NRPE in HA syncs the bind IP
When using a local interface IP as the bind IP in NPRE 3.1_4, once the primary node resumes master after a failover event, the package settings will sync to the secondary. In this case, the secondary will have the primary's local interface IP set for bind. Binding to the CARP VIP is undesirable because it prevents the user from monitoring both nodes independently.
Steps to reproduce:
1. Install NRPE on both primary and secondary in an HA pair
2. Set the Bind IP Address on each node to its own LAN IP address
3. Failover to the secondary and then failback
The secondary's Bind IP Address will sync to be the LAN interface IP of the primary
Updated by Pim Pish about 1 year ago
On top of the listening IP it might be a problem for the NRPE items being synced, too.
I have e.g. on the master a ping check to see if the secondary is reachable via ping.
Since the check is being synced the secondary pings itself, which is not really a sane check.