Scheduled Jobs
The Scheduled jobs page lists every job the server runs on a schedule, in one table: the PA polls, the e-reporting cycle, the received-invoices fetch, the batch document jobs, the daily digests, the auto-retries, the lifecycle checks and the retention purge. For each one it shows the trigger, the state, the last run with its duration and result, and the next run; from the same row a job can be run now, paused, or opened in its editor.
Open it from Synchronization → Scheduled jobs. The Scheduler card of the Tech dashboard links to the page.
At a glance
The jobs
| Kind | Jobs | Trigger set in |
|---|---|---|
| Poll | PA import poll (pending invoices, status 9906), PA status retrieval (lifecycle events), E-reporting cycle (generation and submission of the due flows), Received invoices fetch (supplier invoices received on the platform). | The job's editor: interval in minutes, plus the quiet window for the two PA polls. |
| Batch | Batch document jobs: from the JDE print queue (BIP), from the input directory, or from the watched folders. | The job's editor: interval or daily time, and the batch parameters below. |
| Digest | Daily digests — e-mail of errors or archived invoices. | The digest form. |
| Auto-retry | Auto-retries — daily resend of invoices in an error status. | The auto-retry form. |
| Lifecycle check | Lifecycle checks — daily integrity and ordering check of the statuses. | The check form. |
| System | Retention purge — the daily sweep of the database stores (see Retention purge). | The job's editor; days per store. |
Disabled jobs are listed too, as Off, so they can be enabled from the page.
Columns
| Column | Description |
|---|---|
| Kind | Poll, Batch, Digest, Auto-retry, Lifecycle check, System. |
| Job | Name of the job; a click opens its editor. |
| Trigger | every 5 min, daily 04:00, daily · keep 90 days. |
| State | On, Off, Paused, Running. |
| Last run | When the last run started, with Duration and Result (the one-line outcome of the run). Kept across server restarts. |
| Next run | When the next scheduled run fires. |
| Actions | Run now, Pause / Resume, Run history. |
The toolbar filters by kind, refreshes the table (Refresh, Auto-refresh), creates a job (New job) and exports the grid. Its summary line counts the active jobs and shows the quiet window, highlighted while it is active.
Actions
- Run now — runs the job at once, outside its schedule, after a confirmation. The pause switch and the quiet window are ignored for this run; a job already running is not started twice. The run is recorded like a scheduled one.
- Pause / Resume — available on every job. A paused job keeps its schedule but its fires are skipped until it is resumed or the server restarts — a pause can therefore never be forgotten for good. Run now ignores the pause.
- Run history — opens the last runs of the job: when, event (start, end, error), duration, result and run id. Every run of a scheduled job — fired by the schedule, by Run now or by the matching command-line switch — writes a start row and an end or error row in the runtime processing log. The run id also appears in the server log lines of that run, so the two can be joined in the Server log. Failed runs show in Diagnostics → Failures.
The editor
A click on the job name opens its editor in a window. Save & apply saves the settings and reloads the schedules at once — a changed hour or interval no longer waits for a server restart; the pauses survive the reload. The digest, auto-retry and lifecycle-check editors also offer Send test and Delete.
Polls
| Job | Fields |
|---|---|
| PA import poll | Import poll interval (min) — minutes between import-status polls for pending invoices (status 9906); 0 = disabled. Quiet window from / to (HH:mm) — see below. |
| PA status retrieval | Status retrieval interval (min) — minutes between lifecycle status retrievals from the PA; 0 = disabled. Quiet window from / to (HH:mm). |
| E-reporting cycle | E-reporting cycle interval (min) — minutes between e-reporting cycles (generation and submission of the due flows); 0 = disabled. |
| Received invoices fetch | Received invoices interval (min) — minutes between fetches of the supplier invoices received on the platform; 0 = disabled. |
The quiet window is a daily time range during which the import poll and the status retrieval skip their runs — set it over a send campaign so per-invoice polls don't pile up while the PA integrates the batch. It may cross midnight; leave both bounds empty to disable it. The same two fields appear in Job defaults.
Batch jobs
| Field | Description |
|---|---|
| Label | Name shown in the list (e.g. BIP invoices). |
| Interval (min) | Minutes between runs. 0 = disabled, unless a daily time is set. |
| Daily at (HH:mm) | Fires once a day at this local time instead of the interval. Empty = use the interval. |
| Field | Values | Description |
|---|---|---|
| Process type | xml / ubl | xml = full pipeline (transform + process); ubl = direct UBL processing (input is already UBL). |
Template (when Process type = xml) | from list | Document to apply. Lists templates of type document. |
Mode (when Process type = xml) | AUTO / SINGLE / BURST / UBL | Processing mode — AUTO lets NomaUBL decide; SINGLE and BURST force the corresponding output strategy; UBL produces UBL only. |
| Source | bip / directory / watch-folders | Where to fetch documents from: bip = JDE Print Queue (JDE-specific); directory = scan the Input Directory of the global settings (a watched folder pointing at the input folder does the same); watch-folders = run the Watched folders pickup — the job then targets one configured directory (picked from a dropdown) or all of them. |
Extract mode (when Source = bip) | input / output / both | Which BIP artefacts to pull: input = source XML only; output = generated output files only; both = both. |
Language filter (when Source = bip) | text | Optional BIP language filter (e.g. FR). Empty = all languages. |
| Replace existing | Y / N | When Y, re-import documents that already exist in NomaUBL (overwrite); when N, skip duplicates. |
Skip PA send (when Process type = xml) | Y / N | When Y, the job never sends to the Plateforme Agréée — it only produces UBL/PDF locally. |
Validate only (when Process type = ubl) | Y / N | When Y, validate the UBL without sending it to the PA. |
Send to PA (when Process type = ubl) | (use settings) / Y / N | Override the per-document-type "send to PA" decision: empty = use default; Y = force send; N = force skip. |
Digests, auto-retries, lifecycle checks
The editor shows the full form of each kind, unchanged: see Daily digest, Auto-retry and Lifecycle checks.
Retention purge
The daily retention purge bounds the stores that nothing else bounds. It runs every 24 hours, the first time five minutes after the server starts.
| Field | Default | Store |
|---|---|---|
| Processing log (days) | 0 | Per-file START / END / error rows of the runtime log (F564237). Suggested 180. |
| Step timings (days) | 0 | Per-step timing rows written by debugProfile (F564237). Suggested 30. |
| Job runs (days) | 0 | Run rows of the scheduled jobs (F564237). Suggested 90. |
| Validation errors of finished invoices (days) | 0 | Errors (F564236) of invoices in a terminal status. Suggested 365. |
| Notifications (days) | 90 | Inbox notifications (F564253). Empty = 90 days; 0 = keep forever. |
Empty or 0 keeps a store forever. The purge is therefore off by default except for the notifications, which keep their 90 days: an upgrade deletes nothing by itself. Count only reports what each store would lose with the days saved on the server — save first, then count. The same values are in Job defaults → Retention; the command line offers java -jar nomaubl.jar -purge <configFile> [--dry-run].
New job
New job opens a window to pick the kind and a name:
| Kind | Creates |
|---|---|
| Batch | Document processing from BIP, a directory or the watched folders. The job starts disabled until its trigger is set. |
| Digest | Daily e-mail of errors or archived invoices. |
| Auto-retry | Daily resend of invoices in an error status. |
| Lifecycle check | Daily integrity and ordering check of the statuses. |
A digest, auto-retry or check name uses letters, digits, dash and underscore: it becomes the name of its configuration file. Create opens the new job in its editor.
Access
Grant Scheduled jobs in the Synchronization group on the Roles page. Run now, Pause and Save & apply require the integration-run action. The former Daily digest, Auto-retry and Lifecycle checks page grants are ignored.
Tips & best practices
- Read the State column before a campaign. Pause the status retrieval or set a quiet window, send, then resume.
- Use Run now to test a new batch job before giving it a trigger.
- Open the run history of a failing job before its editor: the result of each run often names the cause.
- Turn the retention on progressively — start with the suggested days, Count only, then save.