The number of traps that are processed and specified in trapd.conf is a factor in performance. The smaller the number of received traps processed as events, the higher the overall rate of processing.With trap receiver you can import subsets of traps into the Adapter Platform. Large numbers of traps may be sent by network devices, but typically only a small subset are important to network operators. For example, link up and down traps are sent frequently by devices but are not typically important to display to the network operator since the IP Availability Manager already supplies authentic, root-cause problems related to these traps.
To prevent crashes owing to trap storms and restricting the queue size from growing, users are recommended to do the following settings in the smarts/conf/trapd/trapd.conf file:
| Parameter |
Value |
Description |
|---|---|---|
| QUEUE_LIMIT_MEGS |
Default: 0 (No limit) Recommended value: 100 |
Maximum allowable size of the queue before the process of discarding begins for traps from 'chatty' sources. |
| QUEUE_LIMIT_SECONDS |
Default: 0 (No limit) Recommended value: 60 |
The amount of time the traps can remain in the queue before being discarded. |
Alternatively, you can use the dmctl utility (put command) to set the required filtering parameters as follows:
dmctl -s put SNMP_TrapManager::<trap manager
instance>::MegabytesInQueue 100
dmctl -s put SNMP_TrapManager::<trap manager
instance>::SecondsInQueue 60
