Richard Creighton wrote:
Erik P. Olsen wrote:
Hi,
I have installed two sata II disks via a RocketRaid 1720 controller.
They run Fedora 7 in a hardware raid1 configuration. I have
experienced certain problems wich seems to be related to kernel
2.6.21-1.3194.fc7, see bugzilla issue 240982.
The driver from hightech only works with this kernel and the socalled
open source driver compiles with errors. This is too much of a pita so
I would prefer to skip the proprietary driver and run software raid1
instead.
First of all I would like to know if linux supports drivers for
RocketRaid 1720? Secondly, is it possible to convert the hardware
raid1 to software raid1 (dmraid) without touching the already running
OS? Or do I have to create a new system from scratch?
I had the same problem with a RR1740x controller with that kernel. I
was able to successfully compile the driver when I upgraded the kernel
to 2.6.22.... being sure to get the correct syms. Since then, I've
been able to run the controller just fine. My problems started with
>2.6.18.... until .22+ It is continuing to work with .17 and .25 in
the alpha 11 of SuSE. It is a pita, to be sure, but I am no happily
running a 2.5TB raid5 with that controller under SuSE 10.3.
As to conversion: I suspect that it is a real PITA to try. The
formats are entirely different and you have to have a small boot
partition in order to load the drivers for the software raid. I tried
and gave up on that idea, at least until RR decides to actually support
their controllers under Linux the way they do for Microsloth.
I suspect the 172x controller will play out the same way as the 174x
controller did as I believe they use the same chipset, the main
difference being how they handle the physical interface, eg, number of
ports, etc.
Well, there are differences but I do believe the basic functions are the same.
The README file in the source directory talks about running "make oldconfig" and
"make dep" under the kernel source directory. I am compiling the driver against
kernel 2.6.23.15-80.fc7 and "make oldconfig" went OK but "make dep" did not
saying: *** Warning: make dep is unnecessary now. Further I should check
modversions.h for a certain line. However, this file is not found in the kernel
source.
Nevertheless I ran make under the driver source and got:
[root@epohost linux]# make KERNELDIR=/usr/src/kernels/2.6.23.15-80.fc7-i686
ARCH=i386
make[1]: Entering directory `/usr/src/kernels/2.6.23.15-80.fc7-i686'
CC [M]
/downloads/RocketRaid/source/rr172x-linux-src-v1.0/product/rr1720/linux/.build/os_linux.o
/downloads/RocketRaid/source/rr172x-linux-src-v1.0/product/rr1720/linux/.build/os_linux.c:
In function ‘os_sd_changed’:
/downloads/RocketRaid/source/rr172x-linux-src-v1.0/product/rr1720/linux/.build/os_linux.c:315:
error: implicit declaration of function ‘scsi_device_cancel’
make[2]: ***
[/downloads/RocketRaid/source/rr172x-linux-src-v1.0/product/rr1720/linux/.build/os_linux.o]
Error 1
make[1]: ***
[_module_/downloads/RocketRaid/source/rr172x-linux-src-v1.0/product/rr1720/linux/.build]
Error 2
make[1]: Leaving directory `/usr/src/kernels/2.6.23.15-80.fc7-i686'
make: *** [rr172x.ko] Error 2
So clearly the compile procedure does not work. Have you had the same bad
experience and been clever enough to overcome the problem? I'll be happy to hear
if you have a way out otherwise I'll have to work with highpoint to get the
problem resolved.
Once I got the RR1740x to compile under Linux with a less buggy kernel,
I have had zero problems with the card other than lack of any real
support, though the driver source did not need modifications once the
kernel issues were resolved.
--
Erik.
_______________________________________________
Ataraid-list mailing list
Ataraid-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/ataraid-list