Hi,
I am running kamikaze 7.09 on a Linksys wrt54.
I have setup a lot of port forwarding rules to make some servers available to the outer world.
Everything works fine for quite some uptime (usually hours - sometimes days).
After a while (quite likely under heavy traffic load), the internal port forwarding rules get messed up:
* the router starts to redirect packets to different destinations (the destination IP or the destination port changes)
* the firewall rules (as displayed by "iptables -L -vn" and "iptables -t nat -L -vn") still seem to be correct
The problem has the following symptoms:
1) packets do not get forwarded -> the service is unreachable
2) "DROP" messages of the specific packets start to appear in "dmesg"
Restarting the firewall scripts does not help.
Reloading of ipt-modules does not help.
Reboot helps -> afterwards the router works as expected again.
I tried exactly the same setup with two different Linksys WRT54. Both show the same behaviour.
Somehow I have the feeling, that the internal representation of the iptables rules gets messed up under certain conditions.
I don't see any error logs in dmesg (except for the DROP packets).
My current (ugly) "workaround" for this problem is a cron job, that runs every minute and reboots the machine, if it greps the specific DROP packets in "dmesg". That results in quite a lot of downtime (while rebooting) - but at least there are no hours of downtime (this happened without reboot).
The same devices worked flawlessly with white russian before - with a similar setup.
Do you have any ideas, how I can debug this issue?
Should I attach some information (ouptut of "iptables -L -vn" or dmesg)? It only works inline in this forum - right?
Any helping advice would be great!
thanks for your time,
Lars