Health Checks

Health Checks can be created for the cluster, to enable traffic to be re-routed in the event of an issue with a member of the cluster.

To configure health checks, click on the "Manage Health Checks" link from the Cluster's control panel, which will display a page similar to the one below:

The health check options available:

• Always up. Assume that the server is always available. Will still failover in the event of a datacentre wide outage.
• ICMP ping. Pings the server to make sure its up.
• TCP port. Checks that a TCP port can be opened.
• HTTP String. Performs a HTTP GET on a URL, and checks that the expected text appears in the response.

Please note that with the HTTP String option, the URL is re-written for each server, i.e. for my server fostoaa12, configuring the HTTP String option with URL ​​http://example.com/test would be re-written behind the scenes as ​​http://78.31.100.184/test with “Host: example.com” added as a header in the request.

If any of these health checks fail, the server will be considered down and will be taken out of the cluster until all health checks start passing again.

Health checks are completely separate to the checks carried out by Port Patrol, and as such, you will not be notified if a health check begins to fail. To get alerts for this, Port Patrol rules should be added to monitor these checks. It is advisable to monitor the status of the cluster as a whole using Port Patrol as well.