Check Interval (seconds)

The number of seconds between health check attempts. This defines the length of the ‘health check cycle’ and applies to both GEO clusters and FQDNs. The valid range for this field is between 9 and 3600. The default value is 120.

Note: The interval value must be greater than the timeout value multiplied by the retry value (Interval > Timeout * Retry + 1). This is to ensure that the next health check does not start before the previous one completes.
Note: If the timeout or retry values are increased to a value that breaks this rule, the interval value is automatically increased.

Connection Timeout (seconds)

This is the allowed maximum wait time for a connection to be established and for the receipt of a reply to a health check. If no reply is received before the timeout expires, then the connection is re-attempted as long as the Retry Count has not been reached. If any HTTP response is received in response to the health check, then no retries are attempted in the current health check cycle. The valid range for this field is between 4 and 60. The default value is 20.

Retry Attempts

This specifies the number of times the health check will be re-attempted if the timeout above is reached before receiving a response. The default retry attempts is 2. The retry count does not apply if any response is received in response to the health check. If, for example, either a 200 or a 404 response is received, then no retries are attempted in the current cycle.

The maximum detection window for failed clusters of FQDNs is the Check Interval + (Connection Timeout * (Retry attempts + 1)). On average, the maximum time is half that.

The timeline diagram below illustrates what happens from the time a resource IP is added or enabled, to when it goes down and then comes back up again:

  1. When a resource IP is enabled/created, an ICMP request is sent by the LoadMaster to the resource IP. Assuming it responds, the resource is marked UP.
  2. After 120 seconds has elapsed (the default Check Interval), an ICMP request is sent to the resource IP. If 20 seconds (the default Connection Timeout) elapses and the IP fails to respond, the LoadMaster will send up to two additional requests (the default Retry Attempts) and wait for 20 seconds between each. If all three of these requests receive no response, then the resource is marked down, and the Check Interval timer is reset.
  3. After 120 seconds elapses, the LoadMaster attempts to send an ICMP request to the resource IP. If the resource has now come back up and responds before the Connection Timeout elapses, the LoadMaster marks it UP and resets the Check Interval timer.