megasas: MFI FW status 0x3

[Michael E Brown]

On Thu, Apr 26, 2007 at 11:44:23AM +0200, Robert von Bismarck wrote:
> Please correct me if I'm wrong, but I was under the impression that the
> best IO disk scheduler for hardware RAID was NOOP. So that the
> "intelligence" of the controller would not clash with the "intelligence"
> of the scheduler.

Nope. Technically plausible sounding enough, but nevertheless incorrect.

IO scheduler does things like decide which process can submit IO,
something that a RAID controller has no idea of. It also balances reads
vs writes, so things like the anticipatory scheduler can delay IO from
certain processes if it thinks another process is going to do a
sequential IO (which improves throughput by reducing seeks).
--
Michael