OpenTelemetry - RunsOn

This page covers RunsOn metrics end to end: the per-job runner metrics you get in the GitHub Actions UI, and the authoritative OpenTelemetry reference for everything RunsOn exports over OTLP — what it emits, when, and where the automatic behavior stops.

At a glance

RunsOn surfaces per-job resource metrics (CPU, memory, disk, network, I/O) through three paths, from zero-config inline charts to full OTLP export. Server-side control-plane signals go out over OTLP only.

Path	Where it appears	Setup	Best for
Built-in inline charts	`Complete runner` step + `metrics.jsonl` in S3	None — on by default	Default rightsizing and single-job troubleshooting
`runs-on/action@v2` CloudWatch	`Post Run runs-on/action@v2` step	Add the action with `metrics:`	AWS-native per-job metrics — see CloudWatch
Runner OTLP export	Your OTLP backend	`extras=otel` + stack OTLP endpoint	Long-term, fleet-wide observability

These are not mutually exclusive — a job can render inline charts and export to OTLP at the same time.

Per-job runner metrics

Use these to rightsize runners and spot job-level constraints without leaving the GitHub Actions UI.

Built-in inline charts

This is the default path on current RunsOn runners:

The runner collects local host metrics during the job.
RunsOn writes those samples into metrics.jsonl on the runner.
The Complete runner step renders ASCII charts from that file.
The same step uploads metrics.jsonl into your RunsOn S3 bucket for later inspection.

This flow is separate from remote OTLP export — you get the Complete runner charts and the uploaded metrics.jsonl file without enabling extras=otel.

Example excerpt from a real Complete runner step

Runner metadata

For each runner you also get detailed metadata about the EC2 instance, RunsOn installation, and runner timings — expand the “Set up job” section in the GitHub Actions UI:

Server-side behavior

When OtelExporterEndpoint is configured, the RunsOn server exports OTLP logs, metrics, and traces from the server process. That includes structured server logs, the RunsOn-defined server metrics inventory below, and the Go runtime appendix.

Current v3 installs do not expose a Prometheus /metrics endpoint. Use OTLP export through OtelExporterEndpoint for server-side metrics.

Runner-side behavior

Runner-side OTEL behavior is available starting with RunsOn v2.12.0.

When the stack has an OTLP endpoint configured and a job opts in with extras=otel, the runner exports its host metrics (CPU load and utilization, memory, disk I/O, network I/O) and Go-runtime metrics to your observability backend, alongside the per-step job spans. Add extras=otel to a job label, or to the runner spec used by a pool:

runs-on: runs-on=${{ github.run_id }}/runner=2cpu-linux-x64/extras=otel

The collector always writes local metrics.jsonl for the built-in inline charts, independent of OTLP — enabling one does not require the other.
Remote OTLP export only happens when the runner has OTEL enabled and the stack OTLP endpoint is configured.
RunsOn exports the same default runner host-metrics set on Linux and Windows.

Logs

When the stack OTLP endpoint is configured, RunsOn server logs are exported automatically from the server process. They also remain available in CloudWatch through the usual AWS log groups.

RunsOn forwards the runner bootstrap log file (output.log) through the local collector when runner OTEL is enabled.

This is the RunsOn bootstrap and agent log path. It is not the full GitHub workflow job log stream.

If the EC2 instance log group is configured, the same bootstrap log file also remains available through the instance log group path in CloudWatch.

Traces

RunsOn emits server traces plus agent and instance lifecycle traces.

Per-step job spans are also automatically emitted to your OTLP backend after a job completes. Those spans are emitted from the RunsOn server and you don’t need to set extras=otel on your jobs to get them.

Automatically emitted per-step job spans

Reference: metric inventory

RunsOn server metrics

Metric	Kind	Unit	Description	Attributes	Transport
`runs_on_jobs_total`	`counter`	`{job}`	Total number of jobs by status.	`conclusion` `instance_family` `instance_lifecycle` `instance_type` `interrupted` `org` `pool_name` `pool_type` `repo_full_name` `status` `workflow_name`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_internal_queue_duration_seconds`	`histogram`	`s`	Time from job queued in RunsOn to scheduled.	`conclusion` `instance_family` `instance_lifecycle` `interrupted` `pool_name` `status`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_overall_queue_duration_seconds`	`histogram`	`s`	Time from job queued by GitHub to started.	`conclusion` `instance_family` `instance_lifecycle` `interrupted` `pool_name` `status`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_job_duration_seconds`	`histogram`	`s`	Time from job started to completed.	`conclusion` `instance_family` `instance_lifecycle` `interrupted` `pool_name` `status`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_pool_instances_total`	`observable_gauge`	`{instance}`	Current number of pool instances by state.	`installation_id` `org` `pool_name` `state`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_rate_limiter_tokens`	`observable_gauge`	`{token}`	Available tokens in a rate limiter.	`limiter`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_rate_limiter_burst`	`observable_gauge`	`{token}`	Burst capacity of a rate limiter.	`limiter`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_spot_circuit_breaker_active`	`observable_gauge`	`1`	Whether the spot circuit breaker is active; 1 means active and 0 means inactive.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_github_operations_total`	`counter`	`{operation}`	Total number of GitHub API operations by operation name.	`github_app_label` `github_operation`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_aws_operations_total`	`counter`	`{operation}`	Total number of AWS API operations by operation name.	`aws_operation`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_scaling_boost_active`	`gauge`	`{app}`	Current number of configured GitHub boost apps.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_registration_jit_duration_seconds`	`histogram`	`s`	Time spent generating a GitHub JIT runner configuration.	`github_app_label`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_jobs_queue_fetched`	`histogram`	`{message}`	Number of job messages fetched from the jobs queue per poll iteration.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_reconciler_backlog`	`gauge`	`{job}`	Current number of jobs waiting to be reconciled.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_provisioning_backlog`	`gauge`	`{job}`	Current number of provisioning jobs waiting in the launch or registration queue.	`runs_on.provisioning_stage`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_provisioning_inflight`	`gauge`	`{job}`	Current number of provisioning jobs actively being processed in the launch or registration stage.	`runs_on.provisioning_stage`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_provisioning_stage_duration_seconds`	`histogram`	`s`	Wall-clock time spent in a provisioning stage worker pass.	`runs_on.provisioning_stage`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`runs_on_provisioning_registration_retries_total`	`counter`	`{retry}`	Total number of retryable provisioning registration attempts.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.

Runner host metrics exported to OTLP by default

Notes:

Remote OTLP export requires extras=otel, a configured OTLP endpoint, and JobEnabled=true.
Applies to Linux and Windows.
Disk metrics are limited to the detected primary/root disk, network metrics to the detected primary network interface, and filesystem metrics are not exported by default.

Metric	Kind	Unit	Description	Attributes
`system.cpu.load_average.15m`	`gauge`	`{thread}`	Average CPU Load over 15 minutes.	-
`system.cpu.load_average.1m`	`gauge`	`{thread}`	Average CPU Load over 1 minute.	-
`system.cpu.load_average.5m`	`gauge`	`{thread}`	Average CPU Load over 5 minutes.	-
`system.cpu.utilization`	`gauge`	`1`	Difference in system.cpu.time since the last measurement per logical CPU, divided by the elapsed time (value in interval [0,1]).	`cpu` `state`
`system.disk.io`	`sum`	`By`	Disk bytes transferred.	`device` `direction`
`system.disk.operations`	`sum`	`{operations}`	Disk operations count.	`device` `direction`
`system.memory.utilization`	`gauge`	`1`	Percentage of memory bytes in use.	`state`
`system.network.io`	`sum`	`By`	The number of bytes transmitted and received.	`device` `direction`

Go runtime appendix

Metric	Kind	Unit	Description	Attributes	Transport
`go.config.gogc`	`observable_up_down_counter`	`%`	Heap size target percentage configured by the user, otherwise 100.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`go.goroutine.count`	`observable_up_down_counter`	`{goroutine}`	Count of live goroutines.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`go.memory.allocated`	`observable_counter`	`By`	Memory allocated to the heap by the application.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`go.memory.allocations`	`observable_counter`	`{allocation}`	Count of allocations to the heap by the application.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`go.memory.gc.goal`	`observable_up_down_counter`	`By`	Heap size target for the end of the GC cycle.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`go.memory.limit`	`observable_up_down_counter`	`By`	Go runtime memory limit configured by the user, if a limit exists.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`go.memory.used`	`observable_up_down_counter`	`By`	Memory used by the Go runtime.	`go.memory.type`	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`go.processor.limit`	`observable_up_down_counter`	`{thread}`	The number of OS threads that can execute user-level Go code simultaneously.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.
`go.schedule.duration`	`histogram`	`s`	The time goroutines have spent in the scheduler in a runnable state before actually running.	-	Exported through OTLP metrics when the stack OTLP endpoint is configured. Structured metric snapshots are also emitted to logs when metrics logging is enabled.

Reference: attribute inventory

RunsOn server metric attributes

Context	Attribute	Action	Description
	`aws_operation`		AWS API operation name.
	`conclusion`		GitHub job conclusion for completed jobs; empty for other states.
	`github_app_label`		GitHub app label: primary for the control-plane app, or the configured boost app key.
	`github_operation`		GitHub API operation name.
	`installation_id`		GitHub App installation identifier.
	`instance_family`		EC2 instance family derived from the resolved runner instance type.
	`instance_lifecycle`		Runner pricing lifecycle such as spot or on-demand.
	`instance_type`		EC2 instance type when a job has been assigned to a runner.
	`interrupted`		Whether the job was interrupted before completion.
	`limiter`		Rate limiter identifier such as github_api or ec2_run.
	`org`		GitHub organization name.
	`pool_name`		Pool name when the job is served from a pool-backed runner.
	`pool_type`		Pool standby type when applicable.
	`repo_full_name`		Repository full name in owner/repo form.
	`runs_on.provisioning_stage`		Provisioning stage, either launch or registration.
	`state`		Current pool instance state.
	`status`		Job lifecycle status such as queued, scheduled, in_progress, or completed.
	`workflow_name`		GitHub workflow display name.

RunsOn server resource attributes

Context	Attribute	Action	Description
	`cloud.region`		AWS region when configured.
	`deployment.environment`		Stack environment name when configured.
	`service.instance.id`		Server host identifier when the hostname can be resolved.
	`service.name`		Always runs-on-server.
	`service.namespace`		Stack name when configured.
	`service.version`		RunsOn app version when configured.

Runner metric attributes exported to OTLP by default

Context	Attribute	Description
`cpu`	`cpu`	Logical CPU number starting at 0.
`cpu`	`state`	Breakdown of CPU usage by type.
`disk`	`device`	Name of the disk.
`disk`	`direction`	Direction of flow of bytes/operations (read or write).
`memory`	`state`	Breakdown of memory usage by type.
`network`	`device`	Name of the network interface.
`network`	`direction`	Direction of flow of bytes/operations (receive or transmit).

Runner resource attributes

Attribute	Action	Description
`service.name`	`insert`	Inserted as runs-on-agent when the incoming telemetry does not already define a service name.
`service.namespace`	`insert`	Inserted as runs-on for runner-generated telemetry when the incoming telemetry does not already define a service namespace.
`service.instance.id`	`insert`	EC2 instance identifier.
`deployment.environment`	`upsert`	Configured RunsOn environment name.
`deployment.environment.name`	`upsert`	Configured RunsOn environment name.
`stack_name`	`upsert`	RunsOn stack name.
`region`	`upsert`	AWS region.
`org`	`upsert`	GitHub organization name.
`instance_type`	`upsert`	EC2 instance type.
`instance_lifecycle`	`upsert`	EC2 pricing lifecycle such as spot or on-demand.
`availability_zone`	`upsert`	EC2 availability zone.
`ami_id`	`upsert`	AMI identifier used by the runner instance.
`pool_name`	`upsert`	Pool name when the runner comes from a pool.
`pool_type`	`upsert`	Pool standby type when the runner comes from a pool.
`repo_full_name`	`upsert`	Repository full name in owner/repo form.
`workflow_path`	`upsert`	GitHub workflow file path.
`job_name`	`upsert`	GitHub job name.

Go runtime metric attributes

Context	Attribute	Action	Description
`go-runtime`	`go.memory.type`		Type of Go runtime memory, currently stack or other.