kernel crash

[Gilboa Davara]

On Tue, 2010-08-17 at 09:44 -0400, Steve Blackwell wrote:
> I leave my computer on 24/7 so that my backups can run at night.
> Lately, it has been crashing during the night usually leaving no trace
> of what happened. Last night it crashed but left this
> in /var/log/messages:
> 
> Aug 17 01:04:56 steve kernel: INFO: task kjournald:1960 blocked for more than 120 seconds. 
> Aug 17 01:04:56 steve kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. 
> Could a hard drive get shut down because it was getting too hot? What would be a normal temp for a hard drive that has just completed a backup? 124C seems really hot. The HD cooling fan had been  broken so I replaced it this past weekend but it doesn't seem to have helped. Too late? Permanent HD damage already done?
> Any other comments or suggestions?

Hello Steve,

This is not a crash.
The kjournald kernel process (which handles various file-system task).
You assumption that the HD went into some type of sleep/suspend mode
during write sounds reasonable to me.

124C seems -very- hot. Even during heavy I/O.
Two things spring into mind:
A. Is it a normal desktop SATA drive or high-speed SCSI/SAS drive?
B. Please post the SMART log of the drive. (smartctl -a /dev/sdX). 

- Gilboa