Product Documentation

Load Balancing Virtual Server and Service States

Apr 27, 2017

A load balancing virtual server that does not have a backup virtual server can take the following states, depending on the states of the service(s) bound to it and whether it is administratively disabled:

  • UP: At least one of the services bound to the virtual server is UP.
  • DOWN: All the services bound to the virtual server are DOWN, or the load balancing feature is not enabled.
  • Out of Service (OFS): If you administratively disable the virtual server, it enters the OFS state but its effective state is DOWN. Transitioning to the OFS state from the DOWN or UP state, or to the DOWN or UP state from the OFS state, is controlled by the administrator.

The state and effective state of a virtual server are the same if a backup virtual server is not configured. However, if a backup virtual server or a chain of backup virtual servers is configured, the effective state is derived from the states of the services that are bound to the primary virtual server and the backup virtual server(s). If any of the backup virtual servers in the chain is UP, the effective state of the primary virtual server is UP, even if all the services bound to the primary virtual server are DOWN.

The following diagrams show the conditions under which a virtual server transitions from one state to another.

localized image

A service can take the following states:

  • UP: If probes from all the monitors bound to the service are successful.
  • DOWN: If monitoring probes to the service are not answered within the configured time limit.
  • OUT OF SERVICE: If you administratively disable the service, or if you gracefully shut down the service and there are no active transactions to the service
  • GOING OUT OF SERVICE (TROFS): If you administratively disable the service with delay, or gracefully shut down the service and there are active transactions to the service. For more information, see Graceful Shut down of Services.
  • DOWN WHEN GOING OUT OF SERVICE (TROFS_DOWN): A monitoring probe fails while the service is in the GOING OUT OF SERVICE state.

A service in the process of transitioning from UP to OFS is in the GOING OUT OF SERVICE state. A service transitioning from DOWN to OFS is in the DOWN WHEN GOING OUT OF SERVICE state. For example, if a service is DOWN and you disable it with delay, the service transitions to DOWN WHEN GOING OUT OF SERVICE and then to the OUT OF SERVICE state. If a service is UP and you disable it with delay, the service transitions to GOING OUT OF SERVICE. During this time, if a monitoring probe to the server fails, the service transitions to DOWN WHEN GOING OUT OF SERVICE and, after the delay time expires, enters the OFS state.

Note

You can configure spillover to a backup virtual server by setting the "healthThreshold" parameter to a non-zero positive value. Then, if a single service bound to the primary virtual server transitions to the DOWN WHEN GOING OUT OF SERVICE state and the health threshold is not reached, the primary virtual server is marked DOWN and new connections are directed to the backup virtual server.

The following diagrams show the conditions under which a service transitions from one state to another.

localized image
localized image
localized image
localized image