Re: (PM) Fixing the port numbering scheme on the PM4?!?!

Robert Boyle (robert@garden.net)
Sun, 14 Feb 1999 00:19:36 -0500

At 04:56 PM 2/13/1999 -0800, Thomas Kinnen wrote:
>On Sat, 13 Feb 1999, Robert Boyle wrote:
>
>> is being used. I KNOW that we could potentially work around this by
>> using the session id and port number or ip address and port number and
>> changing our database design, but that is a kludge and completely
>> patchwork solution. Is a fix for this in the works? Has anyone else
>> found this to be a problem? Please let Lucent know too. Thanks!
>
>Note that even if the PM4 were to use a unique number for each port in the
>chassie that still would not completely remove the problem you are having.

Yes it would.

>It would only reduce the occurrences. RADIUS accounting is not time
>senestive and you could easily receive a start record after the stop. The
>only way to insure a unique key for a database is to use a combination of
>the user-Name, Acct-Session-Id, NAS-IP-Address, NAS-Port, and if the
>Session ID can not be assured to be unique (unable to roll over, be reset,
>etc) you need to apply some other supplementary id by way of possibly the
>Class attribute or some other way.
>
>Also if you are using triggers and they do not factor in the delay time
>you can also have a major number of problems.

We do calculate the exact call time based on (start time - Acct-Delay-Time)
At any given point in time, there can only be ONE call on a given port on a
given PortMaster. Unfortunately, because of the vague accounting records
sent by the PM4, I can't use its accounting information... :( All I need to
fix the problem is for Lucent to acknowledge a design flaw and recode the
port numbers. I would recommend not adding a "BladeNumber" attribute. :(
(Although I could potentially work around this by using PortNumber =
((BladeNumber-1)*96)+PortNumber. I think that to make legacy applications
not break, you should just recode the port numbers starting at 0 and ending
at the number of blades * 96. Any thoughts?

>Overall adding either an attribute for the blade number or making the
>numbers unique across the chassie would be nice and reduce the problem but
>it would not remove it.

It would in our case. Please let me know.

-Robert

Tellurian Networks (Formerly Garden Networks)
Quality Nationwide Internet Access since 1995
Access, Co-location, Development & Consulting
http://www.tellurian.net (973)300-9211 Ext.103
Earth first! We'll stripmine the other planets later.
"The penalty for censoring what your children are
taught is children who are no brighter than you."
Finger robert@tellurian.net for PGP public key

-
To unsubscribe, email 'majordomo@livingston.com' with
'unsubscribe portmaster-users' in the body of the message.
Searchable list archive: <URL:http://www.livingston.com/Tech/archive/>