Re: [Emerald] Work on SNMP Concurrency Checking

Dale E. Reed Jr. ( (no email) )
Sun, 16 Apr 2000 17:51:24 -0700

TIA wrote:
>
> Right you are, of course, since you built it.. ;^) Here are the results I
> see with radius -x31. It looks like it is supposed to work. It works for me,
> but that's like the taking your car to a mechanic thing, I have never had
> any trouble of any kind dialing in, authenticating, etc. It must be some
> real crappy Rocksmell chip modems or something. I would then ask what
> specifically the On-line table in Emerald is looking at so that I could use
> a SQL query to clear the "hung" connections instead of going into the
> On-line Table and hitting the Clear button one at a time. Is there a SQL
> query that will do this?

RadiusNT/Emerald look for entries in the ServerPorts tables with
AcctSessionID
of 1 to tell who is on-line. Just set the AcctSessionID from 1 to 0 to
clear the port in that table.

> SQL Statement: {CALL RadCheckOnlineSNMP('tvkoch')}
> Server:rtr1-clyde SNMPUser:.1.3.6.1.4.1.9.2.9.2.1.18.13 User:

What this shows is the OID and server we sent an SNMP get to. The
string after User: (here nothing) is what we got back. Therefore,
RadiusNT ignored the on-line entry and let the user on-line.

> Radius -x31 with Cisco 2511 & Concurrent user confirmed already on:
>
> radrecv: Request from host cefd3812 (rtr1-clyde) code=1, id=6, length=76
> NAS-Identifier = 206.253.56.18
> NAS-Port = 13
> NAS-Port-Type = Async
> User-Name = "bohunk"
> Password = "s\314U\314\002\255\011\274\225\275\245\357\220E\333P"
> User-Service = Framed-User
> Framed-Protocol = PPP
> Checking for duplicate logins.
> SQL Statement: {CALL RadCheckOnlineSNMP('bohunk')}
> Server:rtr1-clyde SNMPUser:.1.3.6.1.4.1.9.2.9.2.1.18.7 User:bohunk

Here we confirm that indeed bohunk is on-line. What I see happening is
that RadiusNT 3 had double-check enabled and therefore re-grabs the
user from the DB and tries again (this prevents stale cache entries from
incorrectly denying someone access). However, in this case,
double-check
shouldn't apply. I'll check into it.

> Radius -x31 with 3Com TCM & Concurrent user confirmed already on:
>
> radrecv: Request from host cefd3c14 (tcm-belleville) code=1, id=147,
> Checking for duplicate logins.
> SQL Statement: {CALL RadCheckOnlineSNMP('sps')}
> Server:tcm-belleville SNMPUser:.1.3.6.1.4.1.429.4.10.1.1.18.2282 User:sps
> Server:tcm-belleville SNMPUser:.1.3.6.1.4.1.429.4.10.1.1.18.2283 User:sps

What this says is that user sps was lists to be on-line twice, and
the SNMP check did indeed verify that. 

-- 

Dale E. Reed Jr.      Emerald and RadiusNT__________________________________________IEA Software, Inc.    www.iea-software.com

For more information about this list (including removal) go to:http://www.iea-software.com/support/maillists/liststart