 drives in there right now in Raid-5. How the hell can raid 5 possibly work?

pe2950III Energy Smart

Unless the third drive is like schrodinger's cat, in which case it would be a complete copy of whichever drive happened to be missing, my power of imagination is defeated by the idea that any 1/3rd of my data can be created from the other 1/3 and some meta-data bout the missing stuff. If thats the case why do I need the missing drive at all? It all makes no sense to me, but I'm happy to trust it!


Brian Moon said...

RAID 5 works by writing a checksum bit to the "extra" drive. What is written depends on parity. For even parity, a bit is written to the 3rd drive that makes the sum of the bits across the 3 drives even. If odd parity is in use, it is the opposite. So, if one drive drops out and parity is even, the RAID controller knows what was on the missing drive due to the checksum.

blog comments powered by Disqus

I know nothing, I'm not a fortune teller, and you'd be insane to think that I am. This disclaimer was cribbed from an email footer I once received. It is so ridiculous I had to have it for myself.

Statements in this blog that are not purely historical are forward-looking statements including, without limitation, statements regarding my expectations, objectives, anticipations, plans, hopes, beliefs, intentions or strategies regarding the future. Factors that could cause actual results to differ materially from the forward looking statements include risks and uncertainties such as any unforeseen event or any unforeseen system failures, and other risks. It is important to note that actual outcomes could differ materially from those in such forward-looking statements.

Danny Angus Copyright © 2006-2013 (OMG that's seven years of this nonsense)