James Bottomley wrote:
On Thu, 2008-03-27 at 12:06 +0100, Primoz Kolaric wrote:
I have a strange problem with my SCSI subsystem. After a few days (7-10
days) of normal work the linux kernel starts to report these messages:
Mar 20 08:19:53 xxxx kernel: scsi0: PCI error Interrupt
The adaptec card gives this type of interrupt when it detects an error
on the PCI bus.
This cryptic piece at the end is the actual error:
Mar 20 08:19:53 xxxx kernel: scsi0: Data Parity Error has been reported
via PERR# in DFF1
Mar 20 08:19:53 xxxx kernel: scsi0: Split completion read data parity
error in DFF1
Mar 20 08:19:53 xxxx kernel: scsi0: Signal System Error Detected in DFF1
Mar 20 08:19:53 xxxx kernel: scsi0: Address or Write Phase Parity Error
Detected in DFF1.
But it's claiming an actual PCI bus parity error.
I already tried changing the scsi cables, terminators, running without
terminators, changing the SCSI controler card (unfortunately i only had
exactly same model) but nothing helps and i'm running out of ideas. I
have two identical machines: same motherboard, same scsi controler
connected to almost the same (difference is only in the number of hard
disk bays) external RAID units and it happens on both of them. On one it
happens regularly on the other it only happened twice in one year.
The same SCSI setup (same external raid and scsi controlers) was used
before on different motherboard and it worked ok, so i'm assuming that
the problem isn't between the scsi controler and the external RAID.
I'm afraid if the problem is on the PCI bus, changing the SCSI piece
won't necessarily help. Unless anyone with specific PCI advice can
chime in, about the best you can do is reseat the card (or preferably
move it to a different slot) and hope the error goes away.
I exchanged the machine for a different one (different chipset and cpu)
but left the PCI scsi controler. Since then, there weren't any PCI
parity errors. So i decided to send the Supermicro back for repair (or
at least checkup) since it's still under warranty.
Meanwhile another machine (same type of supermicro server, same scsi
controler, ...) experienced the same PCI parity error. The machine
worked fine for several months before, and nothing vital (no HW,
kernel, ...) was changed, so i'm assuming this error happens upon high
load and that it's not due to broken hardware (PCI bus) but due to some
SW bug.
Regards,
Primoz
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
