Spruce ·
Spruce is the React UI for MongoDB's continuous integration software.
- Clone the Spruce GitHub repository
- Ensure you have Node.js v20+ and MongoDB Command Line Database Tools v100.8.0+ installed
- Run
yarn install
- Start a local Evergreen server by doing the following:
- Clone the Evergreen repo
- From the Evergreen directory, run
make local-evergreen
- Run
yarn run dev
. This will launch the app and point it at the local Evergreen server you just started.
Run yarn run storybook
to launch storybook and view our shared components.
Install the Prettier code formatting plugin in your code editor if you don't have it already. The plugin will use the .prettierrc settings file found at the root of Spruce to format your code.
Follow these directions to enable query linting during local development so your Evergreen GraphQL schema changes are reflected in your Spruce query linting results.
- Symlink the standard definition language GraphQL schema used in your backend
to a file named sdlschema in the root of the Spruce directory to enable query
linting with ESLint like so
ln -s <path_to_evergreen_repo>/graphql/schema sdlschema
- Run
yarn run eslint
to see the results of query linting in your terminal or install a plugin to integrate ESlint into your editor. If you are using VS Code, we recommend ESLint by Dirk Baeumer.
Read more about environment variables here.
We use code generation to generate our types for our GraphQL queries and
mutations. When you create a query or mutation you can run the code generation
script with the steps below. The types for your query/mutation response and
variables will be generated and saved to gql/generated/types.ts
. Much of the
underlying types for subfields in your queries will likely be generated there as
well and you can refer to those before creating your own.
- Create a symlink from the
schema
folder from Evergreen to Spruce using
ln -s <path_to_evergreen_repo>/graphql/schema sdlschema
- From within the Spruce folder run
yarn codegen
- As long as your queries are declared correctly the types should generate
- Queries should be declared with a query name so the code generation knows what to name the corresponding type.
- Each query and mutation should have a unique name.
- Since query analysis for type generation occurs statically we can't place dynamic variables with in query strings. We instead have to hard code the variable in the query or pass it in as query variable.
- Sometimes you may run into an error where a dependency is out of date or in a
broken state. If you run into this issue try running
yarn install
to reinstall all dependencies. If that does not work try deleting yournode_modules
folder and runningyarn install
again. You can use theyarn clean
command to do this for you.
Spruce has a combination of unit tests using Vitest and integration tests using Cypress.
Unit tests are used to test individual features in isolation. We utilize Vitest to execute our unit tests and generate reports.
There are 3 types of unit tests you may encounter in this codebase.
These test React components. We utilize
React Testing Library
to help us write our component tests. React Testing Library provides several
utilities that are useful for making assertions on React Componenents. When
writing component tests you should import
test_utils
instead of React Testing Library; test_utils
is a wrapper around React Testing
Library which provides a series of helpful utilities for common testing
scenarios such as queryByDataCy
, which is a helper for selecting data-cy
attributes, or renderWithRouterMatch
, which is helpful for testing components
that rely on React Router.
Often times you may find yourself writing
custom React hooks. The best way
to test these is using React Testing Library's
renderHook
utility. This allows you to test your custom hooks in isolation without needing
to wrap them in a component. It provides several methods that make it easy to
assert and test different behaviors in your hooks. Such as
waitFor
,
which will wait for your hook to rerender before allowing a test to proceed.
These are the most basic of tests. They do not require any special libraries to run and often just test standard JavaScript functions.
- You can run all unit tests using
yarn test run
- You can run a specific unit test using
yarn test run <test_name>
- You can run Vitest in watch mode using
yarn test
. This will open an interactive CLI that can be used to automatically run tests as you update them.
At a high level, we use Cypress to start a virtual browser that is running Spruce. Cypress then is able to run our test specs, which tell it to interact with the browser in certain ways and makes assertions about what happens in the UI. Note that you must be running the Evergreen server on http://localhost:9090 for the front-end to work.
In order to run the Cypress tests, do the following, assuming you have this repo checked out and all the dependencies installed by Yarn:
- Increase the limit on open files by running
ulimit -n 64000
before running mongod in the same shell. - Start the evergreen back-end with the sample local test data. You can do this
by typing
make local-evergreen
in your evergreen folder. - Start the Spruce local server by typing
yarn build:local && yarn serve
in this repo. - Run Cypress by typing one of the following:
yarn cy:open
- opens the Cypress app in interactive mode. You can select tests to run from here in the Cypress browser.yarn cy:run
- runs all the Cypress tests at the command-line and reports the resultsyarn cy:test cypress/integration/hosts/hosts-filtering.ts
- runs tests in a specific file at the command-line. Replace the final argument with the relative path to your test file
Snapshot tests are automatically generated when we create Storybook stories.
These tests create a snapshot of the UI and compare them to previous snapshots
which are stored as files along side your Storybook stories in a __snapshots__
directory. They try to catch unexpected UI regressions. Read more about them
here.
If you need more data to be able to test out your feature locally, the easiest way to do so is to populate the local db using real data from the staging or production environments.
-
You should identify if the data you need is located in the staging or prod db and connect to them using the instructions in the Evergreen Operations Guide
-
Identify the query you need to fetch the data you are looking for.
Atlas atlas-mxabkq-shard-0 [primary] mci> db.distro.find({_id: "archlinux-small"}) // the full query
-
Write the file to your local system using the following command
Atlas atlas-mxabkq-shard-0 [primary] mci> fs.writeFileSync('output.json', JSON.stringify(db.distro.find({_id: "archlinux-small"})))
A file will be saved to your filesystem with the results of the query.
-
Exit the ssh session using
exit
orCtrl + D
-
You should ensure this file does not contain any sensitive information before committing it to the repository.
-
Once you have this file you can copy the contents of it to the relevant
testdata/local/<collection>.json
file with in the evergreen folder -
You can then run
yarn evg-db-ops --reseed
to repopulate the local database with your new data.
Notes
When creating your queries you should be sure to limit the amount of documents so you don't accidentally export an entire collection. You can do this by passing a limit to the query.
Spruce has a minimal dependency on Logkeeper: it is used by Cypress tests on the Job Logs page. If you'd like to get set up to develop these tests, complete the following:
-
Clone the Logkeeper Repository
-
Run
yarn bootstrap-s3-logs
to download some sample resmoke logs from s3. -
Run the command outputted by the previous step to seed the env variables and start the local logkeeper server with the following command:
LK_CORS_ORIGINS=http:\/\/localhost:\\d+ LK_EVERGREEN_ORIGIN=http://localhost:8080 LK_PARSLEY_ORIGIN=http://localhost:5173 go run main/logkeeper.go --localPath {abs_path_to_spruce}/bin/_bucketdata
Read more about deployment here.