8.1 KiB
Development
Prerequisites
Developing Hive locally requires you to have the following software installed locally:
- Node.js 21 (or
nvmorfnm) - pnpm v9
- Docker version 26.1.1 or later(previous versions will not work correctly on arm64)
- make sure these ports are free: 5432, 6379, 9000, 9001, 8123, 9092, 8081, 8082, 9644, 3567, 7043
Setup Instructions
- Clone the repository locally
- Make sure to install the recommended VSCode extensions (defined in
.vscode/extensions.json) - In the root of the repo, run
nvm useto use the same version of node as mentioned - Create
.envfile in the root, and use the following:
ENVIRONMENT=local
- Run
pnpm iat the root to install all the dependencies and run the hooks - Run
pnpm local:setupto run Docker compose dependencies, create databases and migrate database
Solving permission problems on this step:
export UID=$(id -u)
export GID=$(id -g)
Add "user" field to docker-compose.dev.yml
clickhouse:
user: '${UID}:${GID}'
db:
user: '${UID}:${GID}'
- Run
pnpm generateto generate the typings from the graphql files (usepnpm graphql:generateif you only need to run GraphQL Codegen) - Run
pnpm buildto build all services - Click on
Start Hivein the bottom bar of VSCode - Open the UI (
http://localhost:3000by default) and Sign in with any of the identity provider - Once this is done, you should be able to log in and use the project
- Once you generate the token against your organization/personal account in hive, the same can be
added locally to
hive.jsonwithinpackages/libraries/cliwhich can be used to interact via the hive cli with the registry (Usehttp://localhost:3001/graphqlas theregistry.endpointvalue inhive.json) - Now you can use Hive locally. All other steps in this document are optional and only necessary if you work on specific features.
Development Seed
We have a script to feed your local instance of Hive with initial seed data. This step is optional.
- Use
Start Hiveto run your local Hive instance - Make sure
usageandusage-ingestorare running as well (withpnpm dev) - Open Hive app, create a project and a target, then create a token
- Run the seed script:
TOKEN="MY_TOKEN_HERE" pnpm seed - This should report a dummy schema and some dummy usage data to your local instance of Hive, allowing you to test features e2e
Note: You can set
STAGING=1in order to target staging env and seed a target there. Same for development env, you can useDEV=1
Note: You can set
FEDERATION=1in order to publish multiple subgraphs.
To send more operations and test heavy load on Hive instance, you can also set
OPERATIONS(amount of operations in each interval round, default is1) andINTERVAL(frequency of sending operations, default:1000ms). For example, usingINTERVAL=1000 OPERATIONS=1000will send 1000 requests per second.
Troubleshooting
We recommend the following flow if you are having issues with running Hive locally:
- Stop all Docker containers:
docker kill $(docker ps -q) - Clear all local Docker environment:
docker system prune --all --force --volumes - Delete all generated local
.envfiles:find . -name '.env' | xargs rm - Delete local
docker/.hiveanddocker/.hive-devdir used by Docker volumes. - Reinstall dependencies using
pnpm install - Force-generate new
.envfiles:pnpm env:sync --force
Publish your first schema (manually)
-
Start Hive locally
-
Create a project and a target
-
Create a token from that target
-
Go to
packages/libraries/cliand runpnpm build -
Inside
packages/libraries/cli, run:pnpm start schema:publish --registry.accessToken "YOUR_TOKEN_HERE" --registry.endpoint "http://localhost:3001/graphql" examples/single.graphqlThe registry endpoint is the GraphQL endpoint of your local
serverservice. You can also edit thehive.jsonfile in theclipackage to avoid passing theaccessTokenandendpointevery time.
Setting up Slack App for developing
- Download Loophole CLI (same as ngrok but supports non-random urls)
- Log in to Loophole
$ loophole account login - Start the proxy by running
$ loophole http 3000 --hostname hive-<your-name>(@kamilkisiela I usehive-kamil). It createshttps://hive-<your-name>.loophole.siteendpoint. - Message @kamilkisiela and send him the url (He will update the list of accepted redirect urls in Slack App).
- Update
APP_BASE_URLinpackages/web/app/.envto the proxy URL (e.g.https://hive-<your-name>.loophole.site) - Run
packages/web/appand openhttps://hive-<your-name>.loophole.site.
We have a special Slack channel called
#hive-teststo not spam people :)
Setting up GitHub App for developing
Starting a proxy for GitHub App
- Follow first two steps from
Setting up Slack App for developing(download loophole and log in). - Start web app proxy:
$ loophole http 3000 --hostname hive-<your-name> - Start server proxy:
$ loophole http 3001 --hostname hive-<your-name>
Creating a GitHub App
- Go to
Settings->Developer settings->GitHub Apps, and click on theNew GitHub Appbutton. - Provide a name for your app, and set the
Homepage URLtohttps://hive-<your-name>.loophole.site. Then set the callback URL tohttps://hive-<your-name>.loophole.site/api/github/callback, and post installation's callback URL tohttps://hive-<your-name>.loophole.site/api/github/setup-callback. - Click on
Create GitHub App.
Setting up env variables
-
Server: Set the following env variables in
packages/services/server/.env:INTEGRATION_GITHUB=1 INTEGRATION_GITHUB_GITHUB_APP_ID=<your-github-app-id>You'll find the GitHub App ID and private key in the
Generaltab of your GitHub App.Store the Github private key next to the
.envfile with the namegithub-app.pem(packages/services/server/github-app.pem) -
Web App: Set the following in
packages/web/app/.env:INTEGRATION_GITHUB_APP_NAME=<your-github-app-name>
Installing the GitHub App
Open Hive UI and go to your organization's settings page. Find Integrations section and click on
Connect GitHub. You should be redirected to GitHub where you can grant repository access. After
installing the app, you should be redirected back to Hive.
Testing
- Create a project and a target.
- Create a token from that target.
- Setup a GitHub repo with CI/CD actions like this one: https://github.com/n1ru4l/hive-federation-subgraph/.
- Add the token to the repo's secrets as
HIVE_TOKEN. - Add Hive endpoint to the repo's secrets as
HIVE_ENDPOINT(https://hive-<your-name>.loophole.site/graphql). - Make sure your GitHub app is installed on that repo.
- Push a commit to the repo and check if the CI/CD action is triggered.
Local OIDC Testing
The docker-compose.dev.yml files includes a mock OIDC server that can be used for testing the OIDC
login/logout flow locally. The server tuns on port 7043.
Please make sure to set the AUTH_ORGANIZATION_OIDC environment variables for the server and
app to "1".
You can use the following values for connecting an integration to an OIDC provider.
# Token Endpoint
http://localhost:7043/connect/token
# User Info Endpoint
http://localhost:7043/connect/userinfo
# Authorization Endpoint
http://localhost:7043/connect/authorize
# Client ID
implicit-mock-client
# Client Secret
client-credentials-mock-client-secret
For login use the following credentials.
# Username
test-user
# Password
password
Run Hive
- Click on Start Hive in the bottom bar of VSCode
- Open the UI (
http://localhost:3000by default) and register any email and password - Sending e-mails is mocked out during local development, so in order to verify the account find
the verification link by visiting the email server's
/_historyendpoint -http://localhost:6260/_historyby default.- Searching for
tokenshould help you find the link.
- Searching for