Adovo

Data & Analytics

dbt v2 BigQuery jobs can ignore the old timeout key

Audit BigQuery profiles during dbt v2 migration so a legacy timeout_seconds key does not leave jobs unbounded.

November 17, 2026

The timeout line can remain in profiles.yml while the guard disappears. dbt issue #17151 reports that dbt v2.0.11 accepts the legacy BigQuery timeout_seconds key but does not enforce it; job_execution_timeout_seconds works in the same reproduction. The reporter's slow query ran to completion with the old key. This is a migration risk, not proof that every BigQuery job is unbounded.

Search effective profiles, including CI

Inventory BigQuery output blocks in local, CI and deployment profile templates. Find timeout_seconds and note which workloads depend on it for cost, queue or SLA control. dbt's BigQuery setup documentation names job_execution_timeout_seconds as the profile setting. Do not assume a parsed YAML key was consumed by the adapter.

Prove the timeout with a safe probe

On a non-production target, run a bounded long query with the current profile and check the BigQuery job's cancellation reason and duration. Repeat with the documented key. The issue's one-second test still took about 24 seconds end to end, so distinguish server-side timeout from job startup and cancellation latency. Record the dbt and adapter versions with the result.

Set a migration acceptance check

Replace the legacy key in reviewed configuration, then verify the effective profile used by every job. Add a configuration lint rule or startup assertion so an old key cannot silently reappear. For expensive models, retain a warehouse-side cost or resource guard as a second control.

Adovo’s view

A configuration file is a statement of intent; the running job is the proof. Every engine migration should test the operational limits the business relies on, especially those that fail silently when a familiar key changes meaning.

Continue exploring

Data & Analytics

dbtBigQuery

dbt Fusion can misread numeric UNPIVOT labels as strings

Diagnose a dbt static analysis error on valid BigQuery UNPIVOT SQL with integer labels.

Read more

Data & Analytics

dbtBigQuery

dbt Fusion rejects a valid BigQuery FORMAT call with six values

Distinguish a dbt static-analysis function limit from a BigQuery SQL error.

Read more

Data & Analytics

dbtOpenMetadata

A dbt model with no parents can keep old OpenMetadata lineage

Detect stale dbt lineage after a model loses its final upstream dependency despite overrideLineage being enabled.

Read more
Explore data and analytics services →