Recovery and Failure

Recovery for in-memory workflows works the same as transactional workflows for the most part; a few special cases are handled differently.

The necessary information is persisted to the cache, database, or filesystem for recovery. In-memory workflows can recover from the following failure conditions:

  • Node Crash 
If the application server on which the workflow is running crashes.
  • Subsystem Failure 
If one or more of the subsystems (database, filesystem, JMS, or cache) are not operational.

Based on the transaction scopes (see Transaction Scopes), the required information for a process to be newly created or restarted from the last persisted state is stored.