Sake Blok wrote:
http://ibctech.ca/http-conv.pcap
Yep, that's exactly what I was after!
In the trace you can see (by using "follow tcp stream" on one of the
retransmissions) that there is actually no response from the server.
Aha! I was following stream with one of the successful ones, and not the
transmission. Now I see that there is no response. Thanks.
This means the problem lies between the capturing device and the server
(either the request gets lost or the response gets lost). Since the
fibre is on the 'other' side of your capturing device, you can assume
the fibre is not causing the packet loss. However, since doing the
same test on the other side of the link does not show the problem, the
question is how are the two different.
This has me thinking... Instead of going over to the client site, I'm
going to plug a notebook into a switchport on the same vlan that goes to
the client site and test it from here.
This is very odd however. I go through the same edge device to
tigerdirect.ca as they do. The vlan switch is connected directly to the
edge router. This would indicate the edge router is the problem.
Unfortunately, I do not have a spare port I can span with on it,
however, being connected to the same interface but having success would
mean that....
hmmm...I wonder if I'm not requesting the same stylesheet...
Off to test.
Thanks all,
Steve