Note: You will need to be inside the folder
serverof the respective API when running the commands mentioned below.
nodemon server.js
npm test
Note Before testing you will need to have a local MySQL and redis server running. The MySQL server needs to run on port
3306and have arootuser with the passwordpassword. Also, the MySQL database will need to contain the tables found invr-collaboration-room-db-dev.sql. The redis server needs to run on port6379.
/auth/login/auth/register/auth/auth_check/cadmodels/cadmodels/signedurl/cadmodels/pub-sub/sessions/sessions/{sessionId}/participants/sessions/{sessionId}/invited
To host our MySQL database we used a Google Cloud service called Cloud SQL. This service helps us to ensure an extra layer of security, since only the microservices running on Cloud Run that have a connection to it, will be able to access its data. Likewise, it provides scalability, since its capacity can be automatically adjusted based on usage.
All sensitive data for the database is stored as GitHub secrets. To find the values of these you will need to check the info about the Cloud SQL instance on Google Cloud.
Note: The host of the MySQL is set to the value under
Private IP address. Do not use the public one!
Without enabling Public IP under the settings of the Cloud SQL instance, you will not be able to access the server without it happening through a service running on Cloud Run.
Note: The service running on Cloud Run will not be able to access the MySQL server unless it is connected to it! This happens in the GitHub Action workflow-file, but it can be done manually also (
EDIT & DEPLOY NEW REVISION<CONNECTIONS<Cloud SQL connections).
When Public IP is enabled you will need to select your current IP Address. When your IP has been added, you can access the server through the Cloud Shell by using the following command: gcloud sql connect <NAME-OF-CLOUD-SQL-INSTANCE> -u <USERNAME-OF-THE-CLOUD-SQL-INSTANCE>
To make a connection to the Cloud SQL instance from Cloud Run you will need to make sure that they are both in the same region (e.g. us-west1). In addition, they will need to run on the same network in Google Cloud. In most cases this network will be called default.
We use the Google Cloud service Memorystore to store the tokens that are used to authorise users that want to access specific resources. Unlike Cloud SQL that uses an SSD (or HDD) to store data, redis on Memorystore is placed in the cache. This ensures super-fast read and write operations, making it ideal for temporary data that is frequently accessed. Much like Cloud SQL, Memorystore increases security and ensures scalability.
All sensitive data for Redis is stored as GitHub secrets. To find the values of these you will need to check the info about the Memorystore (Redis) instance on Google Cloud.
Note: The host of the Memorystore (Redis) instance is set to the value under
IP address.
To be able to access the instance from a service running on Cloud Run, you will need to create a VCP connector first (VCP network < Serverless VPC access) and then add it to the Cloud Run instance. This happens in the in the GitHub Action workflow file, but it can also be done manually.
Note: The Memorystore instance will need to be connected to the same network as the VCP connector (
Connection properties<Authorized network). In most cases this network is calleddefault.
Cloud Run is a service on Google Cloud Platform (GCP) that we have used to host our microservices. Using this service, many requests can be sent to each service at a time, and they will be automatically scaled based on the number of requests received.
Each microservice is automatically deployed to Cloud Run on either a push to the development or master branch. The are all configured to be private (no-allow-unauthenticated), use the same VCP connector as the redis database (vpc-connector), have a connection to the Cloud SQL instance (add-cloudsql-instances) and to use our GitHub secrets (set-env-vars). To retrieve these variables being set we use the following syntax in Node.js const MYSQL_HOST = process.env.MYSQL_HOST;.
Note: The API gateway is the only service that is not set to private (
allow-unauthenticated). This is because it routes all traffic between the client and the microservices, and hence it must be available without authentication.
The bucket user-cad-models is configured to publish details about new CAD models being uploaded to it to the topic user-cad-models under Pub/Sub > Topics on Google Cloud. This topic has a subscription named cad-microservice-subscription that will send a POST request to the endpoint /cadmodels/pub-sub with info about which model was uploaded/changed, at what time it happened, etc. In short, the bucket user-cad-models is the publisher and the endpoint /cadmodels/pub-sub the subscriber.
The following part of Google Cloud's documentation was used to create the topic and subscription mentioned above: Create Subscription / Topic
This link was used to configure Pub/Sub for Cloud Storage for the user-cad-models bucket: Pub/Sub Cloud Storage Config