Maintenance ended

Planned database maintenance — all Poppins services will be unavailable.

What: Postgres patch upgrade on the Supabase project (17.6.1.155 →
17.6.1.166), including a disk right-size. No schema or application changes.

Expected downtime: up to 60 minutes (Supabase's stated worst case; usually
much less).

Affected — everything, because all four apps share one database:
• ops.poppins.cleaning (CRM)
• team.poppins.cleaning (cleaner portal)
• account.poppins.cleaning (customer account)
• poppins.cleaning (marketing site + public quote/lead capture)

Expected alerts during the window: /api/health returns 503 on all four
monitors. This is the maintenance, not an incident.

Customer impact: the public quote form cannot capture leads while the
database is down — enquiries in this window are lost, not queued. No cleans
are scheduled in the next 48 hours and no billing runs in the window.

Scheduled deliberately in a quiet slot: no jobs in flight at start, and
clear of every cron — nightly push/sync 02:00/02:30 UTC, reminders 09:00,
dunning 10:00, Friday payout batch 08:00.

Rollback: none — Postgres cannot be downgraded after upgrading. Recovery
would be restore-from-backup (last nightly backup 01:46 UTC; PITR not
enabled).

Post-upgrade checks: health endpoints on all four apps, hand-written DB
types vs live schema, all seven pg_cron jobs scheduled and active, job
processor tick resuming, Xero connection refresh.