Monitoring & Debugging

Configure Service Monitor, Prometheus metrics, profiler, store port, and eBPF tracer.

  2 minute read  

This page covers Helm values for monitoring, debugging, and observability configuration.

Edge Delta agents expose several interfaces for performance monitoring and debugging. The Service Monitor enables Prometheus to scrape metrics automatically in Kubernetes environments. The profiler port allows detailed CPU and memory analysis using pprof. The Prometheus port exposes agent metrics for external monitoring systems. The eBPF tracer provides network-level visibility for service maps and distributed tracing. See Performance Profiling Agents for detailed guidance on using the profiler to analyze agent performance.

Service Monitor

Service Monitor - Enable

Variable: serviceMonitor

Description: Enable service monitor to scrape Prometheus metrics.

Profiler and Metrics Ports

Profiler Port

Variable: profilerPort

Description: Specify the port to use if you install Edge Delta with a profiler to monitor CPU and memory statistics. Alternatively, you can use Prometheus with its dedicated endpoint.

Example: profilerPort=6060

Prom Port

Variable: promPort

Description: Specify the metrics endpoint port number that Prometheus can use to scrape metrics.

Example: promPort=8087

Store Port

Variable: storePort

Description: A port number to expose pipeline metrics storage.

Example: storePort=6062

eBPF Tracer

Tracer Properties (eBPF)

Variable: tracerProps.enabled

Description: Controls whether eBPF-based telemetry sources are enabled. When set to true (default), the agent can use eBPF sources like Kubernetes Trace (k8s_trace_input) and Kubernetes Service Map (k8s_traffic_input) for network-level monitoring and tracing.

eBPF sources provide powerful observability capabilities but consume significant CPU and memory resources. Disabling them can substantially reduce agent resource consumption.

Default: true

Example - Disable eBPF:

helm upgrade edgedelta edgedelta/edgedelta \
  --set tracerProps.enabled=false \
  -n edgedelta

When to Disable:

  • Using alternative APM or distributed tracing solutions
  • Resource constraints are critical in your environment
  • Service-to-service visibility is provided by other means (e.g., Istio, Linkerd)
  • You don’t need the Service Map visualization

When to Keep Enabled:

  • You need the Service Map visualization
  • Network traffic patterns and latency monitoring are critical
  • Troubleshooting inter-service communication issues

For detailed resource optimization strategies, see Reducing Agent Resource Consumption.

Note: Disabling tracerProps does not remove the source nodes from your pipeline configuration. To fully disable eBPF functionality, either set tracerProps.enabled=false or remove the k8s_trace_input and k8s_traffic_input nodes from your pipeline YAML.