For AI agents: a documentation index is available at /llms.txt. Append .md to any page URL for markdown, or send Accept: text/markdown.
Pipeline Activity
Check the status of live and recent Statsig Warehouse Native pipeline runs in the Statsig Console, filter the job list, inspect a run's SQL and job graph, and cancel an active run.
Pipeline Activity reports the state of Statsig Warehouse Native pipeline runs: which runs are executing right now, which recent runs succeeded or failed, how long each one took, and the SQL each job sent to your warehouse. Open it at Data > Pipeline Activity in the Statsig Console when a reload looks stuck, a Pulse result looks stale, or you want to know what the pipeline has been running in your warehouse over the last two weeks.
Use Pipeline Activity for run status across every experiment and gate in the project. Use Pipeline Overview instead when you want to understand the pipeline itself: the stages it runs, the tables it writes into your warehouse, and the metric source and dashboard Statsig provisions for it. For the exposure and metric health of a single experiment, go to Experiment Diagnostics.
Pipeline Activity covers warehouse-native projects. Projects that don't run a Statsig Warehouse Native pipeline have nothing to report, so the page doesn't apply to them.
What the job list shows
The job list holds one row per pipeline run.
| Column | What it shows |
|---|---|
| Entity | The experiment or feature gate the run belongs to. |
| Run type | The kind of reload the run performed. |
| Run ID | The identifier for the run. |
| Status | The outcome of a finished run: Success, Failed, Partial failure, or Cancelled. |
| Jobs | The jobs that make up the run. The heading changes with the view you select. |
| Currently running jobs | The jobs executing at the moment. This column appears only when you narrow the list to active runs. |
| Started by | The person or process that triggered the run. |
| Turbo mode | Whether the run used Turbo. |
| Run duration | Elapsed time from run creation to completion, or the time so far for an active run. Sortable. |
| Days loaded | How many days of data the run covered. Sortable. |
| Schedule | The reload schedule attached to the entity. |
| Actions | Controls for the run, including Cancel. This column drops out of the completed-runs view. |
Run duration is wall-clock time from the moment Statsig created the run, so it includes any time the run spent waiting. It isn't the warehouse execution time for the run's queries.
Filter the job list
Four controls narrow the list:
- Run state: every run, or only the runs that are still active.
- Status: Success, Failed, Partial failure, or Cancelled.
- Entity: a single experiment or feature gate. This filter covers experiments and gates only.
- Started by: the person or process that triggered the run.
Changing any filter returns the list to the first page. The completed-runs list holds 25 runs per page.
Read the job runtime chart
A stacked bar chart sits at the top of the page and plots daily job runtime grouped by job type. The chart breaks out the largest job types and collects the rest into Other. It covers the last 14 days by default, accepts a range of up to 30 days, and excludes today.
Three things shape how the bars add up:
- Each bar sums the durations of the jobs that ran that day. Jobs inside a run execute in parallel, so a single day's bar can total more than 24 hours. Read a bar as total job effort, not as elapsed time.
- Statsig attributes a run to the day it created the run, in your browser's timezone. A run that starts before midnight and finishes after it counts entirely on its start day.
- Where a job has no recorded runtime, Statsig falls back to an estimate, so a bar can mix measured and estimated values.
The chart range and the job list filters work independently. Changing the range doesn't filter the table, and filtering the table doesn't redraw the chart.
Inspect a run
- Find the run in the job list.
- Click View details.
- Select a view in the dialog that opens:
- SQL: the queries the run sent to your warehouse.
- Details: the run's metadata.
- DAG: the job graph for the run.
- Timeline: when each job in the run started and finished.
The DAG and Timeline views appear only for runs that carry that data. From either one, click a job to jump straight to that job's SQL.
Cancel an active run
- Narrow the job list to active runs.
- Find the run you want to stop.
- In the Actions column, click Cancel.
The job list doesn't poll for status changes. Only the elapsed-time clock advances on its own, so refresh the page or change a filter to pull current status, including after you cancel a run.
Limitations
- A clean run history isn't a freshness guarantee. Pipeline Activity reports the runs Statsig dispatched. It can't show a scheduled reload that never got dispatched, so an empty failure list doesn't prove that results reflect your latest warehouse data. Confirm freshness on the entity's Pulse Results.
- Runtime isn't cost. Statsig doesn't read your warehouse billing, so no figure on this page translates into a dollar amount. Compare the runtime trend against your warehouse's own usage and billing views.
- The page doesn't refresh itself. Statuses update when you reload the page or change a filter.
Was this helpful?