While executing a flow Switch causes processes to be performed and jobs to be moved between processes according to the specifications in the flow definition. However there are situations where a process is no longer able to continue processing jobs due to some problem with the process itself. In those cases, Switch makes all jobs in front of the process wait "inline" in the flow until the problem is resolved (it makes no sense to move these jobs to the problem jobs folder since there is nothing wrong with the jobs).
Important examples of why a process may fail include:
There is a network communication problem so jobs can no longer be transferred.
A property for the process is set to an improper value (Switch attempts to detect invalid properties when a flow is activated but this can never be foolproof in all cases).
Switch offers a number of facilities for handling problem processes, from detecting their presence to re-enabling them once the problem has been resolved.