Monitoring i logirovanie veb-prilozheniy

Observability — sposobnost ponimat sostoyanie sistemy cherez metriki, logi i treysy.

Tri stolpa Observability

• Metrics: Chislovye dannye vo vremeni (CPU, memory, RPS)
• Logs: Tekstovye sobytiya s kontekstom
• Traces: Put zaprosa cherez servisy

Prometheus + Grafana

Standart dlya metrik:

// Laravel Prometheus metrikiuse Prometheus\CollectorRegistry;

$registry = new CollectorRegistry(new APCAdapter());$counter = $registry->registerCounter('app', 'requests_total', 'total requests', ['method', 'endpoint']);$counter->inc(['GET', '/users']);

ELK Stack

Tsentralizovannoe logirovanie:

• Elasticsearch: Hranenie i poisk logov
• Logstash: Obrabotka i filtratsiya
• Kibana: Vizualizatsiya

Sentry dlya oshibok

Otslezhivanie isklyucheniy:

// LaravelSentry\Laravel\Integration::captureException($exception);

// JavaScriptSentry.captureException(error);

Distributed Tracing

Jaeger, Zipkin dlya trassirovki zaprosov cherez mikroservisy.

Alerty

Opovescheniya pri problemah:

• Error rate > 1%
• Response time > 500ms (p95)
• CPU usage > 80%
• Disk space < 20%