From the Slack discussion, it doesn't seem like the issues are related to faro (i.e., they are happening in other parts of the "validate_drp_gen3" processing)?

The long-log option controls the formatting of messages, not the messages themselves (it adds extra context). Is that what you mean? Or is this talking about enabling debug logging?

Somehow the pipeline execution in faro is triggering this extra-long runtime which is not apparent in other pipeline executions. But it does not appear to be the metric measurements that are causing the problem.

Agree that it does not look like faro is the root cause, but adding --long-log will help to find out where the problem is. We should also add debug logging to faro

This has been done. Please see PR for review.