When a batch job instance running on a runtime fabric environment with two cluster replicas encounters an unexpected failure, the following sequence of events occurs:

Failure Detection: The runtime fabric environment detects the failure of the replica running the batch job.

Replica Replacement: A new replica is provisioned to replace the failed one.

Batch Job State Recovery: MuleSoft’s batch processing mechanism ensures that the state of the batch job is periodically saved (checkpointing). This allows the new replica to resume processing from the last saved state.

Resumption of Processing: The new replica takes over the batch job, continuing from where the previous one left off, thus processing the remaining 99,000 records without starting from scratch or causing duplicate processing.