Technologies/Prometheus/traefik.entrypoint.requests.total

PrometheusMetric

traefik.entrypoint.requests.total

Total requests at entrypoint

Dimensions:None

Technical Annotations (68)

Configuration Parameters (22)

traefik.http.routers.<name>.rulerecommended: Host(`app.example.com`)

Must use backticks and match exact request Host header

traefik.http.routers.<name>.entrypointsrecommended: websecure

Must match where traffic actually arrives (web vs websecure)

metrics.prometheus.entryPointrecommended: metrics

defines the entrypoint name for exposing metrics

entryPoints.metrics.addressrecommended: :8082

port where Prometheus metrics are exposed

metrics.prometheus.headerLabels

defines custom labels from request headers; must use valid Prometheus label names

entrypoints.<name>.transport.keepalivemaxrequestsrecommended: 1000

Maximum requests per keep-alive connection; 0 means unlimited

entrypoints.<name>.transport.keepalivemaxtimerecommended: 90s

Maximum duration per keep-alive connection; 0 means unlimited

entrypoints.<name>.http2.maxconcurrentstreamsrecommended: 250-1000

Maximum parallel requests per HTTP/2 connection

entrypoints.<name>.http.maxheaderbytesrecommended: 1048576

Maximum request header size in bytes; increase if 431 errors occur

deployment.lifecycle.preStop.exec.commandrecommended: ["/bin/sh", "-c", "sleep 90"]

Delays pod termination to allow AWS NLB deregistration to complete

service.beta.kubernetes.io/aws-load-balancer-healthcheck-protocolrecommended: http

Enables active health checks instead of passive connection monitoring

service.beta.kubernetes.io/aws-load-balancer-healthcheck-portrecommended: 9000

Traefik internal API port for health endpoint

service.beta.kubernetes.io/aws-load-balancer-healthcheck-pathrecommended: /ping

Traefik health check endpoint

service.beta.kubernetes.io/aws-load-balancer-healthcheck-unhealthy-thresholdrecommended: 2

Number of failed checks before marking target unhealthy

service.beta.kubernetes.io/aws-load-balancer-healthcheck-intervalrecommended: 5

Health check interval in seconds

service.beta.kubernetes.io/aws-load-balancer-healthcheck-timeoutrecommended: 3

Health check timeout in seconds

respondingTimeouts.readTimeoutrecommended: p90 to p99 request duration

Forces connection closure after timeout period

respondingTimeouts.writeTimeoutrecommended: p90 to p99 request duration

Prevents indefinite connection hold by slow clients

respondingTimeouts.idleTimeoutrecommended: p90 to p99 request duration

Closes idle keepalive connections automatically

entryPoints.{name}.transport.respondingTimeouts.readTimeoutrecommended: 15s in example; increase for services with large request bodies (>1MiB)

Timeout for reading entire request including body; too low causes legitimate slow uploads to fail with misleading HTTP 499

experimental.fastProxyrecommended: enabled in example

Experimental feature that changes timeout error representation from HTTP 499 to HTTP 504

reusePortrecommended: false

Disable if experiencing TCP connection failures on Linux

Error Signatures (13)

404 page not foundhttp status

431http status

Empty reply from serverlog pattern

curl: (52) Empty reply from servererror code

accept tcp [::]:8443: use of closed network connectionlog pattern

503http status

499http status

499 Client Closed Requestlog pattern

context canceledlog pattern

504http status

504 Gateway Timeoutlog pattern

Error while writing request error="read tcplog pattern

i/o timeoutlog pattern

CLI Commands (8)

docker exec -it traefik sh -lc 'wget -qO- http://127.0.0.1:8080/api/http/routers'diagnostic

docker inspect app --format '{{json .Config.Labels}}' | jq -r 'to_entries[] | "\(.key)=\(.value)"' | sortdiagnostic

curl -sS -D- -o /dev/null -H 'Host: app.example.com' http://127.0.0.1/diagnostic

docker compose configdiagnostic

curl http://traefik-metrics.traefik:8082/metrics | grep traefik_service_requests_totaldiagnostic

kubectl port-forward -n traefik svc/traefik-metrics 8082:8082diagnostic

curl localhost:8082/metricsdiagnostic

while true; do curl https://my.example.host/; sleep 1; donediagnostic

Technical References (25)

EntryPointcomponentRoutercomponentHost()conceptYAMLconceptprometheus@internalcomponentkeep-aliveprotocolHTTP/2protocolHTTP/2 streamsconceptHTTP 431error codeAWS Network Load BalancercomponentpreStop lifecycle hookconceptpod deregistrationconcept/pingfile pathHEAD requestsprotocolkeepaliveprotocolentrypointcomponentHTTP/3protocolQUICprotocol0-RTTprotocolearly dataconceptentryPointcomponentrespondingTimeoutsconceptreadTimeoutconceptfastProxycomponentSO_REUSEPORTcomponent

Related Insights (14)

Router rule mismatch causes persistent 404s despite healthy containerscritical

▸

High 5xx error rate indicating backend or routing failurescritical

▸

Metrics endpoint not accessible preventing Prometheus scrapingwarning

▸

Invalid Prometheus label names prevent all metric exportcritical

▸

Keep-alive connections unlimited by default cause resource leakswarning

▸

HTTP/2 concurrent stream limit causes request queueingwarning

▸

Large request headers rejected with 431 status codewarning

▸

Pod termination before NLB deregistration causes dropped requestswarning

▸

Client connections remain open due to keepalives or incomplete readswarning

▸

Long-lived connections prevent load distribution to scaled Traefik instanceswarning

▸

IP allowlist bypass via HTTP/3 0-RTT with spoofed addressescritical

▸

Slow client request bodies exceeding readTimeout logged as HTTP 499 Client Closed Requestwarning

▸

fastProxy enabled changes readTimeout violations to HTTP 504 Gateway Timeoutwarning

▸

Linux kernel bug causes TCP connection failures with reusePortwarning

▸