RE: questions about accounting functioning

> -----Original Message-----
> From: freeradius-users-admin@lists.freeradius.org=20
> [mailto:freeradius-users-admin@lists.freeradius.org] On=20
> Behalf Of Lucas Aimaretto
> Sent: Friday, May 20, 2005 5:24 PM
> To: freeradius-users@lists.freeradius.org
> Subject: questions about accounting functioning
>=20
> Hi all,
>=20
> Have some questios about accounting.

what accounting are you thinking about? there're some possibilities..
I assume now mysql with the default setup..

>=20
> When Acct-Start arrives, in which cases will freeradius reject or drop
> that packet?
reject ? probably only if i.e. the client secret is wrong, the packet is =
invalid or FR cannot do accounting for some reason.

> When Acct-Stop arrives, in which cases will freeradius reject or drop
> that packet?
that=92s all in the sql.conf, as long as it's valid regarding attribs, =
secret etc. it will either update a record or create a new one.

I've posted something a while ago to ensure accurate accounting with 2 =
servers, failover, radrelay and one mysql-db..
I have for instance set AcctUniqueId in the db-table to "Unique" to =
prevent logging duplicates, when you receive a stop without having a =
start (which's decided with the update statement in =
accounting_stop_query or the insert in accounting_stop_query_alt), it's =
created with the values received in stop as the start was most likely =
missed somehow. same is true for updates..

acct to files is always done if the packet is valid and there's enough =
disk-space to log.

> If no start was found for the arriving stop ( which fields are taken
> into account for the matching? ), why shoul I insert a stop=20
accounting_stop_query =3D "UPDATE ${acct_table2} \
SET AcctStopTime =3D DATE_SUB('%S',INTERVAL =
(%{Acct-Delay-Time:-0}) SECOND), \
AcctSessionTime =3D '%{Acct-Session-Time}', \
AcctInputOctets =3D '%{Acct-Input-Octets} + =
(%{Acct-Input-Gigawords:-0} * POW(2,32))', \
AcctOutputOctets =3D '%{Acct-Output-Octets} + =
(%{Acct-Output-Gigawords:-0} * POW(2,32))', \
AcctTerminateCause =3D '%{Acct-Terminate-Cause}', \
AcctStopDelay =3D '%{Acct-Delay-Time}', \
ConnectInfo_stop =3D '%{Connect-Info:-%{X-Ascend-Data-Rate}}', =
\
AcctPreSessionTime =3D '%{X-Ascend-PreSession-Time}' \
WHERE AcctSessionId =3D '%{Acct-Session-Id}' \
AND UserName =3D '%{SQL-User-Name}' \
AND NASIPAddress =3D '%{NAS-IP-Address}'"

this means the last three attributes AcctSessionId, UserName, =
NASIPAddress are taken into account for matching..

> record ??? I
> don't understand this. I would drop the stop if no start was found ...

I wouldn't as I want accounting as accurate as possible. the start might =
have been missed for whatever reason, the stop-record contains all info =
you need and the session hapenened somehow if you receive a stop-record.

Michael


-
List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html