Problems Installing Core 4 and core 2.6.1

[compufix]

I am relatively new to Linux and would appreciate some help.
I have a flie server with MSI K7N2 Delta Motherboard, AMD Sempron 2500 cpu and 3 SATA hard drives.
As the motherboard only has 2 SATA slots I have a Silicon Image Raid card with 2 further ports, giving 4 in total. The card is rcognised as a SIL 3112 /3112A by the Linux Hardware Browser.
I have had the system running software Raid 5 on Fedora core 2.6.5-1.358 with no problem, or so I thought until what I believe to be a power outage took out 2 of the disks.
I have reassembled with 3 new disks and am trying to get sartd to run to monitor the disks and email warning of any error. This works on a test bed I have set up with same core on a much older setup.
I first noticed the problem when I decided to install a later version 0f Fedora, ie Core 4, as I needed to do a complete install. This failed with the following message:
Loading SCSI driver
Loading Sata_sil driver disabiling IRQ#10
At this point the installation halted, although left for some time to check whether it resolved the problem. Install is by cd.
I next resorted to my original Fedora core 2 cd which allowed an uninterrupted installation. I updated via red Hat Network and have now 2 cores that I can select to boot :
Fedora Core 2.6.5-1.358 which boots up ok
Fedora Core 2.6.10-1.771 which halts during start up with the following message:
IRQ10: nobody cared (try booting with irq polloption.
handler [<e0872a51>] (ata_interrupt +0x0/0x1ec [libatal])
disabing irq #10
At this stage the system halted and failed to move on even when left for 30 minutes.

Reverting to core 2.6.5-1.358 system starts, however I find smartd fails to start.
Examination via services shows'smartd dead but subsys locked.
I also noticed that mdmpd failed to start and the following message is displayed when trying to restrt in services:
mdmpd failed: the errors: stopping mdmpd: [FAILED]
starting mdmod: kernel md module does not support events [FAILED]

smartctl commands appear to work as do the mdadm commands.
I am concerned that automatic monitoring will not take place if smartd is not running.
Also is mdmpd some means of monitoring the raid automaticaly?

The problem seems to point towards the raid card and the allocation of irq#10.
I do not understand how to boot with irq poll.
Any help / ideas would be appreciated.
I need to be able to monitor the state of the system remotely (especially the raid) so that I can take appropriate action in the event of any errors showing up. This I will do by having the error reoprt emailed to me. This part I have sorted out.
Many thanks
Roger

[compufix]

I have solved my problem by installing Fedora Core 6. The problem seemed to be a driver for the Raid Card as FC6 recognised the card and did not throw up the IRQ problem.
I now have FC 6 installed and booting from a Raid 5 arrangement; I have a non raid boot partition on first drive and system boots up fine.
I immediately noticed that whereas before my Sata discs were allocated 'hd-' they are now referenced 'sd-'.
This caused a problem in that although I now have smartd running ok on start up it is responding with 'smart monitoring not supported on this disk' or something similar. As I knew that they are smart capable I persevered. With FC 2 although smartd would not run, the smartctl commands worked ok.
I figured out that the FC 6 system believes the drives to be scsi which was causing improper reporting when using the same smartctl commands previously used successfully on FC 2.
I came across a useful article on gentoo-wiki.com/HOWTO_Monitor_your_hard_disk(s) which has enabled me to use smartctl commands successfully.
The trick is to use -d ata /dev/sd- after the command eg to check the Smart self test error log for sda I would use the following:
# smartctl -l error -d ata /dev/sda
It may be neccessary to enable Smart on each drive, (I did this anyway so I am not sure if it is required). This can be done by:
#smartctl -s on -d ata /dev/sd-
Note that - after sd in the above should have the relevant drive letter substituted.

I also find that mdmpd now starts ok.

I hope this may be of interest to others and hopefully save a lot of time if they find themselves in a sililar situation.

As a side issue I miss the Red Hat symbol in FC 6, I rather liked that sat in the bottom left of the screen!!