Skip to content

Latest commit

 

History

History
39 lines (34 loc) · 6.06 KB

README.md

File metadata and controls

39 lines (34 loc) · 6.06 KB
Raw

Overview

The FDA MyStudies Response datastore provides APIs to store and manage pseudonymized study response and activity data. The Response datastore receives de-identified study responses from the Android and iOS mobile applications, which are then written to a document-oriented database. Researchers and data scientists access this database to perform analysis of the study data, or for export to downstream systems. The Response datastore also receives de-identified activity data, which is written to a relational database. The application is built as a Spring Boot application. The backend database is Cloud Firestore for the response data, and MySQL for the activity data. The Response datastore uses basic authentication client_id and client_secret.

The Response datastore client applications are:

  1. Android mobile application
  2. iOS mobile application
  3. Study builder
  4. Participant datastore

Deployment

NOTE: Holistic deployment of the FDA MyStudies platform with Terraform and infrastructure-as-code is the recommended approach to deploying this component. A step-by-step guide to semi-automated deployment can be found in the deployment/ directory. The following instructions are provided in case manual deployment in a VM is required. Google Cloud infrastructure is indicated, but equivalent alternative infrastructure can be used as well. It is important for the deploying organization to consider the identity and access control choices made when configuring the selected services. If pursuing a manual deployment, a convenient sequence is hydra/→auth-server/→participant-datastore/→participant-manager-datastore/→participant-manager/→study-datastore/→response-datastore/→study-builder/→Android/→iOS/.

To deploy the Response datastore manually:

  1. Create a Compute Engine VM instance with a static IP and read/write access scopes for Cloud Datastore
  2. Verify that your VM instance has the Stackdriver Logging API write access scope (on by default) and that your VM’s service account has the Logs Writer role (off by default)
  3. Check out the latest code from the FDA MyStudies repository
  4. Create a Cloud SQL instance with MySQL v5.7 (instructions)
  5. Configure the Response datastore database on the Cloud SQL instance
  6. Create a Cloud Firestore database operating in Native mode, then grant the IAM role roles/datastore.user to the service account that your Response datastore will use to read/write data (this could be your VM’s default service account)
  7. Deploy the Response datastore container to the VM
    • Create the Docker image using sudo mvn -B package -Pprod com.google.cloud.tools:jib-maven-plugin:2.5.2:dockerBuild -Dimage=response-datastore-image from the response-datastore/ directory (you may need to install Docker and Maven, for example sudo apt install maven)
    • Update the Docker environment file variables.env with values that configure the application.properties file for your deployment
    • Run the container on the VM using sudo docker run --detach --env-file variables.env -p 80:8080 --name response-datastore response-datastore-image
    • If your Hydra instance is a using self-signed certificate, add that certificate to your container’s keystore, for example with sudo docker exec -it response-datastore bash -c "openssl s_client -connect <your_hydra_instance> | sed -ne '/-BEGIN CERTIFICATE/,/END CERTIFICATE/p' > hydra.crt; keytool -import -trustcacerts -alias hydra -file hydra.crt -keystore /usr/local/openjdk-11/lib/security/cacerts -storepass changeit", then restart the container with sudo docker restart response-datastore
  8. Test if the application is running with curl http://0.0.0.0/response-datastore/healthCheck
  9. You can review application logs in the logging directories you specified, or with sudo docker logs response-datastore; audit logs are available in Cloud Logging

Copyright 2020 Google LLC