This directory contains some documentation about the Nomad codebase, aimed at readers who are interested in making code contributions.
If you're looking for information on using Nomad, please instead refer to the Nomad website.
A development environment is supplied via Vagrant to make getting started easier.
-
Install Vagrant
-
Install Virtualbox
-
Bring up the Vagrant project
$ git clone https://github.com/hashicorp/nomad.git $ cd nomad $ vagrant up
The virtual machine will launch, and a provisioning script will install the needed dependencies within the VM.
-
SSH into the VM
$ vagrant ssh
- Install Go 1.22.4+ (Note:
gcc-go
is not supported) - Clone this repo
$ git clone https://github.com/hashicorp/nomad.git $ cd nomad
- Bootstrap your environment
$ make bootstrap
- (Optionally) Set a higher ulimit, as Nomad creates many file handles during normal operations
$ [ "$(ulimit -n)" -lt 1024 ] && ulimit -n 1024
- Verify you can run smoke tests
Note: You can think of this as a
$ make test
smoke
test which runs a subset of tests and some may fail because ofoperation not permitted
error which requiresroot
access. You can usego test
to test the specific subsystem which you are working on and let the CI run rest of the tests for you.
- Compile a development binary (see the UI README to include the web UI in the binary)
$ make dev # find the built binary at ./bin/nomad
- Start the agent in dev mode
$ sudo bin/nomad agent -dev
- (Optionally) Run Consul to enable service discovery and health checks
- Download Consul
- Start Consul in dev mode
$ consul agent -dev
If in the course of your development you change a Protobuf file (those ending in .proto), you'll need to recompile the protos.
- Run
make boostrap
to install thebuf
command. - Compile Protobufs
$ make proto
See the UI README for instructions.
To create a release binary:
$ make prerelease
$ make release
$ ls ./pkg
This will generate all the static assets, compile Nomad for multiple
platforms and place the resulting binaries into the ./pkg
directory.
Only the api/
and plugins/
packages are intended to be imported by other projects. The root Nomad module does not follow semver and is not intended to be imported directly by other projects.
The code for Nomad's major components is organized as:
api/
provides a Go client for Nomad's HTTP API.client/
contains Nomad's client agent code.command/
contains Nomad's CLI code.nomad/
contains Nomad's server agent code.ui/
contains Nomad's UI code.website/
contains Nomad's website and documentation.
The high level control flow for many Nomad actions (via the CLI or UI) are:
# Read actions:
Client -> HTTP API -> RPC -> StateStore
# Actions that change state:
Client -> HTTP API -> RPC -> Raft -> FSM -> StateStore
When adding new features to Nomad there are often many places to make changes. It is difficult to determine where changes must be made and easy to make mistakes.
The following checklists are meant to be copied and pasted into PRs to give developers and reviewers confidence that the proper changes have been made: