Technologies/Traefik/traefik.service.request.duration
TraefikTraefikMetric

traefik.service.request.duration

Request duration at service
Dimensions:None
Available on:PrometheusPrometheus (1)DatadogDatadog (1)Native (1)
Interface Metrics (3)
PrometheusPrometheus
Request processing duration histogram for a service
Dimensions:None
DatadogDatadog
Request processing duration at the service level
Dimensions:None
Native
Request processing duration at service
Dimensions:None
Sources
traefik_service_request_duration_secondsdoc.traefik.io
traefik.service.request.durationdocs.datadoghq.com
traefik.service.request.durationdoc.traefik.io

Technical Annotations (45)

Configuration Parameters (15)
traefik.docker.networkrecommended: <shared_network_name> (e.g., traefik_proxy)
Docker Compose label that explicitly specifies which network Traefik should use for service discovery when container is attached to multiple networks
ADVERTISE_IPrecommended: Internal IP address (e.g., https://192.168.0.100:32400/) rather than external domain
Plex environment variable that may reduce hairpin NAT reliance
metrics.prometheus.bucketsrecommended: [0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1.0, 2.5, 5.0, 10.0]
Histogram buckets for latency distribution measurement
accessLog.filters.minDurationrecommended: 10ms
logs only requests exceeding this duration for performance analysis
serversTransport.forwardingTimeouts.idleConnTimeoutrecommended: shorter than upstream timeout (e.g., 3s if upstream is 5s)
Controls how long Traefik keeps idle connections to upstream servers open
serversTransport.maxIdleConnsPerHostrecommended: default (not -1)
Setting to -1 disables keep-alive entirely, causing performance issues
retry.attemptsrecommended: 3-5
Maximum retry attempts to bound total duration
retry.initialIntervalrecommended: 100ms-200ms
Initial backoff interval between retries
responseForwarding.flushIntervalrecommended: 100ms
Timeout for individual backend request attempts
respondingTimeouts.readTimeoutrecommended: p90-p99 percentile
Start at p90 for diagnosis, tune to p99 for production
respondingTimeouts.writeTimeoutrecommended: p90-p99 percentile
Start at p90 for diagnosis, tune to p99 for production
respondingTimeouts.idleTimeoutrecommended: p90-p99 percentile
Start at p90 for diagnosis, tune to p99 for production
entryPoints.*.transport.respondingTimeouts.readTimeoutrecommended: 600s
Maximum duration for reading request including body for large uploads
entryPoints.*.transport.respondingTimeouts.idleTimeoutrecommended: 600s
Maximum duration an idle connection remains open during large uploads
entryPoints.*.transport.respondingTimeouts.writeTimeoutrecommended: 600s
Maximum duration for writing response for large file operations
Error Signatures (2)
504 Gateway Timeouthttp status
502 Bad Gatewayhttp status
CLI Commands (1)
docker-compose up -d --force-recreate backenddiagnostic
Technical References (27)
Docker providercomponentservice discoveryconceptapache 32bitcomponentactivemqcomponentcronjobcomponentlog rotationconceptdisk I/OcomponentRound Robinconceptload balancingconceptdeadlockconceptsynchronizationconceptPuma 6 chipsetcomponenthairpin NATconceptloopback functionalityconceptlatencyconceptrequest durationconcepterror ratiosconceptServersTransportcomponentforwardingTimeoutscomponentkeep-aliveprotocolp50conceptp90conceptp99conceptDocker SwarmcomponentMTUconceptentryPointscomponentrespondingTimeoutscomponent
Related Insights (16)
Traefik 504 Gateway Timeout after backend container restart on multiple networkscritical
▸
Timeouts occur when proxying to Apache backend despite backend capacitywarning
▸
Log rotation background jobs cause periodic response time spikeswarning
▸
Disk I/O bottleneck delays all requests globallycritical
▸
Single backend server failure causes partial user impact in load-balanced setupwarning
▸
Code deadlocks from excessive synchronization cause service outagescritical
▸
Router loopback bug causes 100x latency increase when accessing Traefik via external domain from internal networkcritical
▸
High P95 latency exceeds acceptable thresholdswarning
▸
High request latency at p99 degrading user experiencewarning
▸
Uptime-based alerts waste debugging time compared to latency spike alertsinfo
▸
Monitoring raw traffic totals delays detection of production health issuesinfo
▸
Idle timeout mismatch between Traefik and upstream causes 502 errorscritical
▸
Unbounded retry configuration extends request latency indefinitelywarning
▸
Timeout configuration should be based on request duration percentilesinfo
▸
Traefik reports significantly higher request duration than backend servicewarning
▸
Large file uploads fail due to default Traefik timeout valueswarning
▸