Hello Everyone,

I'm having troubles with a Dell Poweredge 800 and its PERC 4/SC Raid card running SBS2003.  And 3 SCSI hard drives in a RAID 5.  Things seem to have gone from bad to worse.  Fortunally the client is good about backup, so I believe the worst case scenario would be reinstalling from scratch and relying on backups.The background is this:  Was referred to client when server started beeping.  Immediately I thought it must be a RAID issue.  I booted into the RAID config tool and found 1 of the 3 drives had failed (drive 2.)  Because this is not hot-swapable I shut down the server and removed the drive cage (had to disconnect all the drives to do so) to see brand/size/model number etc for the replacement.  

Reconnected normally and rebooted.  The system got as far as the SBS2003 splash screen just on the first boot attempt.  The RAID beep code switched to a long beep and the system rebooted.  Again I went into the RAID config tool and its now reports drive 0 has also failed, and the RAID has failed as well.  What are the odds??  It is no longer even getting to the Windows splash screen.  

So now I have ordered new drives and am waiting for arrival.  I have not tried to rebuild the RAID.  I wanted to at least replace the first drive to fail before attempting.  Would anyone care to make any suggestions?  It would be really nice not to have to reinstall from scratch.  

The only documentation I found is:  

http://www.cs.uwaterloo.ca/~brecht/servers/docs/PowerEdge-2600/en/Perc4scdc/UG/h...

Also it is a bit confusing which is drive 2.  According to the link, it sounds like the top one is, but in the raid software, it seems to indicate its the bottom one.  Also, the Dell drive cage is labeled top to bottom 0-1-2-3 (there is no 3)

Thank you all,
Jim

I suppose its possible.  I'm not convinced the second is really lost because it didn't fail until I disconnected and reconnected.  I was hoping there is a way to 'refresh' the drive status to see if it will maybe come back to life.

Quote:

Interesting. Food for thought. What is your sample size? (How many drives?)

It's hard for home users like me to come to any insightful conclusions cuz our sample size (usually 1 or 2 or maybe 3 drives) is so small.

jf38081 wrote on Sep 24^th, 2008 at 1:03pm:

The point of a stripped array is speed, not redundancy.  In fact, stripped arrays have *lower* reliability than single hard drives, so it isn't at all uncommon for them to fail.

And, yes, RAID 5 isn't really worth much with only 3 drives.  It would have more reliability than a single hard drive, but you aren't protected against a second drive failing before you replace the first.  RAID 6 helps some, by giving you the ability to have 2 failed drives, but that wouldn't make sense for 3 hard drives (since 2 would be used up just with parity.) 

For home users, there really is no substitute for backups. 

There are other solutions, but not for small installations.  For example, Google invented a distributed file system that keeps multiple copies of the data.  It is sort of like RAID 0+1 but stored over thousands of servers rather than many hard drives in a single computer.  A distributed disk management system keeps track of all the data and ensures it gets automatically copied to a new place if a drive failure brings the number of back-up copies of any data below the required minimum number.

Unfortunately, I don't think Google makes this file system available.  Its too bad, to.  I bet some companies would pay a handsome sum to license it.