Description

This article provides a definition for Probes

Content / Solution:

Probe is a health check used to track the state of a server.  Probes play an integral part in the health monitoring of servers that are used in the configuration of loadbalancing features.  Also referred to as out-of-band health monitoring, the loadbalancer verifies the server response or checks for any network problems that can prevent a client to reach a server. Based on the server response, the loadbalancer can place the server in or out of service, and can make reliable load-balancing decisions.

The loadbalancer identifies the health of a server in the following categories:

  • Passed—The server returns a valid response.
  • Failed—The server fails to provide a valid response to the loadbalancer and is unable to reach a server for a specified number of retries.

By configuring the loadbalancer for health monitoring, the loadbalancer sends active probes periodically to determine the server state.  You can associate the same probe with multiple real servers or server farms

Port Number Inheritance for Probes

If you choose not to specify (recommended) a port number for a probe, the loadbalancer can dynamically inherit the port number specified:

  • From the real server specified in a server farm.
  • From the port specified in the VIP.


This flexibility provides you with an ease of configuration. In this case, all you need is a single probe configuration, which will be sufficient to probe a real server on multiple ports or on all VIP ports. The same probe inherits all of the real server's ports or all of the VIP ports and creates probe instances for each port.

When you explicitly configure a default port through the probe command, the probes will always be sent to the default port. In this case, the probe will not dynamically inherit the port number from the real server specified in a server farm or from the VIP.

We currently provide all Networks with two predefined probes named TCP and UDP

The TCP probe is a transmission control protocol (TCP) probe.  A TCP probe makes a connection, and if the connection through a 3-way handshake (SYN, SYN-ACK, and ACK) is successful, when the loadbalancer receives FIN-ACK from the server, the server is marked as passed. By default, the loadbalancer terminates a TCP connection gracefully by sending a FIN to the server.

The UDP probe is a user datagram protocol (UDP) probe.  The UDP probe sends a UDP packet to a server and marks the server as failed if the server returns an ICMP Host or Port Unreachable message. If the loadbalancer does not receive any ICMP errors for the UDP request that was sent, the server is marked as passed.

You can create your own Probes and attach them to your Server Farms as described in How to Add a Probe on a Network in a MCP 1.0 Data Center Location, and then manage them as described in How to Manage a Probe on a Network in a MCP 1.0 Data Center Location

Recommended additional reading: What is a Real Server or Rserver