This guide will give instructions for configuring and setting up the project on the Google Cloud Platform (GCP) before the installation of the platform.
Currently, the platform uses Linux and Windows GCP instances for running regression tests.
Apart from the central platform server, the platform utilises Infrastructure as a Service (IaaS) through the Google Cloud Platform.
For this create a GCP account or log in to an existing account here.
The next step is to create a Google Cloud Project that the platform would use to manage instances for running regression tests and maintain a GCS bucket for samples, variables, tester, etc.
For creating and managing projects, refer to the official documentation here.
Now create a project, with the project name and project id of your choice.
Working on Google Cloud Platform is not free, therefore it is recommended to either get a free 90-day trial or enable billing for the GCP account to get access the full set of services and increased usage limits.
Now create a GCS bucket with the following settings (refer to the official documentation here):
- Location: Choose as per your preference and budget, however it is advised to keep the bucket and sample-platform VM instance in the same region for faster access.
- Storage Class: Standard
- Access Control: Uniform, and enable
Prevent Public Access - Protection Tools: None
Note: Sometimes while creating an instance through the platform one might receive an error(ZONE_RESOURCE_POOL_EXHAUSTED or ZONE_RESOURCE_POOL_EXHAUSTED_WITH_DETAILS), this is an issue faced quite often, hence it is advised to choose region and zone wisely. Here is a stackoverflow link for the same error.
Note: We will provide downloads of the samples to users when requested via Signed URLs.
Now, create a service account with sufficient permissions (at least "Google Batch Service Agent") that would be used by the platform, VM instances to mount and read-write the bucket contents and also manage VM instances.
If you are not the owner of the GCP project you are working on, make sure you have sufficient permissions for creating and managing service accounts; if not, request the project owner for the same.
- Create a service account here
- Choose the service account name as per your choice, but at least provide the role of "Google Batch Service Agent" to the account.
You might also want to understand roles in GCP, you can find the official documentation here.
Now navigate to the "keys" section of the service account created, create a new key and download the JSON file.
Note: This is a secret key and access to the key would give access to your GCP project.
Note: If you have provided the "editor" role to the service account (not recommended), skip this step.
If you are not the owner of the GCP project you are working on, make sure you have the "Service Account Admin" role; if not, request the project owner for the same.
- Go to cloud storage page, and select the bucket to be used for the platform.
- Now go to the "Permissions" tab and check that the service account you created has "Storage Legacy Bucket Owner" and "Storage Legacy Object Owner" permissions, if not add these permissions by creating a new principal:
- Give the new principal name as the email of the service account created (can be found in the IAM & Admin -> Service Accounts section).
- Select the roles mentioned above and click on "Save".
To reduce the required size of the VM instances, a network mount will be used to make the data accessible to the VM instances.
This network mount is done using:
- Google Cloud Storage FUSE for linux VM instances.
- Rclone and WinFsp for windows VM instances.
These tools are downloaded and installed automatically using the startup scripts on the VM instances.