Problem with Mandrake 10.1 and system crash

[mke]

Hi,
currently i experience i little problem with my mandrake 10.1 setup. The system is based on my old athlon 1300 with 768pc133 and a 20gb maxtor disk. I have tested all hardware with a variety of tools and came up clean with no problems. The problem is that every fifth-sixth day the system stops responding correctly. I have set it up as a gateway where my internet connection resides and all traffic passes through it. The first problem that comes up every week is that the internet connection dies, altough i have set it up to reconnect. Another thing is that any application i might try to load then, freezes and stops responding. Next thing is the computer to stop responding at all (besided the mouse beying visible and moveable). If i try a hard restart the system then fails to detect my hd. If i switch off my pc for 10seconds and then load it up again most of the times it detects the hd and continues loading till it stops at a point when it loads rules and stuff. Switching it off for another 10-40 seconds resolves the problem again and continues to load. Again when everything has loaded nicely and worked for 5-6 days, the system starts the same thing all over again.
I thought that this could be a problem with the os but i have reinstalled it already once and took me over 3days to configure everything from start, but same thing happens again. I have changed cables, psu, and motherboard but everything seems to happen again.
Has anyone an idea what this could be? Finally, i am thinking of chaning my hd to a newer one that i have on store (60gb Maxtor) but i cant again reinstall everything from scratch cause it will take me a long time and i cant at this stage do this. Is there a utility that will clone my 20gb disk to the 60gb one and not cause any problems with the os?
Thanks again and sorry if i kept this long but didnt know how else to explain it.
If you have any alternative step/solution to this i would be glad to hear about it.

byez

 

[Nodsu]

Plenty of cloning utilities.

After the crash, please examine the system logs - messages and dmesg.
When the system "dies", can you still use the command line?

Monitor the system. Check basic things like memory usage, number of open files, running processes etc regularly. Linux is not that good at surviving rogue applications and will die if some app is too buggy.

 

[mke]

hi, i have checked few logs from some apps i use (verlihub, valknut, mysql) and all came clean. Are there any other specific logs i should also be checking cause i dont know where to start. When the system "dies", i cant use anything else, the system just stops responding to any command. Memory usage is quite low, have average 470mb free physical, and 100% free swap.
System opens over 200-300 connections for verlihub that i use. I have had though with this app no problems (and nore do the logs report anything) in the past as i use this over a year with 400-600 connections opened.
Sometimes though, when the system crashes and i try to restart it i hear a glich in the hd altough microscope diagnostic at full scan reports no problem, nore does the factory scan tool.
My last resort i think is the disk but i need to clone the whole drive somehow and i am not familiar with this at all.
anyways. thanks for replying to my post. if there is anything else that comes in mind please let me know.

byez

 

[Nodsu]

You can easily copy the the drive contents using the dd command if you don't feel like getting some special piece of software.

Since it is the system crashing you ought to look at the system logs. /var/log/messages is a good place to start. The memory useage may be low during normal operation, but we are interested in the usage statistics right before the crash.

 

[DelJo63]

have you updated the NIC drivers?

you can shutdown the network with
/etc/rc.d/init.d/network stop
and see if you're more stable

check swap space size ( should be 2x ram).
swapon -s
displays swap file size and usage (usage should be low but must no be >80%)

ps -aux | more
displays process + memory usage (under the VSZ and RSS columns as # of pages)