RE: [RadiusNT] Radius 3.0.184 , caching and concurrency

New Message Reply Date view Thread view Subject view Author view
Serkan SUBASI (ssubasi@turk.net)
Fri, 22 Dec 2000 00:54:16 +0200



From: "Serkan SUBASI" <ssubasi@turk.net>
Subject: RE: [RadiusNT] Radius 3.0.184 , caching and concurrency
Date: Fri, 22 Dec 2000 00:54:16 +0200
Message-ID: <HPEFKLMFNBAIKFMPCAKBEEKCCFAA.ssubasi@turk.net>

I will try to implement snmp concurrency checking , but at first I want to
implement snmp statistics , but I could not be succesfull yet

I installed this on NT machines, and they are working, however in windows
2k , we could not get the snmp statistics
earlier I posted a message about this condition

Serkan Subasi

> -----Original Message-----
> From: radiusnt-request@iea-software.com
[mailto:radiusnt-request@iea-software.com]On Behalf Of Denny Figuerres
> Sent: Thursday, December 21, 2000 8:22 PM
> To: radiusnt@iea-software.com
> Subject: RE: [RadiusNT] Radius 3.0.184 , caching and concurrency
>
>
> Um... for part of this check the advanced SNMP features.
>
> With SNMP active Radius will do an SNMP call to ask that Terminal if in
fact
> that user is on-line and if not let them login.
> Also I seem to recall that if you do that then the serverports table
will
> clean-up on it's own over time.
>
> Check the docs about the needed setup info.
>
>
> -----Original Message-----
> From: radiusnt-request@iea-software.com
> [mailto:radiusnt-request@iea-software.com]On Behalf Of Serkan SUBA?I
> Sent: Thursday, December 21, 2000 8:34 AM
> To: radiusnt@iea-software.com
> Subject: [RadiusNT] Radius 3.0.184 , caching and concurrency
>
> I think caching mechanism must be adjusted carefully when using
> concurrency control
> we are running radius 8.0.184 on a windows 2000 server SP1 (and post sp1
> fixes)
> and radius database is a MS SQL server 7.0 SP2 on a windows 2000 server
> SP1 (and post sp1 fixes)
>
> months we are having so much concurrency problems, we are trying to
> analyze the problem, so far we found that, there can be at least two
> reasons for this
>
> 1. acccounting records are late (one user disconnects from port x, and
> then trying to connect port y, and got an error 'concurrency limit')
>
> we build a trigger like this and see that, not every error has an
> serverports record (AcctStatusType = 1), some
>
> CREATE TRIGGER RadLogs_Insert ON dbo.RadLogs
> FOR INSERT
> AS
> INSERT into serverports_LOG (ServerID, Port, UserName, AccountID,
> AcctStatusType, CallDate, FramedAddress, CallerID, RLCallerID,
> AcctSessionID, ConnectInfo, NASPortDNIS, SNMPUser, RSPCheck, LastReset,
> IPAddress)
> SELECT sp.ServerID, sp.Port, sp.UserName, sp.AccountID,
sp.AcctStatusType,
> sp.CallDate, sp.FramedAddress, sp.CallerID, i.CallerID,
sp.AcctSessionID,
> sp.ConnectInfo, sp.NASPortDNIS, sp.SNMPUser, sp.RSPCheck, sp.LastReset,
> sp.IPAddress
> from serverports sp, inserted i
> where sp.username = i.username AND i.RadLogMsgID = 11
> AND sp.AcctStatusType <> 2
>
> 2. caching mechanism may lately send update serverports queries , or
> insert radlogs queries
> I set in cache tab, acccounting section (1000 items, 0 sec, 0 items)
>
> Do you have any idea , how can we solve this problem ?
>
> Serkan SUBASI
>
> ------------
>
> This is a user supported list. If you require assistance from IEA
Software's
> Support Engineers, please check out our Support resources at
> http://www.iea-software.com/support.
>
> For more information about this list (including removal) go to:
> http://www.iea-software.com/support/maillists/liststart



New Message Reply Date view Thread view Subject view Author view
This archive was generated on Thu Dec 21 2000 - 14:40:32 Pacific Standard Time