Andy Green wrote:
Roger wrote:
Jul 16 19:50:18 asterix kernel: ata1: port reset, p_is 0 is 0 pis 0
cmd 4c017 tf 7f ss 0 se 0
Hm there it seems to be resetting the device
starts to just die and then i guess switches to read-only mode and i
cant even use ssh to get into the machine, such that i always reset
it, but when i do it just runs for 4 days max and then just dies, do
you reckon it to be the hard drive that is now weak or what.
I read some time ago there is some design tradeoff for the bearings on
hard drives, they can be optimized for running a long time or for
stopping and starting like on a laptop drive. Maybe that is something
to do with it.
Your problem does sound about the same as what I experienced, also on
x86_64, mine was an nVidia chipset DFI Lanparty board.
I have since replaced that server with a smaller machine running on
FC3 with but uses a lot less memory and does not complain at all.
My box was also up 24hrs being a mailserver. I replaced it with an
embedded ARM board that draws 1.5W and a USB flash pen. I wrote an
article about it here: http://warmcat.com/_wp/?p=5 But of course it
just does my mail, not 1000 users.
Do you reckon it could be the SATA Drivers which kill the hard drives!
Sorry I don't have any answers, I dropped that machine after the second
drive was killed and it is sitting in the corner looking lonely. But
then I would think more people would be hitting this problem if it was
simply a driver issue. Maybe it requires extended uptimes on the box in
addition?
-Andy
I am running an AMD 3800 x2 without any problems on the newest kernel on
and gigabyte GA-K8NF-9 without any problems. I have two IDE drives and
two SATA drives and everything is fine. So I have to say there is no
problem with the x86_64 kernels. I have been happy so far except for the
32 bit only stuff like flash.
Peter
