Original question was:
> I was trying to built a huge RAID-5 disk on the 3 channel PCI
> RAID controller (with 6 Seagate 9G drives). I wanted to turn all
> 6 drives into one huge RAID-5 disk, but the standalone config
> utility refused to do it. The automatic option failled (no details,
> just that it failled). Manual config works, you can group all 6 drives
> into one geroup, but when you try to actually define the disks,
> the program does not accept any size above 32768 M (suspicous).
> So: is it a feature, or what? I know I can work around this (like
> creating two RAID-5 disks an combining them with AdvFS or LSM),
> but it's not quite what I want.
> Had anybody tried to go above 32G with officially supported drives
> (like 9xRZ29 or 17xRZ28)? Did it work?
One response from john_at_iastate.edu, warning me that you can only glue
together 8 disks, so even with the 4G RZ29 disks, there is no supported config
that goes above 32G. Probably that's the reason why this feature is
not known to anybody.
As I heard (about half year ago) Dec is experimenting with the 9G disks,
so they will eventually find the same limit and fix it. In the
mean time, I will try to group all 6 disks together, and create 2 RAID-5
volumes (each less that 32G), and use 'addvol' or LSM to combine them
(LSM does not even need a license to do this). Puts in an extra layer,
but since it's still the RAID that does the strping, the performance hit
on the CPU should be minimal. I don't even have to build a kernel, since
by default it supports 2 RAID disks (will be an issue if one goes above
64G -- just 2 more 9G drives).
Gyula
Received on Tue Aug 27 1996 - 14:20:11 NZST