Configuring a PowerConnect Cloud Cluster
Overview
The PowerConnect Cloud agent can be run in a cluster. Hazelcast is used as the clustering technology. The agents in a cluster communicate and share configuration and scheduling information through a distributed shared memory grid. If an agent in the cluster becomes unavailable due to a failure or planned downtime, other agents in the cluster can assume the workload and continue to collect and process existing events and metrics.
Clustering the PowerConnect Cloud agent allows for:
High availability - if one of the PowerConnect Cloud agents fails, the others in the cluster can continue to collect and process events and metrics
Scalability - the PowerConnect Cloud agents in the cluster can share the workload to allow for high volume event and metric collection
Requirements
TCP port 5701 open between PowerConnect agent hosts
Configuration
To configure the PowerConnect Cloud agent as part of a cluster do the following:
Install the PowerConnect Cloud Agent as per the installation instructions
On each host where the PowerConnect Cloud agent is installed:
Set the environment variable POWERCONNECT_MEMBERS to a comma separated list of ip addresses where the PowerConnect Agents will/are running. For example if you have two PowerConnect Cloud agents running on hosts with ip addresses 192.168.1.20 and 192.168.1.21 you would set the POWERCONNECT_MEMBERS variable as follows:
export POWERCONNECT_MEMBERS=192.168.1.20,192.168.1.21
Restart the PowerConnect Cloud Agent on each host:
/opt/powerconnect/bin/powerconnect.sh restart
Login to the PowerConnect Cloud Agent web UI and you will see each member of the cluster in the Dashboard under the Cluster section e.g:
The first PowerConnect Cloud Agent to start (i.e. the oldest) is the primary cluster member. All members will sync their configuration from this host.
If the primary node goes down, another member in the cluster will take over the role