VM Connector supports both transient and persistent asynchronous queues, and their recovery behavior differs fundamentally. MuleSoft specifies that transient queues provide better performance but are not reliable when the runtime or host fails. Persistent queues are slower because their content must be persisted, but they provide recovery across failures.
For a standalone Mule runtime, persistent VM queues serialize and store queued message data on disk. Consequently, if the Mule runtime or underlying system stops before an in-flight queued message is consumed, the message can remain available and processing can resume after the runtime becomes operational again.
A transient queue stores its queued state in volatile runtime memory. A system failure destroys that state, so an unconsumed transient message cannot be assumed to survive restart.
Therefore, neither "all messages are lost" nor "all messages are recovered" accurately describes VM behavior; durability depends specifically on the queue type.
For workloads where asynchronous events must survive process or host failure in a supported standalone runtime architecture, a persistent VM queue is therefore the appropriate choice.
Reference topics: VM Connector; persistent queues; transient queues; asynchronous messaging; durability and recovery.
Official documentation: https://docs.mulesoft.com/vm-connector/latest/vm-reference
===============================================================
