I think this is not a nistnet problem, but rather just tcpdump incompatiblity with nistnet. Unless you can show the problem with a tcpdump on one of the end points of the communication, or an intermediate router which isn't the nistnet box, I don't believe you have a problem.
W/o nistnet, packet flow is from interface A to interfaceB's outbound queue. It wouldn't surprise me that tcpdump is capturing the packet when it is placed in the queue before nistnet intercepts it for the delay and then a 2nd time after the delay. On Sun, 11 May 2008, Monia Ghobadi wrote: > Hi, > > As you mentioned, the problem appears only when the dumps are captured on > the NISTNET machine. > > Something interesting to mention is that: > Assuming in your simple topology machA is sending TCP traffic to machB: > > [machA] <-------------eth1-<NISTNet>-eth2----------> <machB> > > and the delay is set to be: cnistnet -a "machB-IP" "machA-IP" --delay 100ms > (on the direction of Ack packets) > > dumping on eth1 would only result in occasional duplicate packets. > However, if the delay was set as cnistnet -a "machA-IP" "machB-IP" --delay > 100ms (on the direction of data packets) > then if I dump on eth1, every single packet is reported twice. > > It seems that dumping in the same direction as the delay is more of a > problem with NISTNET ! > > > Monia > > > > > On Fri, May 9, 2008 at 9:16 PM, David Morris <[EMAIL PROTECTED]> wrote: > > > > > A couple folks have reported duplicate packet issues associated with > > setting a delay. > > > > In at least one report, the tcpdump capture was performed on the nistnet > > system. I just attempted to recreate the problem. With: > > > > [machA] <-> <NISTNet> <-> <machB> > > > > I performed a ping from A to B and was running etherial on machA to > > capture packets. I used a simple "--delay 1000" for each direction. > > I saw no indication of duplication. > > > > I suspect this is a case of the packet capture being incompatible > > with the way NISTNet hooks the network on the NISTNet system. > > > > Can either of you reproduce the problem when capturing packets external to > > the NISTNet machine? If so, please describe your setup in enough detail > > for me to replicate it. In particular, what were the 'cnistnet' commands > > used? > > > > Dave Morris > > _______________________________________________ > > nistnet mailing list > > nistnet@antd.nist.gov > > http://www-x.antd.nist.gov/mailman/listinfo/nistnet > > > _______________________________________________ nistnet mailing list nistnet@antd.nist.gov http://www-x.antd.nist.gov/mailman/listinfo/nistnet