What happens in dbt when a single resource fails to build?

When a single resource fails to build in dbt, the process of building models is halted immediately, leading to the termination of any running connections. This behavior is by design, as dbt prioritizes data integrity and consistency. If one component encounters an error, it signifies that the overall process may not yield reliable results. Therefore, dbt does not continue with the execution of other models to prevent partial or inconsistent data states. This immediate exit ensures that developers are aware of the issue that needs to be resolved before any further actions or builds can be attempted, allowing for a clean slate for troubleshooting and resolution.

In contrast, the other responses do not align with dbt's operational principles. Notifications may exist for certain workflows but are not the default behavior of dbt on build failures. Continuing to run other models would compromise data integrity, and automated retries would not address the underlying issue that caused the failure in the first place.