This Repository is NOT a supported MongoDB product

drivers-evergreen-tools

This repo is meant for MongoDB drivers to bootstrap their Evergreen configuration files. It contains set of scripts for tasks that most drivers will need to perform, such as downloading MongoDB for the current platform, and launching various topologies.

Using In Evergreen

The bundled .evergreen/config.yml file contains a suggested template for drivers to use. This file can either be taken and used as is, or used as a recipe to copy&paste from.

It is recommended to copy the entire directory and modify the following scripts:

• install-dependencies.sh - Install any platform dependencies not currently available on Evergreen
• run-tests.sh - How to run the test suite
• compile.sh - Configure any alternative environment routes
• compile-unix.sh - Instructions to run the compile stage on *nix
• compile-windows.sh - Instructions to run the compile stage on Windows
• make-docs.sh - Instructions how to compile the driver docs
• make-release.sh - Instructions how to package and release the driver

The normal matrix (e.g. all tasks with the exception on those in the ** Release Archive Creator buildvariant) runs the following two shell scripts:

• The install-dependencies.sh file is always executed by all tasks.
• The run-tests.sh is run by all tasks, except for the ** Release Archive Creator.

The ** Release Archive Creator buildvariant is special, and does not run the "standard test matrix", but in stead runs the following:

• The compile*.sh is executed by the release-compile and release-compile-cmake tasks. These are no commonly used by drivers, so feel free to ignore.
• The make-docs.sh is executed by the make-docs task
• The make-release.sh is executed by the make-release-archive task

See also: https://evergreen.mongodb.com/waterfall/drivers-tools

Using With GitHub Actions

This repository includes a metadata file for GitHub Actions to allow downloading MongoDB and launching topologies from a GitHub Action Workflow. To use this action, use the following template:

    steps:
      # [...]
      - id: setup-mongodb
        uses: mongodb-labs/drivers-evergreen-tools@master
        # Set configuration
        with:
          version: ${{ matrix.mongodb-version }}

The following inputs exist:

Name	Description
version	MongoDB version to install
topology	Topology of the deployment (server, replica_set, sharded_cluster)
auth	Whether to enable auth
ssl	Whether to enable SSL
storage-engine	Storage engine to use
require-api-version	Whether to start the server with requireApiVersion enabled (defaults to false)

These correspond to the respective environment variables that are passed to run-orchestration.sh.

After the cluster is started, its URI is exposed via the cluster-uri output. In addition, the action also exposes the path to crypt_shared via the crypt-shared-lib-path output, unless the installation was not requested or failed. This configuration snippet environment variables with the cluster URI and crypt_shared lib path returned from the setup-mongodb workflow step when running tests:

    steps:
      # [...]
      - name: "Run Tests"
        run: "run-tests.sh"
        env:
          MONGODB_URI: ${{ steps.setup-mongodb.outputs.cluster-uri }}
          CRYPT_SHARED_LIB_PATH: ${{ steps.setup-mongodb.outputs.crypt-shared-lib-path }}

Starting and Stopping Servers Locally or on an Evergreen Host

There are two options for running a MongoDB server configuration. One is to use docker. The other is to run ./evergreen/run-orchestration.sh locally. For convenience, you can run make run-server and make stop-server to start and stop the server(s). For example:

TOPOLOGY=replica_set MONGODB_VERSION=7.0 make run-server

See (run-orchestration.sh)[./evergreen/run-orchestration.sh] for the available environment variables.

Linters and Formatters

This repo uses pre-commit for managing linting and formatting of the codebase. pre-commit performs various checks on all files in the repo and uses tools that help follow a consistent code style.

To set up pre-commit locally, run:

brew install pre-commit
pre-commit install

To run pre-commit manually, run:

pre-commit run --all-files

To run an individual hook like shellcheck manually, run:

pre-commit run --all-files shellcheck

Setup and Teardown

For convenience, there are two scripts that can be used for setup and teardown of assets and services used by drivers-evergreen-tools.

$DRIVERS_TOOLS/.evergreen/setup-test.sh will handle common setup actions that have previously been spread through multiple pre: tasks in drivers.

$DRIVERS_TOOLS/.evergreen/teardown-test.sh will clean up common assets and services. This script will also collect all logs files recursively found in the ${DRIVERS_TOOLS} directory into a single ${DRIVERS_TOOLS}/.evergreen/test_logs.tar.gz file for convenient uploading.

Subfolders that have setup and teardown requirements are encouraged to also provide setup-test.sh and teardown-test.sh.

NOTE: The subfolder setup/teardown scripts requires users to have configured support for Secrets Handling.

Secrets Handling

See the Secrets Handling readme for more information on how secrets are managed locally and on on Evergreen.

Python CLIs

We make some of our Python scripts available as self-contained clis that do not require setting up a Python virtual environment. For example, after running .evergreen/setup.sh you can run the resulting .evergreen/mongodl, .evergreen/mongosh-dl and .evergreen/socks5srv scripts directly.

Env Files

This repo supports the use of .env files, which can be placed in $DRIVERS_TOOLS and in the sub-directories. The script will first read $DRIVERS_TOOLS/.env if it exists, and then $SCRIPT_DIR/.env if it exists, to give the local file higher precedence. This pattern can be used to replace the use of legacy ${PREPARE_SHELL} invocations in a shell.exec Evergreen command, enabling the use of subprocess.exec instead.

evergreen_config_generator

This repo also contains a Python package for use in scripts that generate Evergreen config files from Python dicts. See evergreen_config_generator/README.