(RADIATOR) Hacking around an upstream issue.

[Hugh Irvine]

Hello Martin -

Its a bit hard to see how Radiator could deal with this situation in any 
sensible fashion.

I think I would still be inclined to use a seperate Handler for Alives 
and use a special AddQuery to check the existence of a session before 
trying to update it - although this will still be a problem if you have 
missed a Start record.

I would also check how and why you are missing accounting records in the 
first place and fix whatever problem is causing this to happen 
(saturated links?).

Keep in mind that the "sanity" of the session database depends directly 
on the sanity of the accounting records that are used to maintain it. 
Clearly what you describe is not "sane".

regards

Hugh


On Wednesday, September 4, 2002, at 10:18 AM, Martin Edge wrote:

> Hi Hugh,
>
> The normal processing of Alives is fine. The fact we get Alives 
> ensures I
> can check the sanity of the SessionTable.
> Each time an Alive record comes in, it updates the SessionTable for each
> user's session
>
> The NAS's appear to be sending an Alive packet for a Session after we 
> have
> received the Stop packet for the _same_ Session.
>
> This is due to the first attempt to send the Alive packet failing, the 
> NAS
> waits 10 seconds for a retry. During this ten seconds, the user 
> disconnects,
> the NAS sends a stop record.
>
> The NAS then sends off the second attempt for the Alive packet.
>
> Consequently, the session is 'reopened' in my SessionTable, as the Alive
> packet triggers a delete/insert .. for all intensive purposes I see a 
> dead
> session, which was actually already closed off.
>
> Thanks
> Martin
>
> -----Original Message-----
> From: Hugh Irvine [mailto:hugh at open.com.au]
> Sent: Wednesday, September 04, 2002 9:50 AM
> To: Martin Edge
> Cc: Radiator
> Subject: Re: (RADIATOR) Hacking around an upstream issue.
>
>
>
> Hello Martin -
>
> What exactly is the problem?
>
> If you just want special processing for Alives, do something like this:
>
> <Handler Acct-Status-Type = Alive>
> 	....
> </Handler>
>
> regards
>
> Hugh
>
>
> On Wednesday, September 4, 2002, at 09:23 AM, Martin Edge wrote:
>
>> Hey Guys,
>>
>> Want to see if anyone has any ideas of how I should deal with this
>> situation.
>>
>> We are currently getting the following for the 'occasional' user 
>> session
>> from our Upstream RADIUS..
>>
>> Order	Amount	Type
>> -------------------------------------------------
>> 1	1 		Start
>> 2	Many 		Alive's (every 15 min)
>> 3	1		Stop (0 sec, Acct-Delay-Time)
>> 4	1		Alive (9 seconds afterwards, Acct-Delay-Time=10)
>>
>> We are told that what is happening, is the first attempt is made to
>> send the
>> first Alive packet. By coincendence, the user disconnects between these
>> retries, and the Stop packet is fired off. The Retry Alive packet is
>> sent
>> after the Stop packet for that session has arrived.
>>
>> Until they can come up with a network-fix for the problem (to prevent
>> Additional Packets for the Same Session to be sent before completely
>> failing
>> the current packet (until $x times tried), I'm going to have to
>> develop a
>> hack around to work out on what basis to ignore these extra Alives.
>>
>> I was thinking I have two options
>>
>> 1:- Make Radiator see whether the SessionID is still active for an 
>> Alive
>> packet (could be costly on DB work each instance)
>> 2:- Make Radiator store recent sessionIDs it has closed off in a DB or
>> DBM
>> file, and check incoming Alive packet -isn't- in the recently expired
>> list.
>>
>> Which would be the best? Where should I start? :)
>>
>> Thanks,
>> Martin
>>
>> ===
>> Archive at http://www.open.com.au/archives/radiator/
>> Announcements on radiator-announce at open.com.au
>> To unsubscribe, email 'majordomo at open.com.au' with
>> 'unsubscribe radiator' in the body of the message.
>>
>>
>
> --
> Radiator: the most portable, flexible and configurable RADIUS server
> anywhere. Available on *NIX, *BSD, Windows 95/98/2000, NT, MacOS X.
> -
> Nets: internetwork inventory and management - graphical, extensible,
> flexible with hardware, software, platform and database independence.
>
>
>

--
Radiator: the most portable, flexible and configurable RADIUS server
anywhere. Available on *NIX, *BSD, Windows 95/98/2000, NT, MacOS X.
-
Nets: internetwork inventory and management - graphical, extensible,
flexible with hardware, software, platform and database independence.

===
Archive at http://www.open.com.au/archives/radiator/
Announcements on radiator-announce at open.com.au
To unsubscribe, email 'majordomo at open.com.au' with
'unsubscribe radiator' in the body of the message.