Ive done some more testing on this, and whats happening is that a network
disconnection causes the TCP communications between Radiator and the SQL server
to time out (eventually). The time it takes to timeout depends on the TCP
implementation, but it can be a several minutes.
I am looking for a way to improve this, but its very much under the control of
the kernel, rather than Radiator.
I know thats not very helpful, but we are trying to find a solution.
On Mar 19, 12:05pm, [EMAIL PROTECTED] wrote:
> Subject: Re: (RADIATOR) What happens when the database server goes away?
> Have you gotten any answers on this? I've had similar problems.
> When my mssql is not reachable or is down for some reason,
> radiator just hangs there not going to the next authby, which
> happens to be <AuthBy TEST>. I have ContinueWhileIgnore set
> for the handler. The thought behind doing it this way is that we
> won't deny service to our customers when our database is down.
> We have our radiators running on an NT4 as well as RedHat linux
> 5.0. Both behave the same in the face of an mssql failure.
> I've asked Mike for help on this, but he wants the log trace 4
> output. I can't give him this since my main concern when our
> database is down is to get it going again, not figuring out why
> radiator won't go on to the next authby. I don't run trace4 for long
> periods of time because it slows down radiator and generates a lot
> of output in a very short amount of time.
> I've tested my handler setup on another linux 5.0 system with
> radiator on it. The test worked properly when I intentionally set the
> password to the database incorrect. It saw the failure and went on
> to authby test and did great, but if the database is just not there, it
> has problems with this.
> John Kicklighter
> Internet 2xtreme
> Date sent: Fri, 19 Mar 1999 10:24:39 +1300 (NZDT)
> From: Andrew Ruthven <[EMAIL PROTECTED]>
> To: [EMAIL PROTECTED]
> Subject: (RADIATOR) What happens when the database server goes
> > Hi guys,
> > I currently have Radiator setup using Auth::Emerald to authenticate
> > against a MS-SQL server. Radiator is running on a Linux box and I'm using
> > DBD::Sybase and the free Sybase drivers.
> > If the MS-SQL server goes away (say, for example, we unplug the machine
> > from the network) how long will it take before Radiator times out and goes
> > on to the next AuthBy?
> > We tested this, and after waiting a couple of minutes without it dropping
> > to the next AuthBy we plugged the server back into the network. Can
> > anybody recommend any timeouts we can tweek?
> > (I'm tempted to modify the code so that if the SQL Server times out
> > Radiator waits 10 minutes before trying it again...)
> > Cheers!
> > Andrew Ruthven
> > Actrix Networks Limited
> > At Work: [EMAIL PROTECTED]
> > At Home: [EMAIL PROTECTED]
> > ===
> > To unsubscribe, email '[EMAIL PROTECTED]' with
> > 'unsubscribe radiator' in the body of the message.
> To unsubscribe, email '[EMAIL PROTECTED]' with
> 'unsubscribe radiator' in the body of the message.
>-- End of excerpt from [EMAIL PROTECTED]
Mike McCauley [EMAIL PROTECTED]
Open System Consultants Pty. Ltd Unix, Perl, Motif, C++, WWW
24 Bateman St Hampton, VIC 3188 Australia Consulting and development
Phone, Fax: +61 3 9598-0985 http://www.open.com.au
Radiator: the most portable, flexible and configurable RADIUS server
anywhere. SQL, proxy, DBM, files, LDAP, NIS+, password, NT, Emerald,
Platypus, Freeside, external, etc etc on Unix, Win95/8, NT, Rhapsody
To unsubscribe, email '[EMAIL PROTECTED]' with
'unsubscribe radiator' in the body of the message.