otel-cli is a command-line tool for sending OpenTelemetry traces. It is written in Go and intended to be used in shell scripts and other places where the best option available for sending spans is executing another program.
otel-cli can be added to your scripts with no configuration and it will run as normal but in non-recording mode and will emit no traces. This follows the OpenTelemetry community's philosophy of "first, do no harm" and makes it so you can add otel-cli to your code and later turn it on.
Since otel-cli needs to connect to the OTLP endpoint on each run, it is highly recommended to use a localhost opentelemetry collector that can buffer spans so that the connection cost does not slow down your program too much.
We publish a number of package formats for otel-cli, including tar.gz, zip (windows), apk (Alpine), rpm (Red Hat variants), deb (Debian variants), and a brew tap. These can be found on the repo's Releases page.
On most platforms the easiest way is a go get:
go get github.com/equinix-labs/otel-cli
To use the brew tap e.g. on MacOS:
brew tap equinix-labs/otel-cli
brew install otel-cli
Alternatively, clone the repo and build it locally:
git clone [email protected]:equinix-labs/otel-cli.git
cd otel-cli
go build
# run otel-cli as a local OTLP server and print traces to your console
# run this in its own terminal and try some of the commands below!
otel-cli server tui
# configure otel-cli to talk the the local server spawned above
export OTEL_EXPORTER_OTLP_ENDPOINT=localhost:4317
# run a program inside a span
otel-cli exec --service my-service --name "curl google" curl https://google.com
# otel-cli propagates context via envvars so you can chain it to create child spans
otel-cli exec --kind producer "otel-cli exec --kind consumer sleep 1"
# if a traceparent envvar is set it will be automatically picked up and
# used by span and exec. use --tp-ignore-env to ignore it even when present
export TRACEPARENT=00-0af7651916cd43dd8448eb211c80319c-b7ad6b7169203331-01
# create a span with a custom start/end time using either RFC3339,
# same with the nanosecond extension, or Unix epoch, with/without nanos
otel-cli span --start 2021-03-24T07:28:05.12345Z --end 2021-03-24T07:30:08.0001Z
otel-cli span --start 1616620946 --end 1616620950.241980634
# so you can do this:
start=$(date --rfc-3339=ns) # rfc3339 with nanoseconds
some-interesting-program --with-some-options
end=$(date +%s.%N) # Unix epoch with nanoseconds
otel-cli span -n my-script -s some-interesting-program --start $start --end $end
# for advanced cases you can start a span in the background, and
# add events to it, finally closing it later in your script
sockdir=$(mktemp -d)
otel-cli span background \
--service $0 \
--name "$0 runtime" \
--sockdir $sockdir & # the & is important here, background server will block
sleep 0.1 # give the background server just a few ms to start up
otel-cli span event --name "cool thing" --attrs "foo=bar" --sockdir $sockdir
otel-cli span end --sockdir $sockdir
# or you can kill the background process and it will end the span cleanly
kill %1
# server mode can also write traces to the filesystem, e.g. for testing
dir=$(mktemp -d)
otel-cli server json --dir $dir --timeout 60 --max-spans 5
Everything is configurable via CLI arguments, json config, and environment variables. If no endpoint is specified, otel-cli will run in non-recording mode and not attempt to contact any servers.
All three modes of config can be mixed. Command line args are loaded first, then config file, then environment variables.
CLI argument | environment variable | config file key | example value |
---|---|---|---|
--endpoint | OTEL_EXPORTER_OTLP_ENDPOINT | endpoint | localhost:4317 |
--protocol | OTEL_EXPORTER_OTLP_PROTOCOL | protocol | http/protobuf |
--insecure | OTEL_EXPORTER_OTLP_INSECURE | insecure | false |
--timeout | OTEL_EXPORTER_OTLP_TIMEOUT | timeout | 1s |
--otlp-headers | OTEL_EXPORTER_OTLP_HEADERS | otlp_headers | k=v,a=b |
--otlp-blocking | OTEL_EXPORTER_OTLP_BLOCKING | otlp_blocking | false |
--config | OTEL_CLI_CONFIG_FILE | config_file | config.json |
--verbose | OTEL_CLI_VERBOSE | verbose | false |
--fail | OTEL_CLI_FAIL | fail | false |
--service | OTEL_CLI_SERVICE_NAME | service_name | myapp |
--kind | OTEL_CLI_TRACE_KIND | span_kind | server |
--status-code | OTEL_CLI_STATUS_CODE | span_status_code | error |
--status-description | OTEL_CLI_STATUS_DESCRIPTION | span_status_description | cancelled |
--attrs | OTEL_CLI_ATTRIBUTES | span_attributes | k=v,a=b |
--tp-required | OTEL_CLI_TRACEPARENT_REQUIRED | traceparent_required | false |
--tp-carrier | OTEL_CLI_CARRIER_FILE | traceparent_carrier_file | filename.txt |
--tp-ignore-env | OTEL_CLI_IGNORE_ENV | traceparent_ignore_env | false |
--tp-print | OTEL_CLI_PRINT_TRACEPARENT | traceparent_print | false |
--tp-export | OTEL_CLI_EXPORT_TRACEPARENT | traceparent_print_export | false |
--no-tls-verify | OTEL_CLI_NO_TLS_VERIFY | no_tls_verify | false |
Valid timeout units are "ns", "us"/"µs", "ms", "s", "m", "h".
otel-cli deviates from the OTel specification for endpoint URIs. Mainly, otel-cli supports bare host:port for grpc endpoints and continues to default to gRPC. The optional http/json is not supported by opentelemetry-go so otel-cli does not support it. To use gRPC with an http endpoint, set the protocol with --protocol or the envvar.
- bare
host:port
endpoints are assumed to be gRPC and are not supported for HTTP https://
andhttps://
are assumed to be HTTP unless --protocol is set togrpc
.- loopback addresses without an https:// prefix are assumed to be unencrypted
We want working on otel-cli to be easy, so we've provided a few different ways to get started. In general, there are three things you need:
- A working Go environment
- A built (or installed) copy of otel-cli itself
- A system to receive/inspect the traces you generate
Providing instructions on getting Go up and running on your machine is out of scope for this README. However, the good news is that it's fairly easy to do! You can follow the normal Installation instructions from the Go project itself.
If you're planning on making changes to otel-cli, we recommend building the project locally: go build
But, if you just want to quickly try out otel-cli, you can also just install it directly: go get github.com/equinix-labs/otel-cli
. This will place the command in your GOPATH
. If your GOPATH
is in your PATH
you should be all set.
otel-cli can run as a server and accept OTLP connections. It has two modes, one prints to your console while the other writes to JSON files.
otel-cli server tui
otel-cli server json --dir $dir --timeout 60 --max-spans 5
Many SaaS vendors accept OTLP these days so one option is to send directly to those. This is not recommended for production since it will slow your code down on the roundtrips. It is recommended to use an opentelemetry-collector locally.
Another option is to run the local docker compose Jaeger setup in the root of this repo with
docker-compose up
. This will bring up a stock Jaeger instance that can accept OTLP connections.
If you're not sure what to choose, try otel-cli server tui
or docker-compose up
.
Just run docker-compose up
from this repository, and you'll get an OpenTelemetry collector and a local
Jaeger all-in-one setup ready to go.
The OpenTelemetry collector is listening on localhost:4317
, and the Jaeger UI will be running on
localhost:16686
. Since these are the expected defaults of otel-cli
, you can get started with no further configuration:
docker-compose up
./otel-cli exec -n my-cool-thing -s interesting-step echo 'hello world'
This trace will be available in the Jaeger UI at localhost:16686
.
We've provided Honeycomb, LightStep, and Elastic configurations that you could also use, if you're using one of those vendors today. It's still pretty easy to get started:
# optional: to send data to an an OTLP-enabled tracing vendor, pass in your
# API auth token over an environment variable and modify
# `local/otel-vendor-config.yaml` according to the comments inside
export LIGHTSTEP_TOKEN= # Lightstep API key (otlp/1 in the yaml)
export HONEYCOMB_TEAM= # Honeycomb API key (otlp/2 in the yaml)
export HONEYCOMB_DATASET=playground # Honeycomb dataset
export ELASTIC_TOKEN= # Elastic token for the APM server.
docker run \
--env LIGHTSTEP_TOKEN \
--env HONEYCOMB_TEAM \
--env HONEYCOMB_DATASET \
--env ELASTIC_TOKEN \
--name otel-collector \
--net host \
--volume $(pwd)/local/otel-vendor-config.yaml:/local.yaml \
public.ecr.aws/aws-observability/aws-otel-collector:latest \
--config /local.yaml
Then it should just work to run otel-cli:
./otel-cli span -n "testing" -s "my first test span"
# or for quick iterations:
go run . span -n "testing" -s "my first test span"
- add some shell examples for:
- using bash trap(1p) to send events
- examples for connecting collector to other vendors' OTLP endpoints
- span background doodles: https://gist.github.com/tobert/ceb2cd9b18ab7ab09e1ea7e3bf150d9d
Please file issues and PRs on the GitHub project at https://github.com/equinix-labs/otel-cli
Releases are managed by goreleaser. Currently this is limited to @tobert due to rules in the equinix-labs organization. For now releases are not automated, but will be by the time a v1.0 rolls out and the test suite is robust enough that we feel confident.
Testing the release: goreleaser release --snapshot --rm-dist
To release, a GitHub personal access token is required. The release also needs to be tagged in git.
git checkout main # make sure we tag on main
git pull --rebase # get the latest HEAD
git tag v0.1.1 # tag HEAD with the next version
git push --tags # push new tag up to GitHub
goreleaser release --rm-dist
Apache 2.0, see LICENSE