Hello,
We are experiencing radius authentication packets retransmisions due to
slow response from NASes (and possibly SNMP packets drops in the
backbone).
The problem appears when Radiator makes a SNMP query to a NAS with
snmpget command, and no answer is received (or received too late). This
action blocks the next radius authentication packets that are arriving
at Radiator. They are queued and eventually dropped or retransmitted to
Radiator from a proxy-radius.
We think that the problem could be solved when we change the timeout
period (3 seconds) and the # or retries (2) that the snmpget command has
defined by default (when a SNMP query to a NAS fails, Radiator locks 6
seconds!!!). Changing that parameters could make another problem when
Radiator is checking a simultaneous-use of a user if snmpget fails to
respond "on-time" that users can connect twice or more. This is less
problematic that the first.
In radiator.cfg there aren't parameters to change this behavior of
snmpget, aren't it? so we think we need to modify the source code of
Radiator (SNMP.pm), the subs snmpget and snmpwalk to introduce -r
<retries> -t <timeout> to the command line that calls snmpget.
Do you think we can do this change? Do you recommend that?
Thanks in advance,
Anton
===
Archive at http://www.starport.net/~radiator/
Announcements on [EMAIL PROTECTED]
To unsubscribe, email '[EMAIL PROTECTED]' with
'unsubscribe radiator' in the body of the message.
