(RADIATOR) Redback is sending too many Access-Requests

[Toomas Kärner]

Hi,
Just out of my experience with messing around with redback and PPPoE I have
put some comments below, maybe you'll find them useful.

----- Original Message ----- 
From: "Mishari Al-Faris" <mishari26 at gmail.com>
To: <radiator at open.com.au>
Sent: Tuesday, December 20, 2005 7:28 AM
Subject: (RADIATOR) Redback is sending too many Access-Requests


Hello All,

Our Redback is sending us Access-Requests for DSL users way too frequently,
can reach up to 1-2 requests/sec.
#####
This is totally normal if since some routers configured with PPPoE have
"no_sence" in waiting in between the requests and just "DOS" the pppoe
server and auth_system behind it.
#####

Ofcourse the Timeout and Retry timers on Redback are set for reasonable
values like 1 retry after 30 second timeout.
All the requests coming to Radiator are being accepted, yet still sometimes
when too many users are that way our customers begin complaining about error
"718" (the computer you're dialing is not responding), in other words, the
requests are not being accepted fast enough.
#####

This may be a result of redback configuration that limits max sessions per
pvc also. As far as I you can get this error only if PPPoE server does not
reply to PPPoE discovery packets but I might be wrong here. If you have a
session limit per connection set in redback configuration and that limit is
full then redback will not reply any more to any PPPoE discoveries from that
connection resulting in that error.
#####

We saw that after locking and unlocking the DSL port coming to Redback's ATM
interface that the user is able to login once and the requests stop.
#####
This will kick out all sessions from that connection and then session limit
is not reached - users will get in with no problem until limit is reached.
#####

Trying to trouble shoot this, I was wondering does Radiator have a queue for
the requests its handling? and if its possible to check its status at a
point in time? Also, when I enable Trace 4 and see an access-request appear
in the log, does that mean Radiator only received it at that exact instant?
(being reasonably close again, not talking absolute exactness) or does it
only appear in the log after it had acknowledged the request from the
Redback?

I ask this because when we get "718" errors with users and I try to login
myself, and ask a friend to have an eye on the Radiator log to see when the
request concerning me arrives.

Also, we get these in the logfile:
Tue Dec 20 08:27:36 2005: ERR: There is no value named /rb01/11/0/27/675 for
attribute Calling-Station-Id. Using 0.
Tue Dec 20 08:27:36 2005: ERR: There is no value named /rb01/11/0/27/675 for
attribute Calling-Station-Id. Using 0.

Is there a way to stop this please?

Thanks so much for you patience in reading this,
Mishary Al-Faris

#####
If the problems that you have match exactly what I described above then the
solution that Hugh offered should not help you. Also I'm not too sure that
accepting all in case of flooding is good idea (but it was clever to add
Session-Timeout into reply in such case).
I hope I gave you some ideas what to test.
ps. session limit on connection is set with "bind" command "maximum"
attribute ...
#####



--
Archive at http://www.open.com.au/archives/radiator/
Announcements on radiator-announce at open.com.au
To unsubscribe, email 'majordomo at open.com.au' with
'unsubscribe radiator' in the body of the message.