Launcher, a Lightstep Distro for OpenTelemetry 🚀

NOTE: This is in beta and is expected to GA in Fall 2020.

What is Launcher?

Launcher is a configuration layer that chooses default values for configuration options that many OpenTelemetry users want. It provides a single function in each language to simplify discovery of the options and components available to users. The goal of Launcher is to help users that aren't familiar with OpenTelemetry quickly ramp up on what they need to get going and instrument.

Getting started

pip install opentelemetry-launcher

Configure

Minimal setup

from opentelemetry.launcher import configure_opentelemetry
from opentelemetry import trace

configure_opentelemetry(
    service_name="service-123",
    access_token="my-token",  # optional
)

tracer = trace.get_tracer(__name__)

with tracer.start_as_current_span("foo") as span:
    span.set_attribute("attr1", "valu1")
    with tracer.start_as_current_span("bar"):
        with tracer.start_as_current_span("baz"):
            print("Hello world from OpenTelemetry Python!")

Additional tracer options

configure_opentelemetry(
    service_name="service-123",
    service_version="1.2.3",
    access_token="my-token",
    span_exporter_endpoint="ingest.lightstep.com:443",
    metric_exporter_endpoint="ingest.lightstep.com:443",
    log_level=debug,
    span_exporter_insecure=False,
    metric_exporter_insecure=False,
)

Usage with Auto Instrumentation

OpenTelemetry Python includes a command that allows the user to automatically instrument certain third party libraries. Here is an example that shows how to use this launcher with auto instrumentation.

First, create a new virtual environment:

cd ~
mkdir auto_instrumentation
virtualenv auto_instrumentation
source auto_instrumentation/bin/activate
pip install opentelemetry-launcher
pip install requests
pip install flask
opentelemetry-bootstrap -a install

Once that is done, clone the opentelemetry-python repo to get the example code:

git clone git@github.com:open-telemetry/opentelemetry-python.git
cd opentelemetry-python

Set the environment variables:

export LS_SERVICE_NAME=auto-instrumentation-testing
export LS_ACCESS_TOKEN=<the access token>

Run the server:

cd docs/examples/auto-instrumentation
opentelemetry-instrument python server_uninstrumented.py

Run the client in a separate console:

cd docs/examples/auto-instrumentation
python client.py testing

This should produce spans that can be captured in the Lightstep Explorer. This should also produce System Metrics that can be captured in the Lightstep Dashboards.

Configuration Options

The configuration option for propagators accepts a comma-separated string that will be interpreted as a list. For example, a,b,c,d will be interpreted as ["a", "b", "c", "d"]. The configuration option for resource_attributes accepts a comma-separated string of key=value pairs that will be interpreted as a dictionary. For example, a=1,b=2,c=3,d=4 will be interpreted as {"a": 1, "b": 2, "c": 3, "d": 4}.

Principles behind Launcher

100% interoperability with OpenTelemetry

One of the key principles behind putting together Launcher is to make lives of OpenTelemetry users easier, this means that there is no special configuration that requires users to install Launcher in order to use OpenTelemetry. It also means that any users of Launcher can leverage the flexibility of configuring OpenTelemetry as they need.

Validation

Another decision we made with launcher is to provide end users with a layer of validation of their configuration. This provides us the ability to give feedback to our users faster, so they can start collecting telemetry sooner.

Start using it today in Go, Java, Javascript and Python and let us know what you think!

Made with @ Lightstep