Automatic synchronization and retention

Dashloom can synchronize Cloudflare Workers, Cloudflare R2, Cloudflare Pages, Cloudflare Queues, Google, Stripe, Lemon Squeezy, Creem, Polar, Paddle, Supabase, GitHub, Vercel, mapped D1 business metrics, and versioned Custom REST metrics on a workspace schedule. Open Data sources → Automation, choose a provider and interval, and save the schedule.

The Worker cron checks due schedules every 15 minutes. The top-level execution is claimed by the exact Cron expression and Cloudflare scheduled time, so at-least-once duplicate delivery cannot run the same batch twice. A running claim that remains unfinished for 30 minutes can be reclaimed by a later retry. Each due schedule is also claimed with a compare-and-update before the provider call. Successful runs record freshness; failed runs keep the error and retry with exponential backoff capped at six hours. After synchronization and alerts, the same sequential task repairs missing completed-action baselines and refreshes Agent outcome measurements from later-dated evidence.

Hosted operators can inspect the latest 50 scheduled and authenticated manual maintenance executions at /dashboard/operator. The ledger shows the trigger, task-level outcome, and a stable error code; it never stores provider error messages, credentials, or task payloads. Manual endpoints for reports, synchronization, alerts, retention, and billing reconciliation write to the same ledger.

Most history-capable synchronizations refresh the preceding 60 days plus the current day so a monthly Agent report can compare two complete 30-day periods. Cloudflare Queues is deliberately different: its REST endpoint exposes an approximate realtime snapshot, so Dashloom keeps the latest snapshot per day and never presents it as complete message-operation history. Cloudflare Workers Analytics is queried in provider-compliant chunks of at most 30 inclusive days; other history-capable APIs receive an equivalent bounded start date or are filtered to that window. R2 Analytics is limited by Cloudflare to 31 retained days and is always marked as partial Agent evidence. Supabase usage history is also determined by its Management API. Provider pagination caps still apply. High-volume GitHub, Vercel, Cloudflare Pages, Creem, Polar, or Paddle responses create a partial synchronization record when the bounded scan is exhausted; revenue connectors also propagate partial coverage into Agent evidence.

The server enforces each plan's minimum interval: Free daily, Solo hourly, Studio and Agency every 15 minutes. Community self-hosting may configure 15-minute schedules using its own infrastructure.

Workspace owners configure metric, analysis, and report retention under Settings → Data lifecycle. Daily maintenance deletes records older than the selected period and clamps retention to the current plan maximum. Agent outcome cycles follow analysis retention. Billing, subscription, membership, and audit records are not removed by this cleanup job.

For production, keep Cloudflare D1 Time Travel available for operational recovery and separately test workspace JSON export and deletion. Retention cleanup is not a substitute for a disaster-recovery backup.