High HTTP 4xx/5xx error alert
This alert is triggered when the HTTP request success rate drops below 95 percent. The alert threshold can be changed to your desired error alerting requirements.
Ingress certificate expiration alert
This alert is triggered when there are less than 30 days remaining for certificate expiration. The alert threshold can be changed to your desired expiration alerting requirements.
Nginx-ingress config load errors alert
This alert is triggered when there is an error loading ingress configuration
P99 latency alert
This sample alert is triggered when the controller request processing duration for the 99th percentile of requests goes above 100 ms. The alert threshold can be changed to your desired latency for alerting requirements.
Why monitor the NGINX ingress controller?
The NGINX ingress controller exposes many useful metrics that can be scraped by a Prometheus server or agent. These metrics can inform you of ongoing process connections and the current ingress load that your system is handling. Furthermore, these metrics can inform you on ingress config reloads and empower alert conditions on config reload errors or unexpected config changes.
NGINX ingress controller quickstart highlights
- Config reload monitoring (errors and last successful reload)
- Throughput faceted by ingress classes
- Request/response insights on payload size and response time
- CPU and memory statistics
- This quickstart is based on metrics sent to New Relic using Prometheus Agent or Prometheus Server.
Monitoring NGINX ingress controller
The NGINX ingress controller has to have metrics enabled before they are available to be scraped. Instructions for ensuring your NGINX metrics are enabled can be found here: https://kubernetes.github.io/ingress-nginx/user-guide/monitoring/#wildcard-ingresses