How does Amplify detect 5xx errors?

The agent parses the access log in real time and emits per-status-class counters: nginx.http.status.5xx for server errors, nginx.http.status.4xx for client errors. Both are graphed as rates.

Can I alert on error-rate spikes?

Yes. Configure alert rules on nginx.http.status.5xx with absolute or percentage-of-total thresholds. Routes: Slack, Telegram, email, webhooks.

Do you capture 499 timeouts?

Yes. 499 is parsed from the log as a client-closed-connection class and shows up in the 4xx breakdown with its own line.

NGINX Error Rate Monitoring
5xx alerts that fire in seconds, not minutes.

Per-status-code error rate from the access log. Alerts on 5xx spikes, 499 timeouts and 502/504 upstream failures. Routes to Slack, Telegram, email, webhooks.

Start free See demo

Error metrics we graph

nginx.http.status.5xx

Server-error rate. Broken down by 500, 502, 503, 504.

nginx.http.status.4xx

Client-error rate. 404, 403, 499 tracked individually.

nginx.http.status.502

Bad-gateway rate — almost always an upstream problem.

nginx.http.status.504

Gateway-timeout rate — slow upstream or proxy_read_timeout too low.

Example alert recipes

5xx spike: nginx.http.status.5xx > 5 rps for 2 minutes → Slack
502 sustained: nginx.http.status.502 > 1 rps for 5 minutes → Telegram + webhook to PagerDuty
499 storm: nginx.http.status.4xx.499 > 20% of total requests → email (often a CDN / client-side issue)

Know when NGINX starts failing.

Start free View pricing

NGINX Error Rate Monitoring5xx alerts that fire in seconds, not minutes.