bad inode number followed by ext3_abort and remount readonly

[David Shaw]

I have seen this happen a number of times:

Jun 13 13:58:16 n202 kernel: EXT3-fs error (device sda5): ext3_get_inode_block: bad inode number: 9
Jun 13 13:58:16 n202 kernel: Aborting journal on device sda5.
Jun 13 13:58:16 n202 kernel: EXT3-fs error (device sda5): ext3_get_inode_block: bad inode number: 9
Jun 13 13:58:16 n202 last message repeated 6 times
Jun 13 13:58:18 n202 kernel: ext3_abort called.
Jun 13 13:58:18 n202 kernel: EXT3-fs error (device sda5): ext3_journal_start_sb: Detected aborted journal
Jun 13 13:58:18 n202 kernel: Remounting filesystem read-only

Once this happens, things break quickly (/tmp being readonly, as a
start).  Upon reboot, a manual fsck is required, after which the
machine is operational again.

This particular example is a SATA disk, but it has happened to a
regular old IDE disk as well.  It is always the root partition.  The
bad inode number varies (but is always either 3 or 9).  There are no
other errors about the disk in the log.

Kernel: 2.6.11.7
e2fstools: 1.35 (28-Feb-2004)

Any thoughts on how to proceed here?  Unfortunately, I'm not able to
duplicate this at will.

David