> Jul 12 23:24:09 kernel: SCSI disk error : host 0 id 1 lun 0 return code= 2
> Jul 12 23:24:09 kernel: scsidisk I/O error: dev 0812, sector 7372840
> Jul 12 23:24:09 kernel: Kernel panic: EXT2-fs panic (device 8/18):
> ext2_read_inode: unable to read i-node block - inode=921604, block=3686420
>
> 8/18 is my root filesystem on a seagate baracuda. I read that these get
> REAL hot--maybe it is heat induced? This is a normal compaq tower server
> with stock fans, and this error has happened multiple times (with
> different sectors and inodes).
Yep, this would be my guess. Is the pattern of errors consistent with
the heat theory? I.e., do you not see any errors after a long power-off,
but do after sustained activity? I bet that were you to leave it off for
a few hours and reboot it, then you wouldn't see any errors for a while,
and that were you to undertake any heavy file-access programs, then you
would start seeing them.
> Fsck comes up clean after a hard reboot
> (which is impossible but true).
Not impossible, happens to me all the time (well, on the rare occassions
that I have to hard reboot). ext2 really is awesome.
> Please point me to the proper place if I am way off on posting to this
> group. I am not sure what this falls under, but the log says kernel: :)
One of the help groups might have been more appropriate, but indulgence is
granted. 8^) A lot of problems caused by power or heat problems are often
misdiagnosed as kernel problems.
You might want to try moving the drives to external cases with big fans
and cooling off your computer room.
_____________________________________________________________________
Todd Graham Lewis Linux! Core Engineering
Mindspring Enterprises tlewis@mindspring.com (800) 719 4664, x2804