Re-run a Flow Manually

You might want to re-run a flow outside its monthly schedule to:

• Retry a flow that failed in the previous run
• Backfill historical data after an initial setup
• Test that a code change has fixed a problem

---

From the Prefect UI

1. Open the Prefect UI at http://localhost:4200 (SSH tunnel required - see Onboarding)
2. Click Deployments in the left sidebar
3. Find the flow you want to run (filter by tag: economy, society, environment)
4. Click the run button (the play icon) on the right
5. In the Run now dialog, leave the defaults and click Run

The run will appear under Flow Runs within a few seconds. Click on it to watch the live logs.

---

From the command line

Connect to the staging or production VM via SSH, then run the flow directly.

Run with default settings (latest data only)

uv run python -c "
from yhovi_pipeline.flows.economy.earnings import earnings_flow
earnings_flow()
"

Backfill a range of historical data

NOMIS flows accept a time parameter for historical ranges:

uv run python -c "
from yhovi_pipeline.flows.economy.employment_jobs import employment_jobs_flow
employment_jobs_flow(time='2004-12-2024-12')
"

time value	Meaning
"latest"	Most recent period only (default)
"2023-12"	A specific period
"2004-12-2024-12"	A full date range

---

Which flows support historical backfill?

Only NOMIS flows support the time parameter for date ranges. These are:

• employment_jobs_flow
• earnings_flow
• education_attainment_flow

Flows pulling from NHS Fingertips always fetch the full historical series automatically. DWP flows fetch the most recent 12 periods by default.