On Mon, Aug 15, 2005 at 10:41:42PM +0000, David Gore wrote:
It is possible the cores are caused by running two clients on the same host both redirecting to ~hobbitclient/tmp/msg.txt. That would explain my occasional rash of purples followed by greens at least. I am going to work around that.
I dont think so. It is your server modules that are crashing, so what happens on the client shouldn't matter at all.
Unfortunately, the backtrace you've sent from the core-files doesn't reveal much of where the crash happens - which indicates that there's something that thrashes the stack and causes the crash.
To begin with, I'd like you to add "--debug" to the hobbitd command line. This will cause a lot of output to go into your hobbitd.log file, the interesting bits obviously is what happens when it crashes. I'd like to see the full log file, though - I'll e-mail you details of where you can upload it since it's probably too large for e-mail.
If I remove most of the entries from bin/hobbitclient-osf1.sh will the back-end get annoyed and engage in odd behavior? Or will it simply not send cpu, procs, disk, etc.. to the web page?
It should simply stop sending the messages it doesn't have data for.
Regards, Henrik