RAID Controllers and Linux: Ugh!

[Bill McGonigle]

On Jun 29, 2007, at 17:06, Tom Buskey wrote:

> What if the RAID card dies?  Can you get a replacement that can  
> read the
> RAID disks you have now?

Well, that's the real trick, now isn't it?

In high-uptime-requirement sites, I require one of the following:
  * cluster of servers
  * software RAID
  * onsite spare for Hardware RAID controller (we're often too far  
north to qualify for onsite support contracts)

I've seen a company with a smoked hardware RAID card that blew at  
10PM on Friday going into a holiday weekend on a mission-critical  
server (that's why they bought a fancy RAID card for it after all).   
We were using forensics techniques to extract the non-backed-up data  
into Saturday morning.  <-- that sucks

Anecdotally I've had both hardware RAID (Adaptec, 3Ware, Compaq  
SmartRAID) and Linux software RAID (2.6.13, IIRC) hose the RAID  
structure on disk.  Linux Software RAID is particularly bad about  
handling unreliable hardware conditions, like when your el-crapo  
Compaq BIOS throws extra APIC-whoziwhatzas and makes a USB disk go  
away. DAMHINT.

Running a 3ware 95?? as a SATAII/NCQ controller, without RAID turned  
on, then Linux Software RAID on top of it, is so-far, bulletproof for  
my servers.  Anybody here running a 3Ware on OpenSolaris?

Software RAID under linux still lacks scripts to handle auto-rebuild  
under hardware replacement, even though the hardware is now capable,  
which is a sweet feature of hardware RAID cards.  We talked about  
this here before and somebody send a perl script that is a good  
start.  I haven't yet had the time to update it, but the procedure is  
straightforward.

-Bill

-----
Bill McGonigle, Owner           Work: 603.448.4440
BFC Computing, LLC              Home: 603.448.1668
bill at bfccomputing.com           Cell: 603.252.2606
http://www.bfccomputing.com/    Page: 603.442.1833
Blog: http://blog.bfccomputing.com/
VCard: http://bfccomputing.com/vcard/bill.vcf