Documentation

Cluster Checks

About Cluster Checks

The Cluster Check allows you to group interdependent NodePing checks in order to get notifications and track availability of the group of checks as a whole, in addition to each component check individually. This might be a server or database cluster, or a group of web sites that share dependencies, a group of web sites that fail over to each other, or a group of HTTP and DNS checks for which it is helpful to know that both the sites and DNS are unavailable at the same time.

You specify a threshold of how many of those checks must be passing in order for the cluster to be considered 'up'. If the number of those checks passing drops below your configured threshold, the cluster check will fail and send the configured notifications.

Using Cluster Checks

To set up a Cluster check,

1. Create a NodePing check for each item in your cluster - either service specific or availability.
2. Select Cluster from the Check type drop down.
3. Give it a friendly label to identify this check in lists and notifications.
4. Set how often you want the check to run on the Check Frequency field. Every minute is recommended.
5. Choose a check from the drop down under "Checks:" to add it to the cluster check. Click 'Add another check' to include another check in the cluster. At least two checks are required.
6. Set the number of checks required to be passing for this cluster check to pass in the "Number of checks required to pass:" field. This must be a positive integer less than or equal to the number of checks in the cluster.
7. Set the Sensitivity. High is usually appropriate.
8. Set the notifications for this check. More information about notifications.

Common usage:

• To ensure your distributed, spanned, and/or redundant services have enough hosts available to provide the service.
• Monitor reverse proxy: Monitor the hosts behind a reverse proxy to ensure the service is available.
• Monitor Cassandra: Ensure enough nodes are still available for your Cassandra cluster.
• Monitor Redis HA: Ensure enough redis nodes and/or sentinels are available for failover.
• Monitor loadbalancer nodes: availability of hosts behind a load balancer.

Other considerations

Disabled checks show up in the "Checks:" drop down available for the Cluster check but their last status is ignored for determining the state of a Cluster check.