ETL Support

Three steps to land your customer's ETL feed cleanly. Walk them in order on a first pass; once you know the surface, jump anywhere via the numbered cards.

First time here? Walk the loop ↓

Show the 5-step checklist

Configure your ETL hook — Set etl_hook in your config.yaml (or skip — the bundled demo regenerates Sasquatch data when the hook is unset). The Refresh Data run-status banner tells you which one fired last.
Refresh Data — Click Refresh Data to run your hook, then the matview refresh. The coverage report shows every L2-declared primitive (rail / template / chain / metadata key) with the observed row count.
Triage the gaps — If anything missed, the Triage view groups gaps by kind. Each card carries the diagnosis + a deep link to the L2 editor's create-new form, with a one-click "Back to Triage" breadcrumb that survives the save.
Probe a single slice — Use Probe to investigate one specific entity — pick a rail name, set the date window (defaults to All time), and see the L2-declared contract next to the runtime rows. Faster than running the whole pipeline when you're iterating on one fixture.
Re-run + repeat — Edit the L2, click Refresh Data again, watch the coverage tally close. The loop tightens as you go — most operators hit clean coverage on the third pass.

Refresh Data

/etl/run

Execute the ETL pipeline (wipe → hook → matview refresh) and render a per-kind coverage tally so you can confirm every declared primitive landed at least one row.

Triage gaps

/etl/triage

Find + fix gaps — diff declared contracts against observed runtime; each gap renders a card with the diagnosis + a deep link to the relevant L2 editor page.

Probe & fix

/etl/probe

Investigate one L2 slice — pick a rail, template, or chain and see L2-declared column expectations side-by-side with the runtime rows that match.