server

Server deployment using Cloud Run

This directory contains the server component of the analysis runner, which is deployed as a Cloud Run container.

It uses aiohttp instead of flask because Hail Batch's async event queue doesn't mix well with flask's threads.

Each dataset / storage stack has its own permissions groups, which are checked through a single analysis-runner instance.

Deployment happens continuously using the hail_update workflow.

Hail service account tokens need to be copied to a Secret Manager secret separately, after the stacks have been set up.

The Cloud Run HTTPS deployment endpoint is hardcoded in the CLI tool.

Testing locally

See Testing the Container Image Locally for details.

Download a JSON key for the analysis-runner-server service account. Store the file name in the $GSA_KEY_FILE environment variable. Then run:

docker build -t analysis-runner-server .

docker run -it -p 8080:8080 -v $GSA_KEY_FILE:/gsa-key/key.json -e GOOGLE_APPLICATION_CREDENTIALS=/gsa-key/key.json -e DRIVER_IMAGE=$DRIVER_IMAGE analysis-runner-server

This will start a server that listens locally on port 8080.

From another terminal, send a request like this, replacing the JSON parameters accordingly.

TOKEN=$(gcloud auth print-identity-token) curl -X POST -H "Authorization: Bearer $TOKEN" -H "Content-Type:application/json" -d '{"output": "gs://test-bucket/test", "dataset": "fewgenomes", "accessLevel": "test", "repo": "hail-batch-test", "commit": "0fa3abfe59692618578c4e1551b2a9357566d2ad", "script": ["main.py"], "description": "test"}' localhost:8080