In a clustered installation, if an external resource isn't available, how does Sterling B2B Integrator deal with this?

When a process that is trying to access a particular resource fails and does not fail over to another node, manual intervention is necessary to either continue the process or wait for the external resource to become available and the process is restarted or resumed.

The exception to this is the database. Application clusters assume that your database environment is configured for high availability. An application instance can survive short database outages, but it is strongly recommended that you avoid this situation. At the very least, a database outage causes some business processes to fail because they are unable to persist their state.