[RadiusNT Digest]

radiusnt-digest-request@iea-software.com
Thu, 6 Aug 1998 00:01:43 -0700

Message 1: Re: Dr. Watson!!!
from "Reid Sutherland" <reid@isys.ca>

Message 2: Re: Dr. Watson!!!
from "Dale E. Reed Jr." <daler@iea-software.com>

Message 3: Re: Dr. Watson!!!
from "Reid Sutherland" <reid@isys.ca>

Message 4: ODBC errors
from Ted Olson <tolson@ocsnet.net>

Message 5: Re: Access performance
from Ted Olson <tolson@ocsnet.net>

..------ ------ ------ ------ ------ ------ ------ ------ ------ ------.
| Message 1 |
'------ ------ ------ ------ ------ ------ ------ ------ ------ ------'
Subject: Re: Dr. Watson!!!
From: "Reid Sutherland" <reid@isys.ca>
Date: Wed, 5 Aug 1998 09:53:58 -0400

Question is, why would Radius crash to begin with on a long username? Or
does it for that matter? I found when I ran text mode it would puke all the
time, but when I went straight ODBC it worked fine for awhile, then puke. I
turned on Trim Name, seems ok now :/
Either way something is wrong with 2.5, because 2.3 worked fine.

Thanks.

Reid Sutherland
Network Administrator
ISYS Technology Inc.
http://www.isys.ca
Fingerprint: 6552 6469 5320 7475 6568 6c72 6e61 0a64

-----Original Message-----
From: Dale E. Reed Jr. <daler@iea-software.com>
To: radiusnt@iea-software.com <radiusnt@iea-software.com>
Date: Tuesday, August 04, 1998 4:37 PM
Subject: Re: Dr. Watson!!!

>Fox, Thomas wrote:
>>
>> I disagree. The *problem* is that Radius isn't
>> handling it properly. This same thing happens if
>> a user's modem sends a long string of garbage
>> through. It is a bug, imho.
>
>RadiusNT 2.5 correctly handles long user names like this. If
>you enable trim name, RadiusNT *WILL* correctly trim the name,
>just like he said. That *IS* the proper way to handle it.
>RadiusnT 2.5 will alsi discard username longer than 128
>characters and the newest beta has the ability to discard names
>based on their inclusion of a set of characters you can choose.
>
>--
>Dale E. Reed Jr. (daler@iea-software.com)
>_________________________________________________________________
> IEA Software, Inc. | RadiusNT, Emerald, and NT FAQs
> Internet Solutions for Today | http://www.iea-software.com

..------ ------ ------ ------ ------ ------ ------ ------ ------ ------.
| Message 2 |
'------ ------ ------ ------ ------ ------ ------ ------ ------ ------'
Subject: Re: Dr. Watson!!!
From: "Dale E. Reed Jr." <daler@iea-software.com>
Date: Wed, 05 Aug 1998 09:19:04 -0700

Reid Sutherland wrote:
>
> Question is, why would Radius crash to begin with on a long username? Or
> does it for that matter? I found when I ran text mode it would puke all the
> time, but when I went straight ODBC it worked fine for awhile, then puke. I
> turned on Trim Name, seems ok now :/
> Either way something is wrong with 2.5, because 2.3 worked fine.

There was no 2.3 and 2.2 had the same problem. Its a buffer overrun
that causes the problem. The latest 2.5 will reject names over 128
characters even if trim name is enabled or not. That prevents the
buffer overrun and corrects the problem.

-- Dale E. Reed Jr.  (daler@iea-software.com)_________________________________________________________________       IEA Software, Inc.      |  RadiusNT, Emerald, and NT FAQs Internet Solutions for Today  |   http://www.iea-software.com

..------ ------ ------ ------ ------ ------ ------ ------ ------ ------.| Message 3 |'------ ------ ------ ------ ------ ------ ------ ------ ------ ------'Subject: Re: Dr. Watson!!!From: "Reid Sutherland" <reid@isys.ca>Date: Wed, 5 Aug 1998 13:25:42 -0400

Well then any ideas on why it's still crashing?

Reid SutherlandNetwork AdministratorISYS Technology Inc.http://www.isys.caFingerprint: 6552 6469 5320 7475 6568 6c72 6e61 0a64

-----Original Message-----From: Dale E. Reed Jr. <daler@iea-software.com>To: radiusnt@iea-software.com <radiusnt@iea-software.com>Date: Wednesday, August 05, 1998 12:20 PMSubject: Re: Dr. Watson!!!

>Reid Sutherland wrote:>>>> Question is, why would Radius crash to begin with on a long username? Or>> does it for that matter? I found when I ran text mode it would puke allthe>> time, but when I went straight ODBC it worked fine for awhile, then puke.I>> turned on Trim Name, seems ok now :/>> Either way something is wrong with 2.5, because 2.3 worked fine.>>There was no 2.3 and 2.2 had the same problem. Its a buffer overrun>that causes the problem. The latest 2.5 will reject names over 128>characters even if trim name is enabled or not. That prevents the>buffer overrun and corrects the problem.>>-->Dale E. Reed Jr. (daler@iea-software.com)>_________________________________________________________________> IEA Software, Inc. | RadiusNT, Emerald, and NT FAQs> Internet Solutions for Today | http://www.iea-software.com

..------ ------ ------ ------ ------ ------ ------ ------ ------ ------.| Message 4 |'------ ------ ------ ------ ------ ------ ------ ------ ------ ------'Subject: ODBC errorsFrom: Ted Olson <tolson@ocsnet.net>Date: Wed, 05 Aug 1998 10:45:21 -0700

--=====================_902364321==_Content-Type: text/plain; charset="us-ascii"

We upgraded our login server early this morning, dual Pro 200, NT4/SP3,128MB, RadiusNT 2.5.124 running as a service, ODBC mode. All had beenpre-installed and tested for about a week - this morning we just stoppedRadius, copied over the current database (Access97), swapped IP addresses,and brought up the new machine.

Seemed to go very smoothly - for a couple hours at least. Since then, we'vehad 3-4 distinct problems where connections are fine for a while and thenstart being denied with the error below. The bandaid fix is to stop andrestart RadiusNT, then all is OK again (until the next time).

Any idea what might be causing this? It wasn't happening on the oldmachine, and it's the exact same database as before, no changes to tablestructures or indexes. Is there a way to determine which table/index isbeing referred to? (It just happened again...) Here's a longshot, but isthe dual CPU config an issue at all?