We recommend the following approach for monitoring clusters:
- Install the up.time agent on each cluster node and add those agents to up.time.
- Add the shared cluster IP to up.time as a Virtual Node type.
To monitor cluster performance at the OS level, use the agent metrics provided by each cluster node. To monitor application performance across the cluster, add appropriate agent-based monitors to each cluster node (e.g. Exchange, SQL Server, etc.). Note that application monitors can not be added to the virtual node, they must be added to each cluster node individually.
To monitor overall availability of the cluster, we recommend adding agentless monitors (general up/down checks) to the cluster node. This will ensure that the request goes through the cluster and is passed to whichever cluster node is currently active. For example, to monitor a SQL Server cluster, the SQL Server (Basic Checks) monitor can be used to validate that the SQL server port is responding and that a basic query can be run against the DB.