I'm having a very strange problem and I was hoping that maybe someone has seen this before. Note that IP addresses have been sanitized. I have a Red Hat 7.3 firewall using iptables 1.2.5 and kernel 2.4.18-10 separating two subnets. The eth0 subnet contains a DNS server and the eth1 subnet contains a DNS client. The firewall uses connection tracking and the ruleset permits DNS queries from the client to the server. 192.168.10.20 ------ eth0-FIREWALL-eth1 ------ 192.168.3.8 DNS Server DNS Client Rule fragement from FORWARD chain: ACCEPT all -- 0.0.0.0/0 0.0.0.0/0 state RELATED,ESTABLISHED LOG udp -- 192.168.3.8 192.168.10.20 udp dpt:53 LOG flags 0 level 4 ACCEPT udp -- 192.168.3.8 192.168.10.20 udp dpt:53 The DNS client box is running squid, which happens to generate pairs of similar DNS queries for some reason. What is happening is that *one* of the two queries gets dropped most of the time as it crosses the firewall. Observe these tcpdumps: firewall# tcpdump -i eth1 host 192.168.3.8 and port 53 tcpdump: listening on eth1 11:54:15.132008 192.168.3.8.32772 > 192.168.10.20.domain: 18+ A? www.google.com. (32) (DF) 11:54:15.132034 192.168.3.8.32772 > 192.168.10.20.domain: 19+ A? www.google.com. (32) (DF) 11:54:15.171377 192.168.10.20.domain > 192.168.3.8.32772: 18 1/4/2 A www.google.com (152) (DF) firewall# tcpdump -i eth0 host 192.168.3.8 and port 53 tcpdump: listening on eth0 11:54:15.156014 192.168.3.8.32772 > 192.168.10.20.domain: 18+ A? www.google.com. (32) (DF) 11:54:15.171337 192.168.10.20.domain > 192.168.3.8.32772: 18 1/4/2 A www.google.com (152) (DF) Note that the DNS query with query ID 19 has disappeared somewhere between coming into eth1 and exiting eth0. The logging rule I have in iptables does show both of those packets right before they hit the ACCEPT rule: IN=eth1 OUT=eth0 SRC=192.168.3.8 DST=192.168.10.20 LEN=60 TOS=0x00 PREC=0x00 TTL=63 ID=0 DF PROTO=UDP SPT=32772 DPT=53 LEN=40 IN=eth1 OUT=eth0 SRC=192.168.3.8 DST=192.168.10.20 LEN=60 TOS=0x00 PREC=0x00 TTL=63 ID=0 DF PROTO=UDP SPT=32772 DPT=53 LEN=40 I'm kind of at a loss here, does anyone have any idea what could be going on? I'll be happy to provide any additional info that I can. Thanks! -- Steve Snodgrass * ssnodgra@pheran.com * Network and Unix Guru(?) at Large Geek Code: GCS d? s: a C++ U++++$ P+++ L++ w PS+ 5++ b++ DI+ D++ e++ r+++ y+* "If you want to be somebody else, change your mind." -Sister Hazel