Data 360: Understanding Intermittent Failures During Bulk Operations in Ingestion API Data Streams

In the context of Salesforce's Ingestion API, streaming jobs are often interrupted to facilitate bulk operations like large data uploads or deletions. Normally, these jobs are canceled without issue, but if the cancellation takes longer than expected, an error status might be returned, leading to a "Failed" indication in the Data Stream history.
Here's how it works:

• When a bulk job is initiated, ongoing streaming jobs are canceled. This cancellation allows the bulk job to run, complete, and then restart the streaming job
  
  
• If the cancellation process takes too long, the system logs an error, marking it as a "Cancelled Task." However, this doesn't indicate a true failure, as the streaming job will resume after the bulk job is finished
  
  
• Despite the error status, the bulk operation completes as expected, and the data stream will pick up from where it left off without losing any data
  
  

This behavior is expected and doesn't typically require manual intervention. However, it's important to note that Salesforce is exploring options to improve the UI so that intentionally canceled jobs (such as during bulk operations) are not misrepresented as errors.