Re: route learning

[ToBo]

Hi,

> 29 juli 2022 kl. 22:58 skrev Martin Schanzenbach <mschanzenbach@posteo.de>:
> Excerpts from ToBo's message of 2022-07-29 14:23:48 +0200:
>> I set up a three nodes in a very simple configuration: A and B is behind 
>> NAT, so they both have to communicate with each other through C
>> 
>> A   ->   C   <-   B
>> 
>> All nodes are in friend only mode. All nodes have some knowledge of each 
>> other, at least you can see infos in "gnunet-peerinfo" or "gnunet-peerinfo 
>> -f" (what is the exact difference anyway?)
> Do you have any actual connections?
> Try with "gnunet-core".

I mainly used gnunet-transport to monitor that, but yes, gnunet-core
has shown the same.

>> Why does a gnunet-cadet don't find it's way from A to B ?
> This may be due to many reasons. Try compiling gnunet with
> "--enable-logging=verbose" and set the debug level for cadet to DEBUG.
> Then check logs.
> That would be my first approach (IF you have connections at all see
> above).

I recompiled it and tried again, this time it worked for whatever
reason. Sometimes the behavior of GNUnet is hard to understand.

How can I configure a minimal set of deamons to just have a network or session
layer with cadet? I think this will help in understanding the whole thing.


THX