[cobalt-users] Runaway Analog

["Fred Damstra" <fdamstra@xxxxxxxxxxxxxxxx>]

I know others have had this problem, but I have yet to see a solution.

Every morning between 4 and 8am, my Raq4 reboots.  I narrowed it down to
cron, then logrotate, then split_logs, and finallly: Analog. 

Running Analog seems to eat all of the available cpu, and never
finished.  Eventually, the system just reboots.  There is nothing in the
syslogs, but it does appear that it drops core before doing so.

I just started having this problem after the Kernel Update 2.0.1 C33
patch.

The only suggestion I've found is that I disable Analog, which is very
near the bottom of our list of options.  Of course, not rotating the web
logs is also not an option.

Has anybody ever solved the problem of a runaway analog job?  Does
anybody know why the server would reboot when it fails?  Is there a
watchdog process of some sort?

Thanks,
Fred