"Ming-Ching Tiew" <[EMAIL PROTECTED]> wrote: > The problem is, for whatever reason, if the SQL database > server dies and when the database server is restarted and come > back online, the radius server is still unable to continue > to log new accounting information to it, ie it seems the connection > context is obtained during radius server starts up, and it maintains > it throughout the entire life of the radius server and never atttempt > to re-new it when it becomes stale.
The code does retry. See sql.conf, "connect_failure_retry_delay". Set it to a smaller value, and see if that helps. > This will post a MAJOR problem especially when the radius server > goes across the network to log accounting information to a SQL > server, ie the external server and network can fail and it will > definitely fail at some point in time. That's a bad design. It means that the accounting packets will be lost when the external network goes down. You should instead log to the "detail" file, and run radsqlrelay, from the CVS head. It will read the detail file, and put the accounting information in SQL. That will allow the normal RADIUS server to run when the external network is down, and to still log accounting information. When the network comes back up, radsqlrelay will dump the information back into the database. Alan DeKok. - List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
