Production operations
Prepare diagnostics, health checks, error references, alert routing and a verified restore procedure.
Diagnose your environmentLink to section
Run the diagnostic from the installed starter before rollout:
pnpm ops:diagnoseIt checks core configuration, email, paired OAuth credentials, billing completeness and demo/seed isolation. Output contains check codes and remediation rather than secret values. Unconfigured optional billing leaves checkout unavailable.
Configure HTTPS, your own credentials and a migrated database. Disable development fixtures and keep demo deployments on a separate database. Diagnostics do not contact providers to certify their configuration.
Monitor and alertLink to section
Process liveness
Application readiness
Scheduled probe
Error references
Connect the signalsLink to section
Configure your platform's monitoring, an operator and an alert destination for repeated readiness failures and billing webhook failures. The starter supplies hooks; it does not provision a monitoring account or send alerts to a preset recipient.
Exercise an incident in stagingLink to section
Trigger a deliberate failure, verify that the chosen destination receives the alert and check recovery before relying on the integration.
Investigate and replayLink to section
Follow references alongside provider dashboards and failed webhook receipts. Fix the cause before resending an event from Stripe. Duplicate retries are reconciled; manually granting plans or deleting billing history is not a recovery procedure.
Rehearse restorationLink to section
Match the PostgreSQL toolsLink to section
Install pg_dump and pg_restore matching the server's major version.
The verification command uses disposable localhost QA databases.
Run the isolated drillLink to section
STARTER_PRO_TEST_DATABASE_URL=postgresql://postgres:postgres@127.0.0.1:5433/pycolors_workspace_restore pnpm ops:verify-restoreThe command creates two databases, migrates and seeds synthetic data, makes a backup and restores it in a transaction. It cannot target an existing application database.
Review the verification reportLink to section
The command compares every table's row count and fingerprint, retains the report and removes the databases it created.
Verify your production backups
Configure encrypted backups and retention with your operator. Restore an actual backup into a restricted isolated environment and record the recovery point, verification and access procedure. A synthetic drill does not establish the freshness or accessibility of a production backup.
Keep the previous application artifact and review schema/session compatibility
before rollback. The full docs/operations.md remains available in your source.