I've got a DL360G3 running CentOS 5, patched to a fairly current level. The server has a SmartArray 5i with two 73GB 15k drives in a Raid 1 configuration. (Its a twin 2.8GHz Xeon)

The CentOS install put the cciss driver on, and all works fine. And when the system is under a light load, it performs fine.

Here's the issue: When the system gets a heavy I/O load, the load average jumps to 10.0 or 12.0 and the system becomes almost unresponsive. IOwait is taking about 90% of the CPU cycles.

I don't see any errors in the setup, but this seems really strange. The system shouldn't do this in my opinion. Anyone have any ideas on a fix? I know all the hardware is fine as I've got 4 systems all with the same issue. Where can I look for setup issues on the cciss driver, or the drive subsystem, or??

Thanks in advance!

Update, If I run this under Solaris there is no problem. Is Linux still just not ready for prime time??