Grafana Cloud

Configure golang to generate Prometheus metrics

For instructions on how to instrument a Go application with Prometheus exporter, refer to Instrumenting a Go application for Prometheus.

Metrics and KPIs

Memory process_resident_memory_bytes

Utilization in k8s

process_resident_memory_bytes / kube_node_status_allocatable{resource=“memory”}

Utilization in non k8s

process_resident_memory_bytes / node_memory_MemTotal_bytes



Utilization in k8s

rate(process_cpu_seconds_total[5m]) / kube_pod_container_resource_limits{resource=“cpu”}

Utilization in non k8s

rate(process_cpu_seconds_total[5m]) / rate(node_cpu_seconds_total[5m])


The following Go KPI dashboard shows KPIs listed in the table, and the following Go runtime metrics.

Memory metrics

  • Heap Memory go_memstats_heap_alloc_bytes
  • Off-Heap Memory go_memstats_mspan_inuse_bytes
  • Stack Memory go_memstats_stack_inuse_bytes
  • Total Memory go_memstats_sys_bytes

Objects, allocation, and GC metrics

  • Live Object Count go_memstats_mallocs_total-go_memstats_frees_total
  • Object Allocation Rate rate(go_memstats_mallocs_total)
  • Memory Allocation Rate rate(go_memstats_alloc_bytes_total)
  • GC Duration Quantiles go_gc_duration_seconds

Go routines and pointers metrics

  • Number of routines go_goroutines
  • Number of pointer dereferences rate(go_memstats_lookups_total)

Golang runtime