I've seen posts on various forums that have similarities to the one I'm experiencing, but not quite the same.

I used the following iptables rule to preroute port 80 to port 8060:

iptables -A PREROUTING -t nat -i eth0 -p tcp --dport 80 -j REDIRECT --to-port 8060

When I first implemented this one prerouting rule a couple of weeks ago, it worked just fine. Yesterday, the server was shut down for some maintenance and brought up again. The prerouting rule was still in place, but the forwarding from port 80 to 8060 stopped functioning. I can internally reach the application on the server on port 8060 just fine, so it's the prerouting that's failing. I took the time to stop remove the rule, stop and start iptables, add the rule back and save it. It still doesn't work. /var/log/messages reflects that queries are making it to the server, but they're not being forwarded from 80 to 8060.

I've seen some forums that suggest that the ip_forward kernel value should be set to 1. But the fact that this worked initially and that another server with the same setting works (with ip_forward at the default of 0) makes me hesitant to implement the change.

Thanks in advance,
Keith

I have another server that has the same ruleset in place and is working just fine.

Could it be that the application is bound to a specific IP? I believe REDIRECT causes packets to be redirected to loopback (127.0.0.1).

Does telnet localhost 8060 work?

Is traffic to the lo interface allowed in the iptables rule set?

telnet localhost 8060 does connect.

I only have the one rule to preroute port 80 requests to port 8060. Up until the system was rebooted yesterday, this was sufficient to make things work as desired. So there's nothing referencing lo. But your question made me check ifconfig, and I just found that my Ethernet interface says eth2, not eth0. I'm not sure if or how that changed, but it looks like the next thing for me to try is change the interface to eth2 in the iptables rule.

Ser Olmy's suggestion helped me find and solve the problem. I'm not sure how it happened, but doing an ifconfig revealed that the Ethernet interface name changed from eth0 and eth2 between the server being brought down for maintenance and brought up again. When I dropped the rule pertaining to eth0 and inserted a new one using eth2 instead, things started flying again.

Thank you!

PLease mark the thread as Solved.