I have a brand new system with RHEL 5 Server. For some odd reason, some of my network connections go south on me after the system has been on for an hour or 2. Although pings still work, I cannot connect in via VNC (Error: socket error) or remotely administer my Kerio MailServer application (Error: Application Not Running). Since pings work and a reboot fixes the problem, I don't think this is any sort of a hardware problem.

It won't work very well to have to reboot our mail server every hour or so! Any ideas?

Clues in /var/log/messages?

It would appear that my connections are fading in and out. If I am patient, I can keep telling things to reconnect and eventually they do. I am beginning to be concerned about the NIC maybe being strange.

I looked in /var/logs/messages and nothing looked odd to me, just seemed to be normal start-up and shutdown kinds of things. If this is a hardware problem, what would be my best method to diagnose the issue?

Maybe I shouldn't be looking at the NIC, because I can run tons of pings and get 100% responses. It is almost like a TCP problem...I don't know! I think I'm lost on this one.

Is /sbin/ifconfig reporting error packets?

Does restarting the network service 'service network restart' solve the issue ?

Also check iptables rules .

root# iptables -L

thnks

Are these Broadcom Gigabit NICs?

Please post the output of ifconfig for the given interface.
For these type of problem you have to start narrowing it down to which of the 7 layers of the OSI model the problem exists. Given that initially we will all thing network problem, there are many tests to do...
ping...traceroute...mtr...etc etc...that will tell you if you can eliminate the network as being a problem.
In the first instance we should look for issues closer to home on the server.

Like anomie says, first, you will need to look for any increasing error/dropped/frame/carrier count in case of a duplex mismatch.