[TECK]

alexi, since you do not want to post your exact server configuration, I'm going to guess.
Double posts happen when your current Unix setup traceroute dies on a specific hop. More exactly, if you run a traceroute and you get a request timeout. That is the critical moment when you could have a double post initiated... because the data is transmitted in bits.

So it is very simple to solve your problem. Run a traceroute and see how is your response. You will be able to identify right away if you are dealing with a heavily congested network, an error in the routing table or a faulty connection. After you have determined the bad location, you will need to use a utility such as ping to continue troubleshooting.

EDIT: I could resume the above posts to:
Hi, my car is not working. What do you think is wrong?
Well, it could be either the engine or the transmission.
Aha, I thought so, so what do you think is broken?
I don't know, maybe you could start by telling me what is your make and model...