Hey there. Long time reader, first time poster.

I have an OpenSuse 11.1 server running on an Nvidia H/w RAID setup. Every so often (Seemingly random intervals) a process called kraid45d gets called up and the system will hang for ~15-20 seconds. After this, load drops back down to normal.

As I understand it, kraid45d is a daemon that bulk-updates data on the RAID system. Looking at dmraid -r, I get the following output:
/dev/sdd: nvidia, "nvidia_deefaedf", raid5_ls, ok, 1953525166 sectors, data@ 0
/dev/sdc: nvidia, "nvidia_deefaedf", raid5_ls, ok, 1953525166 sectors, data@ 0
/dev/sdb: nvidia, "nvidia_deefaedf", raid5_ls, ok, 1953525166 sectors, data@ 0

Which all seems normal. I have a detailed view of my system resources, and while kraid45d eats ~90% of the CPU time, the user load seems to stay the same. Also, kraid45d uses both cores when it does this action, so there's no CPU time left for users, and all of them experience this slowdown.

I am wondering: Is there a way to get that daemon to run more often so that the "hang time" is much less noticeable? I did something similar to this using the vm.dirty_ratio kernel variable and it improved performance after upgrading the RAM.


System Specs:
AMD Athlon 64 X2 Dual Core 6000+
Nvidia GeForce 8600GT
7.8 GB RAM (4 x ~2GB)

I don't want to have to buy a file server right now, but that might be my only option. Any help would be MUCH appreciated.

-foophy