[Cialug] SataII and Debian Issue

[Chris K.]

I posted this to a forum and as is normal, it was quickly sidetracked 
and taken off subject.  If anyone has any thoughts on next steps to 
trouble shoot this, I'd be interested in hearing them.


I am having a strange issue with an array I am building.
The kit and all is at home, so I don't have *exact* errors I'm getting, 
but I think I describe them fairly well, at least until I get home and 
back to the logs.

The kit is pretty mediocre:
MSI K9AGM2-L AM2 AMD 690V Micro ATX
AMD Sempron 64 3400+ Manila 1.8GHz Socket 754
2x Kingston 512MB 240-Pin SDRAM DDR2 800
4x Seagate Barracuda 7200.10 ST3250620AS 250GB
Promise SATA300 TX4 (not currently in use, for future expansion)
Linksys EG1064 Gigabit Ethernet
Older Enermax 450W PSU

The software:
Debian 4.0 R3 AMD64 (Etch) via netinstall

Thats about it.

Now the weirdness.

When I received the pack of Barracuda's, I ran them through a couple 
iterations of the Seatools extended tests. This did discover a marginal 
drive which I RMA'd. But I continued with the project as I was planning 
on having a hot spare in the end anyway.

I installed all the hardware etc, and plugged the 3 drives into the 
onboard SATA controller (Promise based I think). I created a 3 member, 
5G, RAID1 for the OS and a 3 member RAID5 array using LVM for later 
expansion. The swap for this system is located on the RAID5 in a LV.

One thing I do need to mention is that I am running in Native Sata 
according to the BIOS. When I tried to set it to Legacy IDE, the system 
wouldn't boot even with just the 3 drives.

Everything is fine to this point. This point being when the replacement 
drive arrives.

I run it through the same Seatools tests as the others. Everything looks 
good.

I install the 4th drive, and plug it in to the last Sata connector on 
the motherboard and booted. BIOS sees the drive fine.

Once booted, all the arrays are degraded, which I sort of suspected 
would happen due to drive designator re-ordering (or whatever its 
called). I did the normal mdadm --manage stuff to get the arrays to come 
clean. At this point I was going to install the 4th drive as a spare 
when I noticed alot of *non-response, bus reset errors and waiting for 
drive to spin up notices.* (Will post EXACT errors when I get home)

After a couple of reboots and getting the same thing, but it appears to 
not always be the same drive as my arrays are degraded again. I am able 
to recover using the new 4th drive, because now it couldn't see one of 
the others.

At this point I thought I may be having a power issue so I moved 2 of 
the drives to another 12V leg (PSU supposedly has multiples). Same results.

So then I tried plugging one of the drives into the PCI TX4 card 
thinking that it might be a bad Sata port on the motherboard. Same results.

Could I be dealing with a weak PSU issue?

I would have thought this PSU was still capable of running a single 
processor board with 4 drives attached.