Reaction after a failure

: Click to enlarge the image

One prerequisite of a stable cluster function is a consistent software condition. This means all servers must have the same version of the application software.

After a failure there will be a number of options to choose from:

Register the failure.
Restart the application.
Switchover the application to a backup node.
Cascading-, N-way, or N+1 failover possible.
Manual switchback or fallback.

With LifeKeeper there is an option to configure an automatic switchback after a failover. This option brings the cluster back to the normal configuration after a renewal of the primary node. A useful feature of LifeKeeper is to have the option of an automatic switchover just before a system shut down. This feature could increase the availability of applications in the case of a triggered shut down (such as by an UPS) or an accidental shut down.