Re: RFC 2866 - Accounting ON / Accounting OFF packets
David Mitton wrote:
Accounting-On /-Off events do NOT occur for every session. They are, as the text says, indicators of the global state of the NAS Accounting.
Yes, and that's the point I was making, there can be no Acct-Session-Id because Accounting ON / Off events are not session specific... so the RFC contradicts itself.
Usually you should get an On event when the NAS powers up and RADIUS starts. If a NAS Administrator turns Off accounting, or makes a controlled shut down, then you may see an Off event. The purpose is to indicate the liveliness of your accounting stream, and since RADIUS is not a "stateful protocol" this is where scraping stateful session data from Accounting can get into trouble. To further muddy the water, there is no indication if the Accounting state toggles, that the session state on the NAS is lost or not.
They're confusing, and not that informative. I'm wondering whether to just not log them at all. Our APs appear to generate them after any configuration change made to a BSSID or Radio. Nothing else has ever generated them in sufficient quantities as to cause serious problems. Thanks, Arran -- Arran Cudbard-Bell (A.Cudbard-Bell@sussex.ac.uk) Authentication, Authorisation and Accounting Officer Infrastructure Services | ENG1 E1-1-08 University Of Sussex, Brighton EXT:01273 873900 | INT: 3900
Arran Cudbard-Bell wrote:
Yes, and that's the point I was making, there can be no Acct-Session-Id because Accounting ON / Off events are not session specific... so the RFC contradicts itself.
Yes, well... Most NASes use an Acct-Session-Id of all zeros for Accounting On/Off events.
Usually you should get an On event when the NAS powers up and RADIUS starts ... They're confusing, and not that informative.
Knowing that your NAS rebooted and kicked everyone off is useful. I'm wondering whether to
just not log them at all. Our APs appear to generate them after any configuration change made to a BSSID or Radio. Nothing else has ever generated them in sufficient quantities as to cause serious problems.
That doesn't sound like good practice. If there are still user sessions active on the AP over an Accounting Off/On sequence, then it's *very* bad practice. Alan DeKok.
participants (3)
-
Alan DeKok -
Arran Cudbard-Bell -
David Mitton