Configuring Load Balancers for Horizon Connection Server Health Monitoring

To monitor load balancing health on Horizon Connection Server, follow these best practices.

To avoid flooding Connection Server with a large number of health check requests, set the polling interval to 30 seconds, with a timeout of two or three times that period. Try to have no more than two load balancers sending probes to one Connection Server instance.

The only supported health check is a fetch of favicon.ico. To minimize the cost of the probe, use the HEAD method where possible. Whether or not the check is successful, you must drop the connection after the fetch by either adding a Connection: close header to the request or else using an HTTP/1.0 request. An example of using the HEAD method in the send string of the load balancer:

HEAD /favicon.ico HTTP/1.1\r\nHost: \r\nConnection: Close\r\n

The HTTP status is usually 200. If the Connection Server has been administratively disabled (see Disable or Enable Horizon Connection Server), the status will be 503.

For timeout settings and load balancer persistence values, see the KB article on Timeout Settings and Load Balancer Persistence.