OpenTelemetry AWS X-Ray support for Erlang/Elixir.
It works with the AWS Distro for OpenTelemetry Collector, a version of the OpenTelemetry Collector which has support for AWS services such as X-Ray. This collector accepts standard OpenTelemetry traces, converts them to X-Ray format, and sends them to AWS. You can run the collector as a sidecar container in an ECS task or as a daemon on an EC2 instance.
In AWS X-Ray, the trace_id
is a 128-bit value. The first 32 bits are a Unix
time_t
and the rest are a 96-bit random number. If you use the default
trace_id
, then X-Ray will reject your traces. This library generates ids that
are compatible with X-Ray.
When your app app is downstream from another app or the AWS load balancer, the
upstream app creates the trace for the current request and sends it in the
X-Amzn-Trace-Id
HTTP header. The header includes the trace id and optional
information about the parent span and sampling.
This library includes a propagator which reads the trace id from this header and uses it within your app. It can then pass the same trace id to downstream apps via the header.
NOTE: By default Amazon samples relatively few traces. If you want to ensure that your traces are sampled, make sure that you turn on sampling in your app. A common approach is to turn on sampling for all traces that have errors and for some percentage of normal traces.
This library includes the following modules:
-
An id generator that creates X-Ray-compatible
trace_id
andspan_id
. It implements theotel_id_generator
protocol in the Erlang SDK. -
A propagator that reads and writes AWS X-Ray trace context headers. It implements the
otel_propagator_text_map
protocol in the Erlang SDK.
Links:
- Propagators in general: https://opentelemetry.io/docs/specs/otel/context/api-propagators/
- Erlang SDK propagation: https://opentelemetry.io/docs/instrumentation/erlang/propagation/
- Erlang SDK id generator: https://github.com/open-telemetry/opentelemetry-erlang/blob/main/apps/opentelemetry/src/otel_id_generator.erl
- X-Ray tracing header: https://docs.aws.amazon.com/xray/latest/devguide/xray-concepts.html#xray-concepts-tracingheader
- X-Ray sampling: https://docs.aws.amazon.com/xray/latest/devguide/xray-concepts.html#xray-concepts-sampling
- X-Ray configuration: https://aws-otel.github.io/docs/getting-started/x-ray#configuring-the-aws-x-ray-exporter
- AWS CloudWatch Log Group resource: aws/aws-xray-sdk-python#188
- OpenTelemetry resources: https://opentelemetry.io/docs/instrumentation/erlang/resources/
- OpenTelemetry getting started: https://opentelemetry.io/docs/instrumentation/erlang/getting-started/
- OpenTelemetry intro: https://davelucia.com/blog/observing-elixir-with-lightstep
- Addding Erlang logger filters to new Elixir logger: https://write.as/yuriploc/elixir-logger-and-erlang-filters
Erlang:
Add opentelemetry_xray
to the list of dependencies in rebar.config
:
{deps, [opentelemetry_xray]}.
Elixir:
Add opentelemetry_xray
to the list of dependencies in mix.exs
:
def deps do
[
{:opentelemetry_xray, "~> 0.7"},
]
end
Erlang:
In sys.config
:
{opentelemetry, {
id_generator, [opentelemetry_xray_id_generator],
propagators: [opentelemetry_xray_propagator, baggage]
}}
Elixir:
In config/prod.exs
, configure opentelemetry
to use this library:
config :opentelemetry,
id_generator: :opentelemetry_xray_id_generator,
propagators: [:opentelemetry_xray_propagator, :baggage]
Add resource attributes to the span to connect it to log messages:
export OTEL_RESOURCE_ATTRIBUTES="aws.log.group.names=$AWS_LOG_GROUP"
See phoenix_container_example for a complete Elixir Phoenix app that uses this library.