[rhn-users] kernel panic: journal commit I/O error

[Scott D. Anderson]

I'm running RedHat Enterprise Linux AS Release 3 (Taroon Update 6) on my
main server.  Since August, the /home1 and /home2 partitions have been
using LVM on RAID5 with snapshot partitions being used for tape dumps.
The actual filesystems are ext3, which has journaling.

Twice now, I've gotten the following message on the console:

Message from syslogd at puma at Tue Nov 15 04:02:26 2005 ...
puma kernel: journal commit I/O error
 
Message from syslogd at puma at Tue Nov 15 04:02:26 2005 ...
puma kernel: Assertion failure in journal_flush_Rsmp_e2f189ce() at
journal.c:1356: "!journal->j_checkpoint_transactions"
 
Message from syslogd at puma at Tue Nov 15 04:02:30 2005 ...
puma kernel: Kernel panic: Fatal exception
 
FYI, at 4:02am, the current snapshot is removed, a new one is created, and
a tape dump is begun.  The effects of this error are that (1) I can no
longer "lvremove" the snapshot partitions, because the system says they
are in use, but they're really not; and, (2) I can't do tape dumps because
the "sync" system call hangs.

These can both be fixed by rebooting, but I don't want to be rebooting my
server in order to do tape dumps.  I have two questions:

1. Should I be reporting this kernel error to anyone?  If so, to whom and
how?
2. Any suggestions about how to avoid this or fix it?

Thanks,

Scott

--
Scott D. Anderson
Computer Science Department
Wellesley College
Scott.Anderson at acm.org