Description

This article describes the issue of the switchover trap being sent with a non-master address, when the master-only address is used as the SNMP trap source address.

Symptoms

The configuration to use the master-only address as the SNMP trap source address is as follows:

root@SRX1400> show configuration groups
node0 {
interfaces {
fxp0 {
unit 0 {
family inet {
address 1.2.3.4 /24;
address 1.2.3.6 /24 {
master-only ;
 }
}
}
}
}
}
node1 {
interfaces {
fxp0 {
unit 0 {
family inet {
address 1.2.3.5 /24;
address 1.2.3.6 /24 {
  master-only ;
}
}
}
}
}
}
root@SRX1400> show configuration snmp
trap-options {
source-address 1.2.3.6 ;
}

With the above configuration, the primary node will use the master-only IP address as the source address to send any trap. However, the switchover trap, which is generated at the failover event, may be sent with the non-master only address;  1.2.3.4 or 1.2.3.5 as per the above example.

Pattern 1 :

First, the failover event occurs, after the cluster boots up:

alt

Pattern 2 :

The New primary node gets the role for the first time:

alt

Solution

  • The default address, which is caught by SNMPD, is set when the node takes the mastership.
  • Until the first failover event, the new primary node (ex-secondary node) has not got the mastership. So, the master-only address is not set to the default address.
  • At the second failover and after, SNMPD on the secondary node has already got the address. Then the trap will be sent properly.
  • However, if a node is rebooted for some reason, the node will not send the trap with the master-only address at the next failover event.

This is a limitation of the SNMP infrastructure on the chassis cluster. Currently, no workaround is available for this issue. This article will be updated, as and when a solution or workaround is obtained.