Robust proxy accounting

[Alan DeKok]

Chris Howley wrote:
> I used the example configuration and got the same result.

  Well... it worked the last time I tried it, hence the example.

  The goal is:

 a) have a "fallback" configuration in the pool that writes packets to
the "detail" file

 b) read from the detail file, and try to proxy the packets

  Part (b) will use the "fallback" mechanism if the home servers are
still down.  However, the detail file write will notice that the packet
shouldn't be written back to the detail file, and return "noop".

  At that point, the detail file reader will say "Hmm...it wasn't
processed properly, let me try again in a bit".

  At least, that's how it works on my systems.

  Alan DeKok.