This is from the fivetran support case - Data mismatch between snowflake and databricks – Fivetran Support

Following an internal review by the engineering team, it has been determined that the system is capable of automatically handling these scenarios. To address the issue, we will transition to teleport-based sync and initiate a resync. This will address the issue without the need to failure or rescheduling the connector.
 
As this is expected behavior, we recommend implementing an alert mechanism to monitor data loading from the connector sync. This will enable you to proactively trigger a resync when necessary.
 
We understand you were anticipating a solution for this behavior from our end. However, this is an expected system behaviour.

 

Our downstream processes rely on the fivetran sync's success condition to run the next jobs. We need the sync to fail or reschedule is any such change happens similar to what happens on the schema changes.