On Thu, 18 Dec 2003, vikram mohite wrote: > To understand the problem we removed IPTABLES rules, cleared the cache again > and put ethereal on client machine. When the problem reoccured we captured > the entire TCP stream. We again cleared the cache and opened the page > captured which immediatly reproduced the problem. The problem was also > reproduced on all other client machines accessing the proxy .
Can you please get a ethereal or tcpdump -s 1600 trace of the poisoning traffic on the Squid proxy? tcpdump -s 1600 -w traffic.dump -i any The trace should include 1. The requests which was identified as causing the cache poisoning 2. One request showing that the cache is poisoned. > Strangly I have not been able to reproduce the problem on any other squid > proxy running same versions of squid ( diff hardware config but same > squid.conf ) Odd.. and there is no difference in network connectivity? Including which DNS servers are used, ISP etc.. Regards Henrik