In the context of RSF-1, “Simple-HA” means a restricted mode of operation in which a cluster is configured as normal, but services can not be left in automatic mode, and so will not failover automatically. The machines in the cluster are still aware of each other, and services can be moved around the cluster manually, but will not do so automatically.
This is done by restricting the cluster to run in manual mode. To start a service it may still be put into automatic mode on a server, but as soon as that service is seen to be active anywhere in the cluster, it will revert to manual mode.
Services may be started by briefly putting them in automatic mode. If the service is not active elsewhere, it will be started, and in any case will revert to manual mode. Starting a service in this way retains the checking that RSF-1 does to prevent a service being started on more than a single node in the cluster.
This means that to move a running service to another server, it must first be stopped, and then be marked automatic on the machine that it is to be moved to.
In a normal RSF-1 HA cluster services will normally be in automatic mode on all machines, so that if any machine fails it’s services will be started elsewhere with no manual intervention.
A Simple-HA cluster also has the restriction that each server/service combination may only be put into automatic mode no more than 5 times per hour, on average.
Posted in: General