(RADIATOR) MySQL Database went away

Martin Wallner Martin.Wallner at eunet.co.at
Tue Mar 28 06:53:36 CST 2006


Mishair,
 
I fear you have to use two instances of RADIUS, one bound on the Accounting Port and one on the Authenticating Port. 
 
The reason, why you got stuck is, that accounting and authentication requests got stuck on the same database connection. Index (rebuild or build) tries to be as atomic as possible, so the requests start to queue up towards the database...
 
When you doing two instances of RADIUS, you have one connection from the authenticator process to the database (which is generally a nice read-only select), and a second one, where the accounting will be handled.
 
Additionally there are some more ways to time out the database and write overflowing Accounting requests (as plain log) to a local disk, which you can then integrate in the database by parsing the file (there is such a program in the goodies section)....
 
On a (personal) sidenote: AuthBy SQL-Authentication is good and fun and easy to setup, and it performs up to a distinct size of Users (which generate a given amount of Accounting Requests on the NAS), but it tends to loose performance quickly, when the userbase grows larger and reaches sometimes then the 'critical mass', where it starts to get REALLY SLOW  (and yes, with the right kind of machine running the DB it does perform, but you have to tweak it, pamper it and be very very careful with it...) ... IMHO Authentication via LDAP and Accounting on SQL (and then use 2 different threads for it) is the way to  get things done easy and 'as out of the box as possible'... 
 
=mw=
 


________________________________

	From: owner-radiator at open.com.au [mailto:owner-radiator at open.com.au] On Behalf Of Mishari Al-Faris
	Sent: Dienstag, 28. März 2006 12:52
	To: radiator at open.com.au
	Subject: (RADIATOR) MySQL Database went away
	
	
	Hello All,
	
	We have 2 Radiators talking to a 3rd server running MySQL to store Accounting and Authlog information. Today we saw that our Accounting table has grown pretty large and tried to INDEX the table, it had around 11 million records. This caused the MySQL server to get loaded, however the problem was after a little while both our Radius servers stopped processing Authentication Requests. When we were first trouble shooting the problem and tried restarting radiator it would process a few requests then stop again. Only when we killed the INDEX process on the MySQL server did the 2 radiators continue processing requests. 
	
	Can this behavior be corrected? please direct us in the right path. Because we wish that the Radiators do not stop processing requests if the Acct/Authlog DB becomes slow or unavailable.
	
	Many thanks and regards, 
	


--
Archive at http://www.open.com.au/archives/radiator/
Announcements on radiator-announce at open.com.au
To unsubscribe, email 'majordomo at open.com.au' with
'unsubscribe radiator' in the body of the message.


More information about the radiator mailing list