[632714.975844] ata1.00: status: { DRDY ERR }
[632714.975847] ata1.00: error: { IDNF }
IDNF - ID not found, usually these days means 'asked for a sector out of
range'
Justin, was this error provoked for example by you using a 'dd' command to
try and read stuff past the end of the disk? Because the md device driver
should not try and do something like that. The md device driver knows how
many LBAs are available on the disk and should never issue a read or write
past the end of the device.
Justin, I have the impression that this IDNF error was NOT due to you issuing
dd commands or similar, but was part of 'normal' md operation.
Alan, am I overlooking something? (One feature that might provide such
errors is the 'host reserved address' set of commands that allow you to
'clip' the disk capacity.)
Most of the errors were during normal md/Linux operation. However, I will
also note that with a 16-port 3ware controller, drives were also
having problems, different errors mind you but they were rather nasty.
The errors occured on RAID-1, RAID-5 and RAID-6 and even on a single disk,
no md/Linux, I kept running bonnie++ with different filesystems on one
disk, ext3 and XFS, both of which started erroring out.
When you say 'the errors occured' do you mean specifically the 'IDNF'
errors? Those are the onces I am talking about. I am NOT talking about
other types of errors like UNC or ABORT. I am ONLY talking about the IDNF
errors.
I removed all of the Velociraptors and replaced them with good old
Raptor150s and tested with md/Linux and the 3ware card, no problems with
either configuration. I truly believe something is wrong with the
Velociraptors at this point since both md/Linux and 3ware both had problems
with the drives.
It is easy to check if the IDNF errors are from the disk firmware or
the Linux kernel.
The kernel error message for IDNF should report the LBA which provoked
that IDNF error. Please compare that with the maximum LBA on the disk,
which you can get either via hdparm or smartctl -i (divide total capacity
in bytes by 512).
If the LBA reported in the kernel error message is greater than the
maximum LBA on the disk, then the problem is in the Linux kernel. If the
LBA reported by the kernel is less than the maximum LBA on the disk, then
the problem is the disk firmware (or perhaps the disk has its capacity
'clipped' with the 'Host Protected Area' feature set. But the Linux
kernel should know about this, so that would be a kernel bug again, not a
disk firmware bug).
Cheers,
Bruce
--
To unsubscribe from this list: send the line "unsubscribe linux-ide" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
