[Devel] Re: fix for Radius failed query logging

[Dan Pascu]

On Saturday 18 November 2006 02:08, Peter Nixon wrote:
> FreeRADIUS handles this fine as reverts to INSERT if it desn't find a
> record to UPDATE on Stop.

I think this is highly innefficient. Database writes are very expensive 
and this approach halves the request/second rate that freeradius can 
process. Unfortunately accounting is not something you can distribute to 
take advantage of multiple servers, so every bit of speed counts.

>
> Some vendors generate both Start and Stop with 0 session-time on a
> failed call. This is actually something that you could easily make
> configurable in any case. To give all 3 options (Old "Failed" method,
> "Stop-Only" and "Start-Stop") would at most be 15 or 20 lines of extra
> code...

I think we already know that we can modify openser to adapt it to the way 
freeradius works, but each of the proposed solutions makes less sense 
than using a Failed status for a failed call. This is why people around 
here (at least 2 mentioned this) keep patches to freeradius to add this 
behavior and never thought of modifying openser to achieve this goal, 
even though it would have been easier for them, because the alternatives 
make less sense.

I think the point of the original question was that if freeradius is 
willing to add this missing functionality, not how can we workaround the 
fact it's missing in freeradius. We already do that and I personally will 
continue to keep patches to freeradius because they make more sense, no 
matter what an old RFC (which wasn't updated for the needs of newly 
emerging applications) says. And the funny thing is the RFC even mentions 
this Failed status.

-- 
Dan