Re: different behavior between two ns version

[John Heidemann <johnh@ISI.EDU>]

On Tue, 01 Sep 1998 16:57:40 +0200, Arnaud Legout wrote: 
>John Heidemann wrote:
>> 
>> On Fri, 28 Aug 1998 19:06:23 +0200, Arnaud Legout wrote:
>> >hi,
>> >when I run the attached script with the ns2.1b3 daily snapshot from may
>> >25, 1998
>> >everything is ok.
>> >But when I run the same script with the *real* daily snapshot  I have
>> >the following
>> >message :
>> >warning: classifier _o31 unknown flow s:0, d:256, fid:0, bucket:0
>> >_o31: no target for slot -1
>> >
>> >have you an idea about that?
>> 
>> The default target of an agent was changed from 0 (the first created
>> sink) to -1 (nowhere) in the last week.
>> 
>> I would guess that some source (probably attached to the node with
>> classifier _o31) doesn't have its sink specified.  Use gen-map to dump
>> the agents and which should be apparent.
>> 
>>    -John Heidemann
>
>
>John,
>when I use gen-map I have :
>
>% $ns gen-map
>Node _o10(id 0)
>                classifier__o11(Classifier/Addr)
>                address_0()
>                dmux__o26(Classifier/Addr)
>
>        Agents at node (possibly in order of creation):
>                _o24    Agent/TCP               portID: 0(0)
>
>        Link _o15, fromNode_ _o10(id 0) -> toNode_ _o12(id 1)
>        Components (in order) head first
>                _o30    ErrorModule
>                _o14    Queue/DropTail
>                _o16    DelayLink
>                _o18    TTLChecker
>---
>Node _o12(id 1)
>                classifier__o13(Classifier/Addr)
>                address_1()
>                dmux__o27(Classifier/Addr)
>
>        Agents at node (possibly in order of creation):
>                _o25    Agent/TCPSink           portID: 0(256)
>
>        Link _o20, fromNode_ _o12(id 1) -> toNode_ _o10(id 0)
>        Components (in order) head first
>                _o19    Queue/DropTail
>                _o21    DelayLink
>                _o23    TTLChecker
>---
>
>I have no classifier _o31. So I don't know what to do. 
>It seems to me that the problem comes from the loss module part but as 
>I don't really understand how loss module work I am not able to find the 
>problem (for instance I don't understand what is the meaning of the line
>:
>$em bind $errmodel 1
>in my script).

The next step would be to put a breakpoint in the classifier code
(Classifier::find) and see what it is and hopefully where it came
from.

   -John Heidemann