The APM Server receives data from Elastic APM agents and transforms it into Elasticsearch documents. Read more about Elastic APM at elastic.co/apm.
For questions and feature requests, visit the discussion forum.
To get started with APM, see our Quick start guide.
- Golang 1.16.5
- Fork the repo with the GitHub interface and clone it:
cd ${GOPATH}/src/github.com/elastic/
git clone git@github.com:[USER]/apm-server.git
Note that it should be cloned from the fork (replace [USER] with your GitHub user), not from origin.
- Add the upstream remote:
git remote add elastic git@github.com:elastic/apm-server.git
To build the binary for APM Server run the command below. This will generate a binary in the same directory with the name apm-server.
make
You also need to create all files needed by the APM Server by running the additional command below.
make update
Note that this requires to have python >= 3.7 and venv installed.
To run APM Server with debugging output enabled, run:
./apm-server -c apm-server.yml -e -d "*"
For Testing check out the testing guide
Each beat has a template for the mapping in elasticsearch and a documentation for the fields
which is automatically generated based on fields.yml.
To generate required configuration files and templates run:
make update
APM-Server includes a script to generate an integration package to run with Fleet. To Generate a package run:
make fields gen-package
That command takes the existing fields.yml files and split them into ecs.yml and fields.yml files for each data stream type.
It also generates a README.md with a field reference that will be shown in the integration package.
After generating a package, apmpackage/apm should be manually copied to elastic/integrations.
Then follow instructions in https://github.com/elastic/integrations/blob/master/CONTRIBUTING.md.
To clean APM Server source code, run the following commands:
make fmt
To clean up the build directory and generated artifacts, run:
make clean
For further development, check out the beat developer guide.
See contributing for details about reporting bugs, requesting features, or contributing to APM Server.
See releases for an APM Server release checklist.
APM Server uses Go Modules for dependency management, without any vendoring.
In general, you should use standard go get commands to add and update modules. The one exception to this
is the dependency on libbeat, for which there exists a special Make target: make update-beats, described
below.
By running make update-beats the github.com/elastic/beats/vN module will be updated to the most recent
commit from the master branch, and a minimal set of files will be copied into the apm-server tree.
You can specify an alternative branch or commit by specifying the BEATS_VERSION variable, such as:
make update-beats BEATS_VERSION=7.x
make update-beats BEATS_VERSION=f240148065af94d55c5149e444482b9635801f27
It is important to keep the go-elasticsearch client in sync with the according major version. We also recommend to use the latest available client for minor versions.
You can use go get -u -m github.com/elastic/go-elasticsearch/v7@7.x to update to the latest commit on the
7.x branch.
The beats framework provides tools to cross-compile and package apm-server for different platforms. This requires docker, mage, and vendoring as described above. To build all apm-server packages from source, run:
mage package
This will fetch and create all images required for the build process.
The whole process can take several minutes.
When complete, packages can be found in build/distributions/.
To customize image configuration, see the docs.
To build docker images from source, run:
PLATFORMS=linux/amd64 mage -v package
When complete, docker images can be found through the local docker daemon and at build/distributions/apm-server-*-linux-amd64.docker.tar.gz.
When building images for testing pre-release versions, we recommend setting SNAPSHOT=true in the build environment, to
clearly indicate the packages are not for a specific release.
Documentation for the APM Server can be found in the docs folder.