What is uptime, and how do you measure it in Metabase?

Uptime is the percentage of time a service was up over a window — the time-based reliability number that ends up in SLAs, status pages, and board decks. Its real value is as a budget: 99.9% is 43.8 downtime minutes a month to spend. Measure it in Metabase from synthetic check results synced from UptimeRobot, Better Stack, or Checkly.

What uptime measures

It measures the share of the window during which the service answered — and, read as a budget, how much downtime you have left before breaking a promise. The nines translate to fixed allowances: 99.5% is about 3.6 hours a month, 99.9% is 43.8 minutes, 99.99% is 4.4 minutes. That translation is what makes uptime the executive metric: a single number per service per month, comparable across quarters, with an explicit rule for whether announced maintenance windows count. Probes tell you the service was reachable; whether individual requests succeeded is availability's job.

What data does it need?

• A check_results table: monitor_name, checked_at, is_up, and an is_maintenance_window flag — one row per synthetic check at a constant interval.
• An incidents table with started_at / resolved_at and a planned_maintenance flag, for minute-accurate downtime per outage.
• Source: UptimeRobot, Better Stack, or Checkly check exports via Airbyte or the warehouse; Prometheus blackbox-exporter probe results work too.

SQL patterns

SELECT
  monitor_name,
  date_trunc('month', checked_at) AS month,
  ROUND(
    100.0 * COUNT(*) FILTER (WHERE is_up)
    / NULLIF(COUNT(*), 0), 3
  ) AS uptime_pct,
  COUNT(*) FILTER (WHERE NOT is_up) AS failed_checks
FROM check_results
WHERE NOT is_maintenance_window
GROUP BY 1, 2
ORDER BY 1, 2;

SELECT
  i.incident_key,
  i.service_name,
  i.severity,
  ROUND(
    EXTRACT(EPOCH FROM (i.resolved_at - i.started_at)) / 60
  ) AS downtime_minutes,
  COUNT(c.checked_at) FILTER (WHERE NOT c.is_up) AS failed_checks
FROM incidents i
LEFT JOIN check_results c
  ON c.monitor_name = i.service_name
  AND c.checked_at BETWEEN i.started_at AND i.resolved_at
WHERE i.resolved_at IS NOT NULL
  AND NOT i.planned_maintenance
  AND i.started_at >= CURRENT_DATE - INTERVAL '90 days'
GROUP BY 1, 2, 3, i.started_at, i.resolved_at
ORDER BY downtime_minutes DESC;

Pitfalls

Where this metric applies

• UptimeRobot + Metabase — monitor check results and downtime logs
• Better Stack + Metabase — uptime checks with incident context
• Checkly + Metabase — API and browser check results
• Prometheus + Metabase — blackbox-exporter probe successes per target

Related

Metrics

Dashboards

FAQ