If a molecule node fails unexpectedly, any process executions currently in progress will not automatically continue running on a different node. The molecule provides a highly-available runtime however it does not automatically resume individual executions automatically. Some logic can be built into the process to handle those situations more gracefully.
For scenarios where failure is unexpected or unplanned, you can look into options for adding error handling to a process so that it can retry or recover afterwards. It may depend on the type of process, but some possible options are:
- Adding a Try/Catch shape to the process. You can add logic to the Catch branch to send a document somewhere to be picked up again by a future execution, or to alert a user.
- Configure a Retry Schedule to automatically re-process failed documents.
For expected node downtime and maintenance, see Shutting down an Atom or Molecule node gracefully.