Central job queue worker – run, scheduling, retries

Overview

The single background entry point called every minute: it manages the cron_queue table and calls the controller or library matching the job type. Results: admin cron list, System → Async jobs, system log (channel cron), notifications and optional e-mails. It checks no function permission (system context); where a row carries entitlement_feature_key, the subscription right of the user who created the row decides.

Run steps (masterdata/cronqueue.run)

1. If the module's cron secret is set, the secret parameter must match (hash_equals), otherwise 403. Called by the server crontab every minute, or from the framework's cron list through the cron/cronqueue entry point (which injects the secret from the settings).
2. Lock file (storage/logs/cronqueue.lock, 10 minutes): silent exit if an instance is running; heartbeat file; a shutdown handler removes the lock and logs even on a fatal error.
3. In maintenance mode only the update proceeds (dlbtupdate->cronApply(35)), then exit.
4. Before 7:00 on Monday rows older than 30 days are deleted; daily doc cache clean-up; hourly Drive path refresh; cron self-healing (dlbtcronhealth->healIfDue).
5. enqueueDueTemplates(): runnable rows from due schedule templates (for a company group, for every company of the group), _schedule_id in the payload, template advanced (Dlbtschedule::nextRun).
6. Expired share folders deactivated (expireShares); iPhone backup chain guard (getStuckChains); handleExpired(): expired pending rows deleted and rescheduled in catch-up mode.
7. Global seed (company_id = 0), if enabled and no row is pending: email, email_commands, email_ai, taskdigest, contract_passthrough_scan, otpkelfo_ingest, contir_general, fxrate_sync, avgrate_daily, integrity_check, recovery_fee, interest_doc, subscription_lifecycle, quick_invoice, quick_image, nav_tax_debt, mail_sweep, update_check.
8. NAV sync seed: if the module's NAV sync switch is not 0 (default on), a nav_invoice_sync chain at the next quarter-hour for every company with NAV credentials and digest_status=1; broken chains restored; hourly diagnostic log.
9. Recovery of stuck invoices (recoverOrphanPendingData): a new nav_invoice_data row for NAV invoices left without items.
10. Time-boxed processing (default 45 s, 5–55), at most 200 rows: first Main Job rows, then pipeline rows, finally classic rows.
11. Per row (processClaimedItem): entitlement condition (dlbtentitlement->allowsFeatureKey; without right skipped/entitlement and deletion); dispatch() or dispatchJob(); on success result saved, row deleted, result e-mail or log, next run scheduled, Main Job progress, optional completion e-mail. On error a friendly message, retry limit (permanent NAV error 1, transient network error 3, other 2), exponential back-off (30 s × 2^n, max 600 s), after the last attempt deletion and rescheduling; for NAV sync errors a notification to the company's authorised users.

Tables: cron_queue, job, notification, emailtask, company.

Self-diagnosis (masterdata/cronqueue.status)

From the browser, as plain text, read-only: last worker run from the heartbeat (over 180 s is an error) and lock state; NAV sync switch and interval; NAV companies (credentials, digest_status, seeded); the last 30 rows of nav_invoice_sync and nav_invoice_data; iPhone backup diagnostics; the crontab line to use. Protected by the cron secret (403).

Scheduling rules (scheduleNextRun)