We are experiencing some problems using: - cttproxy, version cttproxy-2.4.18-10, applied without any apparent problem on a Linux 2.4.9-e.10 kernel, derived from the Redhat "Advanced Server" kernel. - zorp 2 We use an udp Plug proxy, defined as follows: def udp(): Service("udp", PlugProxy) Receiver(SockAddrInet("10.0.0.191", 20001), "udp") Receiver(SockAddrInet("192.168.1.1", 20001), "udp") Receiver(SockAddrInet("192.168.7.1", 20001), "udp") and used through iptables rules, like: [root@pif root]# iptables --list -t tproxy Chain PREROUTING (policy ACCEPT) target prot opt source destination TPROXY udp -- dmz.sede-pisa.link.it gondor.sede-pisa.link.it udp spts:1024:65535 dpt:domain TPROXY redirect 0.0.0.0:20001 TPROXY udp -- dmz.sede-pisa.link.it gondor.sede-pisa.link.it udp spt:domain dpt:domain TPROXY redirect 0.0.0.0:20001 ... With tcpdump, we observe that DNS queries coming from the internal networks (both from a DNS server from the 53 port and from DNS clients on non privileged ports) are sometimes redirected to the 20001 port, sometimes not. Obviously, when not redirected, zorp is not activated and the query doesn't work. Do you have any idea of what's wrong in our environment? Thanks a lot, tito.