Lifecycle Checks

The Lifecycle Checks page schedules the two lifecycle maintenance checks so they run daily without an operator — the same operations offered as buttons on the Retrieve Statuses page:

• the integrity check — backfills the status events the Plateforme Agréée delivered but the regular poll missed;
• the ordering check — re-sequences lifecycle events recorded out of order.

Both repairs are idempotent and chronologically ordered: re-running a check over an already-clean lifecycle changes nothing, so a scheduled overlap is safe.

---

At a glance

---

How the page is used

Each row is one scheduled check — several can coexist (for example a nightly repair plus a wider weekly audit in report only mode). Add creates one; the per-row card opens the editor.

The scheduler runs in serve mode; changes take effect on the next scheduled run.

---

Check fields

Field	Description
Description	Free text shown in the list (e.g. Nightly reconciliation).
Enabled	When off, the check stays in the list but is skipped at run time.
Hour / Minute	Time of day the check runs, in the server timezone.
Integrity check (missing statuses)	When ticked, fetches every PA event of the window and backfills the ones missing from the lifecycle.
Compare PA events from the last N days	The integrity window (default: 1). The check re-fetches every PA event of the window; re-checking already-recorded events is a no-op, so a generous overlap is safe.
Ordering check (events out of order)	When ticked, re-sequences lifecycle events recorded out of order — internal platform stages recorded after a standard status, and standard statuses in the wrong numeric order.
Repair automatically	When ticked, the run applies the repairs; left off, it only reports what it would change, leaving the lifecycle untouched.

Report first, repair once confident

Start a new schedule in report only mode and read a few runs on Retrieve Statuses before enabling Repair automatically — the manual page shows exactly which events each mode would backfill or re-sequence.

---

Command line

The same two checks run from the CLI for a cron outside the web server:

# Integrity check — backfill missing PA statuses
java -jar nomaubl.jar -status-integrity <configFile> [--lookback N] [--since YYYY-MM-DD] [--apply] [--notify]

# Ordering check — re-sequence out-of-order events
java -jar nomaubl.jar -status-order-check <configFile> [--apply]

Flag	Effect
--lookback <N>	Integrity window in days (mutually exclusive with --since).
--since <YYYY-MM-DD>	Integrity window start date.
--apply	Apply the repairs; omit to report only.
--notify	Fire the notification rules on backfilled transitions (integrity only).

---

Tips & best practices

• One nightly repair, one wider audit. A 1-day repair every night keeps the lifecycle current; a weekly 7-day report only run catches anything the nightly window missed without changing data on its own.
• The repairs are the Retrieve Statuses buttons. Nothing here does more than the manual integrity and ordering checks — the page only puts them on a timer.
• Leave --notify off for wide catch-up runs. Backfilling a large window with notifications on can flood recipients with historical transitions; enable it only on the tight nightly window.