Informations about home_server_pool with load-balancing or fail-over

Geaaru geaaru at gmail.com
Wed Jul 29 13:48:41 CEST 2015


Thank you very very much for support and informations.


Geaaru

On Mon, 2015-07-27 at 11:09 -0400, Alan DeKok wrote:
> On Jul 27, 2015, at 11:01 AM, Geaaru <geaaru at gmail.com> wrote:
> 
> > Link is homepage:
> 
>   Fixed, thanks.
> 
> > But why if all home_server are in status alive (and first failure 
> > on
> > receive packet) proxy send directly to second home_server without
> > re-transmissions?
> 
>   I don't know.  It depends on how the client retransmission works.
> 
>   The proxy isn't magic.  It doesn't invent packets.  It "fails over" 
> a packet ONLY when it receives a retransmission from the client, AND 
> it notices that the home server WAS alive, but is now marked "dead".
> 
> > RFC how describe fail-over operation ? Can you said me what are RFC 
> > that
> > describe these operations ? Thanks.
> 
>   There is no RFC.  There is only the FreeRADIUS code.  There is no 
> standard for RADIUS proxy fail-over.
> 
> > A client said me that 3GPP specification doesn't expect a response 
> > about a
> > Disconnect-Request if session doesn't exists or is already down. Do 
> > you
> > know if this is correct ?
> 
>   That's not true.  Read RFC 5176 for details.
> 
> > From freeradius side this means that when I receive a Disconnect
> > -Request to
> > an already closed session and proxy forward correctly packet to 
> > correct NAS
> > , proxy doesn't receive Disconnect-Nack or Disconnect-Ack and so 
> > NAS is
> > marked as zombie/dead.
> 
>   No.  The NAS will respond.
> 
>   Alan DeKok.
> 


More information about the Freeradius-Devel mailing list