Re: (ASCEND) Ascend, ~raddb/logfile error messages

[Lasse Andersson <lasse@netcraft.se>]

We are using dbm today already. 
Do you know where to find better explanations of the error messages we are 
getting ?.
That would make decisions to put in more servers easy.

Regards

Lasse


> It sounds to me that you need to either migrate to a better RADIUS server or 
> at least, start using the DBM based code.
> The regular version of the free RADIUS server uses a flat users file.
> On every single access-request, the server has to read the whole file trying 
> to find the user entry.
> Using the DBM variation (build it using 'make dbm'), the flat users
> file is converted into an indexed database, making the searches **much** 
> faster.
> The conversion is extremely simple:
> * recompile the source using 'make dbm'.
> * if you need radipad, compile it using 'make radipad'.
> * copy the executables to /etc/raddb or wherever is the appropriate place. 
> The executables to copy are: builddbm, radiusd.dbm and radipad (if you use 
> this feature).
> * from /etc/raddb, run 'builddbm' to build the database.
> * then, start the server: './radiusd.dbm'
> 
> The only catch is that you need to remember running builddbm everytime you 
> change the users file.
> 
> 
> >From: Lasse Andersson <lasse@netcraft.se>
> >To: ascend-users@max.bungi.com
> >Subject: (ASCEND) Ascend, ~raddb/logfile error messages
> >Date: Thu, 27 Jan 2000 16:57:11 +0100 (MET)
> >
> >Hi,
> >
> >We are using Ascend 1.16 and in the logfile (~raddb/logfile) fills up with
> >quite a lot of these messages from time to time :
> >
> >Thu Jan 27 09:52:32 2000: Warning: backlog of 103 exceeds 100 requests
> >Thu Jan 27 09:52:32 2000: Warning: backlog of 104 exceeds 100 requests
> >Thu Jan 27 09:52:32 2000: Warning: backlog of 105 exceeds 100 requests
> >
> >All authentication and accounting is done through proxy radius servers
> >driven by another organisation.
> >
> >The explanation in the source code that generates the message is this :
> >
> >/* Scan our list of pending requests (authentication, accounting, and
> >    address allocation).  Weed out requests that have died of old age.
> >    If the given AUTHREQ is already on our list, retransmit our reply
> >    if we have one, and discard the request as a duplicate if we don't
> >    yet have a reply.  If we spawned a child to handle the request,
> >    send it a SIGHUP every time the duplicate request arrives here,
> >    prompting the child to retransmit the reply.  If the request was a
> >    duplicate which we handled as above, return TRUE to notify our
> >    caller that no further action is required.  */
> >
> >* Is this due to overload in our radius server or should we increase 
> >backlog or
> >is this normal for 100000+ accounts or we in some kind of trouble?.
> >
> >
> >In the ~raddb/logfile we also get quite a lot of two other messages :
> >
> >Thu Jan 27 10:24:06 2000: ~radius/bin/radiusd:Couldn't open \
> >/etc/raddb/users for reading
> >
> >Thu Jan 27 10:24:06 2000: Authenticate: machine.net.net1645, \
> >id=68: Missing Radius Users File: joe.user@company.com
> >
> >* The users file exist and the user is a legitime user, but this messages
> >keep coming up?.
> >
> >
> >Would be great hear some comments on this.
> >
> >
> >Kindly
> >
> >Lasse Andersson
> >



++ Ascend Users Mailing List ++
To unsubscribe:	send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd:	<http://www.nealis.net/ascend/faq>