System Freeze - like clockwork

[Guest Phat Penguin]

I am running Mandrake 9.0 / 2.4.20-1 kernel on a AMD900 machine.

 

I am getting hard lock system freezes every four days, like clockwork. This sytem runs beautifully - until the fourth day and it hard locks. Same time, same day - every time.

 

I have removed the Nvidia drivers thinking this was the problem, but alas - No.

 

dmesg and messages are not helping.

 

I would be appreciate it if anyone could point me in the right direction and what to look for to solve this frustrating problem. I do not to have to re-install if at all possible.

 

Thanks

[Guest duir66]

Past rebooting every three and a half days there may not be any other option other than a re-install, maybe re-compile.

 

The chipset drivers could be at fault, maybe cpu temp (but unlikely).

 

This is very odd, but does sound like a hardware driver issue.

 

If it worked well with 2.4.19.x you might want to consider going back.

 

Email someone at Mandrake with detailed description of your hardware config. You may have found a bug they (and rest of linux community) can deal with.

 

Good Luck.

[Cannonfodder]

Could there be some job that is running every 4 days?

[Guest Phat Penguin]

Thanks guys for the reply ...

 

Cannonfodder .. I checked and could not find anything the runs a job every 4th day ... odd and frustrating. It bound to be something I have done - but since much tweaking has occured God knows where I have stuffed up.

 

duir66 - full install will be on the cards and I might just grin and bear this (not a mission critical system) until 9.1 appears ... which hopefully is only around the corner.

 

Thanks again,

 

PP

[Guest JaseP]

That just seems to me to be too much like a job that's running to dismiss that that easily.

 

Remember that you have 2 job running apps going in a standard Mandrake install, chrond, and anachron (I think).

 

Maybe one or the other is running the "bad" job.

 

If it's a heating issue, it would happen based on CPU load (more load, more proccessing, more heat), not based on simple timing...

 

Just something to consider.

[Guest joehill]

one example, my machine runs a job every day at 4 am, one night I was up late and my hard drive just started thrashing by itself, scared the shit out of me.

 

it turned out it was something called msec, a security / permissions auditor

 

anyway, don't reinstall, it's gotta be something simple. try running crontab -l and see if there is anything in your user cron file. then try the same as root.

[sglafata]

That just seems to me to be too much like a job that's running to dismiss that that easily.

 

Remember that you have 2 job running apps going in a standard Mandrake install, chrond, and anachron (I think).

 

Maybe one or the other is running the "bad" job.

 

If it's a heating issue, it would happen based on CPU load (more load, more proccessing, more heat), not based on simple timing...

 

Just something to consider.

 

I had the exact same problem, only my machine was locking up on a daily basis. I resolved it by separating each item out in the crontab file.

 

Let me explain: The basic crontab file runs cron.hourly, cron.daily, cron.weekly, and cron.monthly. Inside those directories, which are located in /etc, are separate cron jobs. I listed each cron job from each respective cron.<folder> line by line in crontab.

 

Never had a problem again. Remember that there are other individual cron jobs running that are not listed in crontab. I believe postfix has a cron job. You may need to check your logs to see what is running at that particular time of day.