The radius-server accounting-stop-packet pending-limit optimize command allows the BRAS not to count this packet retransmission into the number of RADIUS accounting stop packet retransmissions if a BRAS fails to transmit a RADIUS accounting stop packet due to a seed application failure and retransmits the packet.
The undo radius-server accounting-stop-packet pending-limit optimize command restores the default configuration.
By default, if a BRAS fails to transmit a RADIUS accounting stop packet due to a seed application failure and retransmits the packet, the BRAS counts this packet retransmission into the number of RADIUS accounting stop packet retransmissions.
This command is supported only on the NetEngine 8000 F1A.
Usage Scenario
When a large number of users go offline, the BRAS sends many RADIUS accounting stop packets to the AAA server. If the AAA server cannot process so many RADIUS accounting stop packets at the same time, the BRAS retransmits the packets. When a large number of RADIUS accounting stop packets are accumulated on the BRAS, the number of RADIUS accounting stop packet retransmissions exceeds the maximum number because RADIUS seed resources are insufficient. As a result, some RADIUS accounting stop packets cannot be sent to the AAA server. When users' going-online status information is not deleted from the AAA server, the users fail to go online again.
To allow the BRAS not to count this packet retransmission into the number of RADIUS accounting stop packet retransmissions if a BRAS fails to transmit a RADIUS accounting stop packet due to a seed application failure and retransmits the packet, run the radius-server accounting-stop-packet pending-limit optimize command. In scenarios where a large number of users go offline, after the command is run, the BRAS successfully sends the users' RADIUS accounting stop packets and the users can go online again.Precautions
In VS mode, this command is supported only by the admin VS.
<HUAWEI> system-view [~HUAWEI] radius-server accounting-stop-packet pending-limit optimize