Uptime Monitoring

Also Known As

TL;DR

Explanation

Synthetic monitoring: external agents (Pingdom, UptimeRobot, Checkly, AWS CloudWatch Synthetics) send HTTP requests on a schedule and alert on downtime or degradation. Checks: simple HTTP ping, multi-step flows (login, checkout), API endpoint verification. Real User Monitoring (RUM): instruments actual user traffic — measures real-world performance, captures Core Web Vitals, detects regional issues. Both complement each other: synthetic catches outages quickly (1-minute intervals), RUM shows real user experience. PHP health endpoint: respond 200 only if DB, cache, and critical services are all healthy.

Common Misconception

Why It Matters

Common Mistakes

• Health endpoint that always returns 200 regardless of dependencies.
• Monitoring only from one geographic location — regional DNS or CDN issues go undetected.
• No alert escalation — single point of contact for 3am alerts causes burnout.
• Checking homepage only — deep application functionality (checkout, API) may be broken while the homepage serves from cache.

Code Examples

// Shallow health check — always 200:
// GET /health
echo json_encode(['status' => 'ok']); // Returns 200 even when DB is down
http_response_code(200);
// Uptime monitor: all green
// Reality: every user request is failing with 500

// Deep health check:
GET /health → 200 if all healthy, 503 if any dependency down:

$checks = [
    'database' => fn() => $pdo->query('SELECT 1'),
    'cache'    => fn() => $redis->ping(),
    'queue'    => fn() => $horizon->isRunning(),
];
$failed = [];
foreach ($checks as $name => $check) {
    try { $check(); } catch (\Throwable) { $failed[] = $name; }
}
$healthy = empty($failed);
http_response_code($healthy ? 200 : 503);
echo json_encode(['status' => $healthy ? 'ok' : 'degraded', 'failed' => $failed]);

References

• ↗ https://www.pagerduty.com/resources/learn/uptime-monitoring/

Tags

Related Terms