Re: [Help-gnu-radius] Reasons for an Accounting-Response to not be logged "fully"?

[JP Hindin]

> > "Good" auth records look exactly the same as "Bad", from as much as the
> > Acct records can tell - they all look like the below:
> [...] 
> > As you can see all the things that the manual suggests are required for
> > Accounting is in this Auth record... So I'm failing to understand what is
> > "missing"...

On Fri, 21 Feb 2003, Sergey Poznyakoff wrote:
> Notice that authentication records have nothing to do with accounting.
> They only mark an attempt to authenticate, nothing more. What you should
> be looking for are *accounting* records, only they are written to rad[uw]tmp.

Yes... I'm quite aware. I clearly wasn't explaining why I was showing you
the Auth record:
In the Gnu-Radius documentation it specifies which fields MUST be in the
Accounting-Response for the Accounting to record it. Logically if a field
appears in an Auth-Request it will also appear in the respecting
Accounting-Response, therefore by showing you the Auth record I was
proving that the fields (User-Name, Acct-Session-Id, etc) were going to be
in the Acct record and that was _not_ why it was not being logged.

As for radius.c:

Feb 21 17:00:30: Main.debug: radiusd.c:1265:rad_spawn_child: adding AUTH 
request to the list. 1 requests held.
Feb 21 17:00:30: Main.debug: radiusd.c:1303:rad_spawn_child: started handler at 
pid 9614
Feb 21 17:00:30: Auth.notice: Login OK: [aasp]: CLID 6515569980 (from nas 
some.addr)
Feb 21 17:00:30: Main.debug: radiusd.c:1331:rad_child_cleanup: child 9614 
exited: 0

I see no ACCT request has been added to the list at all... So does this
mean the NAS didn't send the Accounting-Response at _all_?
(If this is the case things become even more perplexing, since the same
NAS must have done so at some point this morning for it to log the Acct
record like I mentioned in the previous mail)

JP