Telemetry

Every sandbox runs an OpenTelemetry Collector that collects host metrics, logs, traces, and spans and pushes them to an OTLP HTTP endpoint that you configure via environment variables. Your code can also send its own OTLP telemetry to the local collector, which forwards it alongside the host data.

The JavaScript and Python SDKs can also emit OpenTelemetry spans and metrics for SDK calls such as sandbox creation and tool usage. That lets you correlate client-side operations with sandbox-side telemetry in the same backend.

Metrics

Each sandbox automatically collects the following metrics:

Metric	Description
leap0.sandbox.cpu.utilization	CPU usage percentage
leap0.sandbox.cpu.limit	Logical CPU count
leap0.sandbox.memory.utilization	Memory usage percentage
leap0.sandbox.memory.usage	Memory used in bytes
leap0.sandbox.memory.limit	Memory limit in bytes
leap0.sandbox.filesystem.utilization	Disk usage percentage
leap0.sandbox.filesystem.usage	Disk space used in bytes

Logs, traces, and spans

Any OTLP logs, traces, and spans your code sends to the local collector (ports 4317/4318) are forwarded to the configured endpoint.

SDK OpenTelemetry

The JavaScript and Python SDKs can emit OpenTelemetry spans and metrics for SDK operations such as creating sandboxes and calling sandbox tools.

Enable that client-side instrumentation by setting sdkOtelEnabled: true on new Leap0Client(...), sdk_otel_enabled=True on the Python client, or LEAP0_SDK_OTEL_ENABLED=true in your local environment.

If OTEL_EXPORTER_OTLP_ENDPOINT is already set locally, the SDK enables this instrumentation automatically.

This is separate from sandbox telemetry: use otelExport: true in the JavaScript SDK or otel_export=True in Python to forward telemetry produced inside the sandbox itself.

JavaScript SDK and sandbox example

import { Leap0Client } from "leap0"

process.env.OTEL_EXPORTER_OTLP_ENDPOINT = "https://otel.example.com"
process.env.OTEL_EXPORTER_OTLP_HEADERS = "x-api-key=abc123"

const client = new Leap0Client({ sdkOtelEnabled: true })
const sandbox = await client.sandboxes.create({
templateName: "my-template",
vcpu: 2,
memoryMib: 2048,
timeoutMin: 30,
otelExport: true,
})

console.log(sandbox.id)
await client.close()

Resource attributes

Every signal is tagged with leap0.sandbox.id, leap0.org.id, and leap0.template.id so you can filter and group by sandbox.

Configuration

Set these environment variables when creating a sandbox to enable telemetry export.

Your own endpoint

If you run your own observability backend (Grafana, Datadog, etc.), pass the standard OTLP environment variables in the sandbox env_vars:

Variable	Description
OTEL_EXPORTER_OTLP_ENDPOINT	Your OTLP HTTP endpoint (e.g. https://otel.example.com)
OTEL_EXPORTER_OTLP_HEADERS	Comma-separated key=value auth headers (e.g. x-api-key=abc123)

Example

Create a sandbox with telemetry export enabled. Metrics, logs, and traces will be pushed to the configured endpoint as OTLP HTTP.

Python

from leap0 import Leap0Client

client = Leap0Client()
sandbox = client.sandboxes.create(
  template_name="my-template",
  vcpu=2,
  memory_mib=2048,
  timeout_min=30,
  otel_export=True,
)

print(sandbox.id)

JavaScript & TypeScript

import { Leap0Client } from "leap0"

process.env.OTEL_EXPORTER_OTLP_ENDPOINT = "https://otel.example.com"
process.env.OTEL_EXPORTER_OTLP_HEADERS = "x-api-key=abc123"

const client = new Leap0Client()
const sandbox = await client.sandboxes.create({
templateName: "my-template",
vcpu: 2,
memoryMib: 2048,
timeoutMin: 30,
otelExport: true,
})

console.log(sandbox.id)
await client.close()