Re: Possible message multiplexing problem

[Love Hörnquist Åstrand <lha@kth.se>]

>> This is the weird part. send_to_kdc() opens a new socket for each
>> transaction, how is socket reuse possible ?
>
> I failed to mention initially that I am using modified 0.7.2. Did
> send_to_kdc always create a new socket?
>
> I'll send you the capture w/ a frame number. Maybe my analysis is  
> wrong
> and you can make more sense of it.

Yes, send_to_kdc have created sockets for each request for a long long  
time.

>> If you force tcp, does this happen ?
>
> This is very time sensitive. I had to reduce logging and run the  
> test ~10
> times to get it to occur. I highly doubt I could trigger the problem  
> w/
> SYNs and ACKs involved.

Ok, if you send me the dump privately I'll see if I can figure out  
what is going on.

Love