Scheduled checks

Before you start

• A paid ParityRail plan that includes scheduled checks — the Free plan is manual-only and cron skips its projects. See plans & quotas.
• The project has finished onboarding — only eligible, onboarded projects are scanned.
• A Vercel deployment with the cron jobs declared in vercel.json — this is repo configuration, not something you toggle in ParityRail.
• A CRON_SECRET environment variable set in the deployment before going live.

Turn it on

1. In your Vercel project settings, add an environment variable CRON_SECRET with a strong random value:

   CRON_SECRET=<strong-random-value>

2. Make sure vercel.json declares the three cron paths:

   /api/cron/reconcile     0 3 * * *
   /api/cron/recent        0 * * * *
   /api/cron/weekly-digest 0 8 * * 1

3. Deploy. Vercel Cron then calls each endpoint on schedule, sending:

   Authorization: Bearer $CRON_SECRET

4. Confirm the project is on a scheduled-checks plan and has completed onboarding, so it’s eligible.
5. For the weekly digest to actually send, enable at least one notification channel (Slack or email) on the project.

What each schedule does

Daily full check

Once a day, ParityRail checks every customer in each eligible project (/api/cron/reconcile). It’s the same check as a manual full scan — every customer is compared and every access incident refreshed.

Because it checks everyone, it’s metered against your scan quota and skipped once a workspace is over its limit.

Hourly quick check

Every hour, ParityRail re-checks only what plausibly changed — capped per run to stay fast — instead of scanning everyone (/api/cron/recent). It looks at:

• Customers whose Stripe state changed in roughly the last 90 minutes.
• Every customer that still has an open access incident, so unresolved drift keeps getting retried until it clears.

If there’s nothing recent and nothing open, the run finishes immediately without spending credits.

Weekly digest

Every Monday, ParityRail sends a 7-day recap to each alert-enabled project’s Slack and email channels (/api/cron/weekly-digest).

Think of it as your “monitoring is working” heartbeat. It lands even when everything’s healthy, so a quiet week still gets a reassuring all-clear. The digest covers:

• Open access incidents, broken down by severity, and the revenue at risk.
• Incidents cleared in the window (repaired vs. resolved) and the revenue protected.
• How many fulfillment checks ran.

A project only receives the digest if it’s on an alerts-enabled plan and has at least one enabled notification channel — with no channels there’s nothing to send.

Verify it worked

There’s no onboarding-UI signal that scheduled checks are “on.” Confirm indirectly, after the next scheduled window:

• New fulfillment-check runs appear in the project’s run history / incident inbox. The daily full sweep refreshes every customer; the hourly quick check only touches recently-changed customers plus any subject with an open incident, and finishes immediately if there’s nothing to do.
• The weekly digest lands in the project’s enabled Slack/email channels every Monday — even on a healthy week, as an all-clear heartbeat.

Troubleshooting