hello there,

Recently I use the fedora core 5 (the kernel version is 2.6.15) for development.

I have three desktops, one is windows XP, the other
two are fedora core 5 desktops. Often, I connect to linux sytem
through putty from windows XP, or interconnect between fedora5
desktops using ssh.

When (1)I copy large files using scp between two linux desktop or
using sftp between windows xp and linux desktop
(2)use dmesg to display too much messages

The linux kernel will crash and reboot.

I can not fix the problem. Anyone else encounter this problem?

What's wrong?

Thanks a lot.

First of all FC5 is unsupported and unmaintained. If you like Fedora you should keep up with the release schedule. Second what ways does this happen? Only copying from FC5? Only copying to FC5? Both? One FC5 machine? Both? Are both stock machines with stock software and kernels? No other tweaking like overclocking? What's the max size that makes copying not crash the box? What's the actual OOPS message? Did you run it through ksymoops?

In fact, I doestn't like FC series, but I have to use it for current tasks. Please do not always recommend to update the current release encountering any problems if we can not figure out the deep reasons.

Crashing happened due to the way I used "ssh", but I can not find out
the mount of files when the kernel was crased and system was rebooted. Because the crash is not always, it is occeasional. There is no any hint in /var/log/messesages.

Fedora is special in that its team (purposely) rolls out releases at a rapid pace. You're running an unsupported version, and it will not be possible for you to get bug fixes unless you are willing to patch / hack source code yourself.

If you insist on running an old, unsupported version of Fedora, I recommend the following free PDF book to help troubleshoot your issue: http://www.linux-books.us/linux_general_0013.php

Good luck.

You should have posted FC5 was a fixed requirement in your OP. Regardless of the reason or fix I am compelled to post, for those that didn't get it yet, that running old releases has certain consequences. Next to that I asked you very specific questions which you chose to answer in a more or less roundabout way. If you need to have the reboots fixed you'll have to invest time in setting up enough kernel and userland auditing and some test scenario's to actually get messages one way or the other. You could for instance run a debug kernel, run sshd in verbose mode (though I doubt the problem is with SSH), turn up syslog.conf to log everything and maybe output to a separate syslog server if you think messages get lost locally.

Thanks for your suggestions!
I will try.