Runs¶

Runs lists every run of the project: scheduled job runs, catch-up runs, reruns, Run now, and manual runs started from the editor. Open it from the scheduler sidebar. Click a run's row to open it.

Runs list for the last 24 hours with status filters, a running run, failed and successful runs, and the rerun icon on job runs

Choose what to list¶

Time range. The buttons after Showing choose the period: 15m, 1h, 4h, 12h, 24h (default), 7d or 30d. A run is listed when it was queued or finished in that period, or hasn't finished yet. The timeline below always shows the same period.

Status. The checkboxes after Show choose which statuses are listed; All ticks or clears them all. The timeline follows the same filters.

Filter	Statuses	Shown by default
Scheduled	`scheduled`: each enabled job's next run, not started yet	yes
Running	`queued` and `running`	yes
Succeeded	`success`	yes
Failed	`failed`, `cancelled` and `interrupted`	yes
Skipped	`skipped`: slots a job didn't run for, and runs you skipped	no

dbt UI remembers your time range and filters in this browser. At most 1000 runs are listed; choose a shorter period if you need older ones in between.

Run history is kept for 30 days

Finished runs and their logs are deleted after 30 days by default. Your administrator can change this with DBT_UI__SCHEDULER_RETENTION_DAYS (see the configuration reference).

The runs table¶

Column	What it shows
#	The run number.
Started	When the run started (or was queued, if it is still waiting). Empty for scheduled runs.
Scheduled for	The time the run is for: its cron slot. Catch-up runs and reruns keep the original time; Run now uses the time it was started; manual runs from the editor have none. This is the time passed to dbt as dbt vars.
Job	The job name, or manual for runs started from the editor.
Command	The dbt command and selector.
Status	One of the statuses below. For skipped runs, hover the status to see why the run was skipped.
Duration	How long dbt ran. Runs still going count up.
Peak memory	The most memory the dbt process used. See Sizing memory and concurrency.
Rows	Rows affected, as reported by dbt (see Rows affected).
Triggered by	What started the run: Schedule, Catch-up, Run now, Rerun of #… or Manual, with the person's name.
Commit	The commit the run used.

Times are shown relative to now (5 minutes ago, yesterday, in 2 hours, tomorrow); hover one for the exact UTC time. Click any column header to sort by it; click again to reverse the order.

Statuses¶

Status	Meaning
`scheduled`	The job's next run. It starts at the time in Scheduled for.
`queued`	Waiting: for the job's previous run to finish, or for a free slot when the server is already running as many dbt runs as it allows.
`running`	dbt is running.
`success`	dbt finished without errors.
`failed`	dbt reported an error, the run hit the job's timeout, dbt was killed because the server ran out of memory, or the run couldn't start (e.g. fetching the code failed). The error is shown on the run's page.
`cancelled`	You clicked Cancel on the run's page.
`interrupted`	The backend was stopped or restarted while the run was in progress.
`skipped`	The run didn't run for its slot: the job was paused, the slot was missed while dbt UI was down and the job doesn't catch it up, the previous run was still going, or someone skipped it.

Open a run¶

Click a run's row to open it. The page shows:

Started (UTC), Scheduled for (UTC), Duration, Peak memory, Rows affected, Triggered by and Commit.
Results: how many nodes succeeded, failed and were skipped, e.g. 12 ok · 1 error · 0 skipped.
Failure notification: for failed job runs, which channel was notified and when, or the endpoint's error if sending failed. See Failure notifications.
The error message, if the run didn't succeed.
A table of node results: each model, test, seed or snapshot with its status, start time, execution time, rows affected and the dbt thread. Hover a node for dbt's message. The table is sortable.
The log of the run. While the run is going, it updates live and follows the end of the output; scroll up to stop following.

The timeline in the bottom panel switches to the nodes of this run. Click All runs to go back to the list.

A failed run's page: run details, the error, node results with the failed test, the dbt log and the nodes timeline

Rows affected¶

Rows affected is taken from what dbt-core reports for each node in its run results. Whether it is available depends on the database adapter: some report it only for certain operations (DuckDB, for example, only for seeds), others not at all. It is shown as — when not reported.

Cancel a run¶

On the page of a queued or running run, click Cancel. dbt is asked to stop and to cancel the queries it is running in your warehouse; if it doesn't stop within about 30 seconds, it is killed. The run is recorded as cancelled. A queued run is cancelled before it starts.

Rerun runs¶

You can rerun runs of existing jobs, including failed and skipped ones. Manual runs from the editor, runs of deleted jobs and scheduled runs can't be rerun.

Click the rerun icon (Rerun for this slot) at the end of a run's row, or Rerun on a run's page. To rerun several, tick their checkboxes and click Rerun in the bar above the table.
The confirmation lists what each rerun repeats: the command line, selector, target, full refresh, the slot it is for, the dbt vars and the original commit. For a single rerun it also shows the job's current timeout, in minutes.
Under Code, choose Same commit as the original run (default) or the job's current code source (the latest commit of the default branch, or its pinned commit).
Click Rerun.

Rerun confirmation for one run, listing command, slot, dbt vars, original commit and the choice of code

What to expect:

A rerun repeats the original run's command, selector, target, full refresh and dbt vars, so dbt processes the original slot again.
It uses the job's current credentials and timeout.
Reruns wait for any run of the same job in progress. Several reruns run in slot order, one after another per job.
A single rerun opens straight away; several stay in the list. Each is labelled Rerun of #… in Triggered by.

Backfill skipped slots

Tick Skipped in the filters, select the skipped runs (for example the slots while a job was paused) and rerun them. Each runs for its own slot, oldest first.

Skip runs¶

Skip this run records a run that hasn't finished as skipped. Its time is not caught up again, and you can still rerun it later.

A scheduled run is cancelled: the job skips that time, and its following scheduled time becomes its next run.
A queued run won't start.
A running run is stopped at once: dbt is killed without waiting for it to cancel its queries, so queued runs can start. Work it already did in your warehouse stays as it is.

To skip one run, click Skip this run on its page. To skip several, tick their checkboxes and click Skip … runs in the bar above the table. Finished runs can't be skipped. You are asked to confirm, and skipped runs disappear from the list unless Skipped is ticked.

Cancel or skip?

Both stop a run. Cancel gives dbt time to cancel its queries and records the run as cancelled. Skip this run kills dbt immediately and records it as skipped. Neither is caught up again, and neither sends a failure notification.

Select runs¶

Tick the checkbox of a row to select it; the checkbox in the header selects all runs listed for the period. The bar above the table shows how many are selected, with Rerun, Skip and Clear selection. A button is disabled when it doesn't apply to all of the selected runs; hover it to see why.