By default, when Data Collector encounters a stage-level error that might cause a standalone pipeline to fail, it retries the pipeline. That is, it waits a period of time, and then tries again to run the pipeline.

A stage-level error might include a stage not being able to connect to an external system because the system or network is down. 

You can define the maximum number of pipeline retries that Data Collector attempts.

The wait time between retries begins at 15 seconds and doubles in length until reaching a maximum of 5 minutes.