Packet Source Destination Delta Time Summary 77 192.168.1.1 10.1.3.5 0.095424 "Src=60761,Dst= 4000,.A....,S=2329514902,L= 0,A= 726811823,W=32768" 86 10.1.3.5 192.168.1.1 0.062624 "Src= 4000,Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 87 10.1.3.5 192.168.1.1 0.000704 "Src= 4000,Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 90 10.1.3.5 192.168.1.1 0.002688 "Src= 4000,Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 91 10.1.3.5 192.168.1.1 0.000672 "Src= 4000,Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 92 10.1.3.5 192.168.1.1 0.002560 "Src= 4000,Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 94 10.1.3.5 192.168.1.1 0.018656 "Src= 4000,Dst=60761,.A....,S= 726812335,L= 512,A=2329514902,W=11468" 103 10.1.3.5 192.168.1.1 0.115520 "Src= 4000,Dst=60761,.A....,S= 726812335,L= 512,A=2329514902,W=11468" 105 10.1.3.5 192.168.1.1 0.000832 "Src= 4000,Dst=60761,.A....,S= 726812335,L= 512,A=2329514902,W=11468" 106 10.1.3.5 192.168.1.1 0.003520 "Src= 4000,Dst=60761,.A....,S= 726812335,L= 512,A=2329514902,W=11468" 110 10.1.3.5 192.168.1.1 0.060064 "Src= 4000,Dst=60761,.A....,S= 726812847,L= 512,A=2329514902,W=11468" 112 10.1.3.5 192.168.1.1 0.002496 "Src= 4000,Dst=60761,.A....,S= 726812847,L= 512,A=2329514902,W=11468" 113 10.1.3.5 192.168.1.1 0.000672 "Src= 4000,Dst=60761,.A....,S= 726812847,L= 512,A=2329514902,W=11468" 115 10.1.3.5 192.168.1.1 0.002880 "Src= 4000,Dst=60761,.A....,S= 726812847,L= 512,A=2329514902,W=11468" 123 192.168.1.1 10.1.3.5 0.032192 "Src=60761,Dst= 4000,.A....,S=2329514902,L= 0.A= 726812335,W=32768"However, the delta times are very small. It is possible that these transmissions were triggered by the fast retransmission algorithm, but if that were the case I would expect to see ACKs arriving from 192.168.1.1 to trigger the retransmissions, that is not the case.
The data in the complete trace tells the real story. The key is to look at the IP-ID value in each packet. Each unique IP packet should have a unique ID value (at least until the 16 bit value wraps back to 0). In this case all the packets with the same sequence number have the same ID value.This tells me that these are not retransmissions. Now looking at the Time To Live (TTL) value tells me what the problem is. Each time though a router the TTL is decremneted, which is what we have here.
Packet Source Destination Delta Time IP-ID TTL Summary 86 10.1.3.5 192.168.1.1 0.062624 11583 63 "Src= 4000,Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 87 10.1.3.5 192.168.1.1 0.000704 11583 62 "Src= 4000,Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 90 10.1.3.5 192.168.1.1 0.002688 11583 61 "Src= 4000,Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 91 10.1.3.5 192.168.1.1 0.000672 11583 60 "Src= 4000,Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 92 10.1.3.5 192.168.1.1 0.002560 11583 59 "Src= 4000,Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468"Finally, look at the physical addresses which will tell you who is actually sending the packets and makes the looping obvious. We have 00:10:7B:xx:yy:01 sending to 00:0D:65:xx:yy:02 which sends it back to 00:10:7B:xx:yy:01. This happens twice before 00:10:7B:xx:yy:01 sends the packet to 00:0C:85:xx:yy:03 instead of back to 00:0D:65:xx:yy:02.
Packet Source Source Physical Destination Dest. Physical Delta Time IP-ID TTL Summary 86 10.1.3.5 00:10:7B:xx:yy:01 192.168.1.1 00:0D:65:xx:yy:02 0.062624 11583 63 "Src= 4000.Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 87 10.1.3.5 00:0D:65:xx:yy:02 192.168.1.1 00:10:7B:xx:yy:01 0.000704 11583 62 "Src= 4000.Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 90 10.1.3.5 00:10:7B:xx:yy:01 192.168.1.1 00:0D:65:xx:yy:02 0.002688 11583 61 "Src= 4000.Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 91 10.1.3.5 00:0D:65:xx:yy:02 192.168.1.1 00:10:7B:xx:yy:01 0.000672 11583 60 "Src= 4000.Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468" 92 10.1.3.5 00:10:7B:xx:yy:01 192.168.1.1 00:0C:85:xx:yy:03 0.002560 11583 59 "Src= 4000.Dst=60761,.A....,S= 726811823,L= 512,A=2329514902,W=11468"So the issue has nothing to do with fast retransmissions but with a loop in the routing tables. The really interesting thing is that the trace is full of these sequences of packets. Each packet from 10.1.3.5 loops between 2 and 4 times and then gets sent to 00:0C:85:xx:yy:03. What I'd love to do is get a look at the routing table on 00:10:7B:xx:yy:01 to see why it switches between 00:0D:65:xx:yy:02 and 00:0C:85:xx:yy:03.