Recent pipeline runs
Read from the stats JSON every run writes — cron runs and button-launched runs appear here identically.
| Period | Started | Duration | Cat rows | Fuel rows | Result |
|---|
Pipeline
The same script cron runs: scrape → refresh procedures → clear cache → verify → email. It takes the shared lock, so a run started while another is active exits without doing anything and is marked SKIPPED, not success.
Individual steps
Re-run one step without repeating the rest. The scraper
TRUNCATEs its destination table and reloads it, so a scrape on its own
leaves the dashboard tables untouched until the procedures are refreshed.
Failed RTOs only
Pulled from the last scrape's SCRAPE_FAILURE= markers.
Re-scrapes just these offices, in the mode selected above.
RTO coverage
v3 writes state_code/rto_code itself, so an
office is identified by code and survives Vahan renaming it. The baseline is the
previous month in the dashboard table — the raw tables are per-run staging and
hold one period only.
| State | RTO code | Regs last month |
|---|
Category freshness
The analytics portal carries a transport vehicle (goods, passenger, 3W-T) only once its permit and fitness records are complete, so the open month reads short on those and fills in over the following weeks. Absolute totals aren't comparable mid-month, but the mix is: a transport category whose share has collapsed hasn't matured yet. That is what the back-fill run recovers.
| Sub-category | This month | Share % | Prev share % | Ratio |
|---|
Tables in use
Read from .env. Every job this panel launches is handed the
same names, so the panel can never display one table while a job writes another.
| Role | Object |
|---|
Jobs
Jobs survive a restart of the panel; one interrupted by a restart is marked as such rather than reported as finished.
| Started | Action | Status | Duration |
|---|
Log browser
Pipeline stage logs, run logs and job logs.
Select a log on the left.