I've been having BIG problems, and the only thing I can put it down to is my WD HDDs.
To be specific, my non SE WD drives.
I have a Supermicro Dual Tualatin board that I use as my workstation, which has 4 Western Digital 80 GB SE drives in it, have been using Red Hat Linux for umm serveral years, and now Fedora.
Anyways, my workstation works fine - rock solid.
But my server on the other hand has been crashing regularily.
In my servers I use Promie controller cards.
I've got 3 Western Digital 120 GB SE drives, 2 80 GB SE drives.
One 40 GB & one 20 GB non-SE drives. All Western Digital.
I'd been having troubles with my first server, which was an old IBM Netfinity (Dual PII) which I managed to get for nothing. It worked fine to start with, but after a while it started crashing, after 15 - 20 days of uptime. But that 15 - 20 days gradually grew shorter and shorter. When the machine crashed it was a total lock up, nothing was ever written to the logs, it seemed it just froze instantly.
2nd Server MoBo was a Tyan Tiger 100 - Dual slot 1 PIII board.
Crash Crash Crash Crash - every 2 - 5 days.
Next I got hold of a MSI dual FC-PGA motherboard and two PIII 750s. And what do ya know, it crashed, and crashed and chrased. Ya know, I never had that setup up for more than 5 days. Always crashed/locked up before 5 days came around. My server is never to useless.
Right, that was it. I went out and got another Supermicro Dual Tualatin board for my server. A server board, rather than workstation board, though.
I've had the same problem with the Supermicro board, it only stays up for 2 or 3 days.
So, I've done some testing.
I disconnected all disks but the 20GB drive and isntalled Red Hat on it and let it run until it locked up. 2 point something days later the system was down. I tried it again. This time I got closer to 3 days.
So I tried the 40 GB drive by itself.
Exactly the same thing as the 20 GB.
Next I tried one of the 80 GB drives.
After 13 days it was still going. How long should I wait? It might not ever crash again.
I tried it with a 120 GB drive. No crash there after 13 days either.
Basically, I've tried all sorts of combinations of drives.
Non- SE drives working at the same time as SE drives.
OS installed on non-SE drives with SE drives attached.
OS istalled on SE drives with non-SE drives attached.
Anyways, whatever combination of drives I've tried in my server, when ever there is a non-SE drive attached - it will lock up.
All the testing was done on the Supermicro board.
I'm now running all SE drives in my server, and it's still going strong after 52 days.
My conclusion is that the non-SE drives are not to be touched.
Though I still don't understand why it should make a difference.
Any HDD should be able to go for more than 2 or 3 days.
Or the non-SE drives have a problem with Promise controller cards.
Then again, I have tried the drives while connected to the motherboards on board IDE connectors with the same results.
