Re: [ns] trace for packet drop

[Tarik Alj <aljtarik@cholla.inrs-telecom.uquebec.ca>]

>Date: Thu, 22 Jun 2000 08:48:11 -0700 (PDT)
>From: Yan Yu <yanyu@catarina.usc.edu>
>To: ns-users <ns-users@isi.edu>
>Subject: [ns] trace for packet drop
>MIME-Version: 1.0
>
>Hi, there,
>   I have a wierd record entry for packet drop in my trace file. Please
>see the wierd trace entry and its explanation at the end. 
>My question is that can I trust the trace file? where can I add some code
>or to trace in debugger to verify that it really write the corrrect
>information to the trace file? since except wierd trace record,
>everything else seems normal in my simulation, I suspect the trace entry
>is not correct..
>Thanks in advance!
>yan
>
>
>--------------------------
>the topology is 
>2(FH) == 0(BS) -- 3(LAN) - -- 1 (MH)
>
>the traffic is from 2-->1, but the record for drop is something like
>
>d 4.855052 1 3 tcp 1000 ------- 0 2.0 1.0 783 1576
>
>the 3rd field "1" means src in MAC layer header,
>the 4th field "3" means dst in MAC layer header,
>2 in "2.0" means  packet src(end-to-end src, different from MAC layer,
>hop src)
>1 in "1.0" means  packet dst(end-to end dst) 
>              
>the wierd thing is that data packet should go from 3-->1 instead of
>"1-->3" in the trace. From my understanding, the correct trace entry for
>"drop"  should be the 3rd and 4th field is switched, since the data pkt is
>going from 2-->1, so it should be:
>
>d 4.855052 3 1 tcp 1000 ------- 0 2.0 1.0 783 1576 


as I understand it, this drop happened on the LAN; is 3 the adress of the 
LanNode, e.g. the LAN itself or are node 0, 3, 1 on the LAN? Anyways, the LAN 
tracing function is in vlan.tcl, LanIface::trace; the order of trace might be 
reversed for some reason (this used to be the case, but it should have been 
fixed by now, check your file version with the CVS tree) you can swap it there 
and see if that fixes it. Beware, it can be confusing since the LAN itself has 
an address...

Tarik