Service Graph does not show ingress-nginx to actual service requests #3133
-
Hello world, The cluster contains nginx-ingress and Grafana/Loki/Tempo/Prometheus/Grafana Agent/Pyroscope all configured and ready to go with one I have also developped a React webapp instrumented with Faro connected to Grafana Agent and a Golang Gin API with GORM + Postgres database instrumented with OpenTelemetry SDKs from the code (because Beyla wasn't being cooperative on Minikube on Windows 😄). Both these are also applied via Terraform using their own Helm Charts I have configured OpenTelemetry in My issue is the Service Graph view of Tempo does not show the connections between Thank you for your help ! |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
Tempo looks for client/server pairs by using the "kind" field on the span. At a quick glance it looks like the nginx span is marked "server" as is the "wall-api" span. This would prevent the metrics generator from correctly pairing these up as a client/server pair. Generally we would prefer not to attempt to match up every span for performance reasons, but perhaps that's necessary in situations like these to get an accurate service graph? |
Beta Was this translation helpful? Give feedback.
Tempo looks for client/server pairs by using the "kind" field on the span. At a quick glance it looks like the nginx span is marked "server" as is the "wall-api" span. This would prevent the metrics generator from correctly pairing these up as a client/server pair.
Generally we would prefer not to attempt to match up every span for performance reasons, but perhaps that's necessary in situations like these to get an accurate service graph?