Issue with determination of source vs. dest.

Carter Bullard carter at qosient.com
Fri Jun 13 12:31:29 EDT 2008


Hey Nick,
Yes, the assumption is that the Syn and SynAck packets are indicators
of the initiators of the connection.  If we see the Syn, then we  
should be
able to say that the source address in the Syn packet is the initiator  
of
the TCP connection (assuming that all is well).    if we did not see  
the Syn,
then we can infer who is the server and who is the client based on the
direction of the SynAck ( we should reverse the flow if we see only
a SynAck).   This helps us to know what may be going on in  
asymmetrically
routed paths.

Is this breaking something, or is it broken?

Carter

On Jun 13, 2008, at 12:13 PM, Nick Diel wrote:

> This might be related to my localized filtering bug, but I wanted to  
> put it a separate tread in case it wasn't.
>
> In the output from RA, it is switching the client/source and the  
> server/dst for TCP connections where Argus did see at least one  
> syn.  The tcp state ra reports also indicates the client and server  
> are switched.  This is happening for some unidirectional flows where  
> the end of the flow wasn't captured and also a few bidirectional  
> flows where the initial syn was not captured but the synack was.   
> Since RA dir column isn't a ?> I am assuming it is using syn and  
> synack states to determine client/server.
>
>  Sport   Dir  Dport        State
> 80        -> 3293              S
> 80        -> 4437              S
> 80        -> 4438              S
> 80        -> 2155              S
> 80        -> 53523            SE
> 80        -> 25639             S
> 80        -> 43676            SE
> 80        -> 59239             S
> 80        -> 1222              S
> 80        -> 1267             SE
>
>  Sport   Dir  Dport        State
> 80        -> 3293            SA_
> 80        -> 4437            SA_
> 80        -> 4438            SA_
> 80        -> 2155            SA_
> 80        -> 53523         SA_PA
> 80        -> 25639           SA_
> 80        -> 43676         SA_PA
> 80        -> 59239           SA_
> 80        -> 1222            SA_
> 80        -> 1267         SPA_PA
>
> Nick
>




More information about the argus mailing list