clj-otel
clj-otel
provides a small idiomatic Clojure API for adding telemetry to your libraries and applications using OpenTelemetry, an emerging standard for telemetry in cloud-native software, enabling effective observability.
clj-otel
is tested with Clojure 1.11.1 and is based on the reference OpenTelemetry for Java implementation, which supports Java 8 and higher.
clj-otel
is highly configurable and may be used in many ways.
This quickstart briefly outlines getting started in a local environment.
Find more in-depth information on clj-otel
in the documentation and examples.
To add manual instrumentation to your library or application at design time
Add project dependency
deps.edn
{;; ...
:deps {com.github.steffan-westcott/clj-otel-api {:mvn/version "0.2.4.1"}}}
To add traces telemetry, use Clojure functions such as steffan-westcott.clj-otel.api.trace.span/with-span!
to create spans with attributes
(defn validate-profile [profile]
(span/with-span! {:name "Validating profile"
:attributes {:system/profile-id (:id profile)}}
(validate profile)))
To add metrics telemetry, use Clojure functions in steffan-westcott.clj-otel.api.metrics.instrument
to create instruments and take measurements with attributes
(defonce set-password-failure-count
(instrument/instrument {:name "app.set-password-failure-count"
:instrument-type :counter}))
(instrument/add! set-password-failure-count {:value 1
:attributes {:reason :too-short}})
To export telemetry data (from both manual and automatic instrumentation) from an application at runtime
Download the latest OpenTelemetry instrumentation JAR opentelemetry-javaagent.jar
from the OpenTelemetry instrumentation agent releases page.
Add the following JVM options to your application, assuming export of traces telemetry only to an OTLP compliant backend such as Jaeger
"-javaagent:opentelemetry-javaagent.jar" "-Dotel.resource.attributes=service.name=NAME-OF-YOUR-SERVICE" "-Dotel.metrics.exporter=none" "-Dotel.logs.exporter=none"
To receive exported telemetry data
Prepare a telemetry backend such as Jaeger
docker run --rm \
-p 16686:16686 \
-p 4317:4317 \
jaegertracing/all-in-one \
--collector.otlp.enabled=true
To explore application behaviour described by the received telemetry data
Use telemetry backend features such as the Jaeger user interface at http://localhost:16686/search
For demonstration configurations that export traces and metrics telemetry, see the examples. |
Tutorial : A walk-through of instrumenting a small Clojure program and viewing its telemetry.
Guides : Common task recipes for adding telemetry to your Clojure libraries and applications, then configuring and running applications with telemetry.
API & Reference : API documentation for all clj-otel
modules.
Concepts : A primer on observability, OpenTelemetry and what this project clj-otel
enables for Clojure libraries and applications.
Find complete example applications in the examples
directory.
The examples aim to show:
Adding automatic and manual instrumentation to applications
Configuring and running applications that export telemetry data
Viewing telemetry data in backends
clj-otel
is a young, alpha grade project with limited use in a production setting.
Breaking API changes may still be made, but there should be few, if any.
For manual instrumentation:
Coverage of the Traces API is complete.
Trace semantics conventions support for recording exceptions is complete.
Trace semantics support for HTTP spans in applications run without the OpenTelemetry instrumentation agent is limited.
Support for wrapping asynchronous Clojure code in spans is complete.
The API is minimal and low-level, supporting any async library that works with callbacks.
Perhaps with community feedback, this will be expanded to add more specialised support for popular async libraries.
Code for creating spans around core.async
channels can be found in the examples, specifically the <with-span-binding
macro.
Coverage of the Metrics API is complete.
Metrics HTTP semantics support for applications run without the OpenTelemetry instrumentation agent is very limited.
There is currently no coverage of the Logs API.
For the programmatic configuration of the OpenTelemetry SDK:
Coverage of Traces TracerProvider
is complete.
Coverage of Metrics MeterProvider
is in progress.
Most configuration options are supported, but some public details of the OpenTelemetry Java SDK are not yet stable.
There is currently no coverage of Logging LoggerProvider
.
For manual instrumentation:
Add Logs API support.
Consider supporting more trace semantics conventions.
Maintain parity with the latest version of opentelemetry-java
.
Implement integration tests using clj-test-containers or similar.
Consider ClojureScript OpenTelemetry support in the browser and node.js using opentelemetry-js
; this will likely be a separate project.
The most needed contribution is experience reports of clj-otel
use in production systems.
I am keen to hear of usages of clj-otel
and any problems and successes.
clj-otel
is a very young project, so now is an ideal time to provide feedback on the API design as improvements can be made freely.
I will be happy to consider pull requests for minor changes, but I may not accept more significant changes while I make a start on some items in the TODO list.
For questions or feedback on clj-otel
, contact me on the #clj-otel
channel in Clojurians Slack, user steffan
.
To develop clj-otel
, you should first install the following tools:
zprint
executable binary (1.2.7 or later)
Get information on available build scripts with this command:
clojure -A:deps -T:build help/doc
Before making any pull requests, please ensure the source code has been linted and formatted with these commands:
clojure -T:build lint
clojure -T:build fmt
I want to thank:
You (yes, you) for having the curiosity to look into this project. Thank you.
My friends Golnaz and Nimmo, for pointing me in the direction of observability and OpenTelemetry. Without them, I wouldn’t have had the idea to do this project.
The OpenTelemetry community and all makers of telemetry backends for making the effective observability of systems a tangible reality. Cloud-native software is so complex now, we need all the help we can get to understand how well it is (or is not) working.
The Diátaxis Documentation Framework, for a simple way to structure technical documentation.
Copyright © 2021-2023 Steffan Westcott
Distributed under the Apache License v2.0
Can you improve this documentation?Edit on GitHub
cljdoc is a website building & hosting documentation for Clojure/Script libraries
× close