FreeRadius 3.0.15 - Help with accounting

[Aurélio de Souza Ribeiro Neto]

Dear ALL,

     Many thanks for help!!

     I found my problem.

     My checkrad was not working. I fixed and then my problem gone away.

Aurelio

Em 10/08/2017 10:01, Fajar A. Nugraha escreveu:
> On Thu, Aug 10, 2017 at 6:55 PM, Aurélio de Souza Ribeiro Neto
> <netolistas at mpc.com.br> wrote:
>>> Short version:
>>> - get a debug log where your NAS is actually sending the problematic
>>> accounting package
>>> - find out why the NAS does not response (i.e. the 'timeout' above)
>>> - take a look at your simultaneous use check, the problem might be there.
>>>
>>      Maybe an SNMP failure cause this for me.
>>
>>      No problem for me the zap If the snmp for simultaneous-use fails.
>>
>>      My question is: why the radacct registry is incomplete, without
>> nasporttype and callingstationid?
> As far as I can tell, those are related. You wrote:
>
> '
>      Sometimes I receive a complete Start request, but in my database
> (MySQL)  the data are inserted incomplete, without nasporttype and
> callingstationid.
>
>      When this occours I receive an automatic stop  and a new start
> packet and then all is ok.
> '
>
> When can you 'receive an automatic stop'? The only time I can think of
> is when session_zap does its thing. Which isn't really 'receive an
> automatic stop', but rather 'FR generates it'.
>
> When does session_zap happen? If it detects that the session is no longer alive.
>
> How does FR detect that the session is no longer alive? In your case,
> it might be SNMP to the NAS. IIRC if it times out, FR decide that the
> session is no longer alive.
>
> Why does session check happens in the first place? Because you
> configure simultaneous use check.
>
>
> So the key points should be:
> - Did you REALLY 'receive a complete Start request, but in my database
> (MySQL)  the data are inserted incomplete'? If so, a debug log when
> THAT happens should tell you why.
> - Did you REALLY 'receive an automatic stop'? It shouldn't be the case.
> - Why did you get timeout? My GUESS is if you fix that, it will solve
> this particular problem.
>
> Again, some of what I wrote are guesses. You should be able to
> determine the actual cause accurately if you can capture the correct
> debug log, or solve the timeout issue.
>