Skip to content

tycoon fire / firehouse

Failing tests, as text. These commands read the same data the 3D city renders (the observability metadata DB for run history, falling back to dbt's own target/run_results.json), so you can check what is failing without opening the browser.

What counts as a fire

In the city, a failing test sets its building on fire and the firehouse sends one fire truck to it. These commands keep the same rule:

In the data In the city In the CLI
A test with status fail or error Its building is on fire; the firehouse sends one fire truck over the roads tycoon fire lists it
All current fires The firehouse on the civic strip tycoon firehouse counts the trucks

A truck on the street means a problem is awaiting response. It never means a fix is running.

A model that failed to build is not a fire, and the city doesn't set its building on fire. The commands don't stay quiet about it, though: when models failed, both print how many and point at tycoon data history, where the failed run and its error are listed. They never report a broken build as healthy.

tycoon fire

One row per failing test:

tycoon fire                 # failing tests now
tycoon fire --run 480db459  # failing tests in one run

Without --run, each test counts by its newest result across all captured runs, the way the city folds run history. A later dbt run that runs no tests leaves the last test results in place, so a failure stays listed until a run that includes the test passes it. A test or model that is renamed or removed from the project drops out of the list, as it has no building in the city.

--run shows a single run. It takes the run id tycoon data history prints for a dbt run, or a prefix of dbt's own invocation id, and the output names the run it shows (Run: dbt test (6110ef1d)). A prefix that matches more than one run is an error that lists the matches.

The command reads the metadata DB when the project has one (tycoon data transform run captures it) and falls back to dbt's own target/run_results.json, which holds only the most recent run. Tests are named the way dbt names them (not_null_stg_widgets_id), without the project prefix or the hash suffix dbt appends to a test's unique id.

tycoon firehouse

The number of fire trucks on duty: one per failing test, counted the same way as tycoon fire.

tycoon firehouse

There is no "unreachable buildings" stat: the planner guarantees every lot fronts the one connected street network, and the renderer independently re-checks by routing each vehicle over roads.

The contractor vans (one amber van per source past its freshness SLA) are not listed yet.