PRIMECLUSTER Installation and Administration Guide 4.4

11.3.1 Changing Time to Detect CF Heartbeat Timeout

If CF cannot perform a heartbeat for 10 minutes, it is determined that a heartbeat fails. In the environment where the system is overloaded and a heartbeat failure occurs often, tune the heartbeat time.

Note

If you set the heartbeat time long, it takes long to detect an error. Therefore, tune the heartbeat time carefully.

To tune the heartbeat time (10 seconds), perform the following procedure:

1. Add the following to the end of the "/etc/default/cluster.config" file on all the nodes configuring a cluster system. To restore the older file version, take a note of the contents before changing it.

   CLUSTER_TIMEOUT "second"
   Example: Changing it to 30 seconds
   CLUSTER_TIMEOUT "30"

2. To enable the setting value, you need to execute cfset -r at the same time on all the nodes configuring a cluster system.

   # cfset -r

3. Execute cfset -a to confirm the setting value.

   # cfset -a
   From cfset configuration in CF module:
   
   KEY: CFCP       VALUE: cfcp
   KEY: CFSH       VALUE: cfsh
   KEY: CLUSTER_TIMEOUT    VALUE: 30