Changing the host failure detection time

Fast failure detection is crucial to a Db2 pureScale® instance because the faster a host failure is detected, the earlier the recovery for the failure can begin. That said, an aggressive setting for the host failure detection time is not suited for high-latency environments, so users might want to adjust the setting.

Before you begin

  • The Db2® instance and the shared file system cluster must be shut down on all hosts.
  • To perform this task you must be the Db2 cluster services administrator.

About this task

You can use the db2cluster command to adjust the fast failure detection time for your Db2 pureScale instance. The command specifies how long it takes to detect a host failure or network partition. To determine the current setting use the command: 
db2cluster -cm -list -HostFailureDetectionTime

Procedure

To change the host failure detection time, use the command: 
db2cluster -cm -set -option HostFailureDetectionTime -value value
If this command returns an error that indicates the cluster file system is still active, resubmit the command with the force option:
db2cluster -cm -set -option HostFailureDetectionTime -value value -force
Including the force option ensures that the shared file system cluster remains down for the duration of setting HostFailureDetectionTime. The cluster file system automatically restarts when the command completes.

Alternatively, you can stop the shared file system manually with the following command prior to executing the db2cluster command to set HostFailureDetectionTime without -force option :

db2cluster -cfs -stop -all

Results

Once the command completes, the new setting applies to all Db2 pureScale instances in the Db2 cluster services domain.