Plugin(AWS): Design AWS resource groups for Nitric Stacks

[tjholm]

┆Issue is synchronized with this Jira Story by Unito

[tjholm]

Pulumi resource group documentation:
https://www.pulumi.com/docs/reference/pkg/aws/resourcegroups/group/

In essence they appear to be an index of common AWS resources.

So assuming we properly tag deployed resources we can associate them to a particular stack within an AWS account:

All resources withing a nitric stack:

const test = new aws.resourcegroups.Group("test", {
    resourceQuery: {
        query: `{
  "TagFilters": [
    {
      "Key": "x-nitric-stack",
      "Values": ["my-stack"]
    }
  ]
}
`,
    },
});

All collection (dynamodb) resources within a stack:

const test = new aws.resourcegroups.Group("test", {
    resourceQuery: {
        query: `{
  "ResourceTypeFilters": [
    "AWS::DynamoDb::Table"
  ],
  "TagFilters": [
    {
      "Key": "x-nitric-stack",
      "Values": ["my-stack"]
    }
  ]
}
`,
    },
});

[tjholm]

List of supported AWS resources:
https://docs.aws.amazon.com/ARG/latest/userguide/supported-resources.html

Note that resource groups can also be in resource groups, which allows us to get up to all sorts of shenanigans :).

[tjholm]

Initial thoughts for a basic design:

Add stages as a concept for AWS deployments (other clouds currently handle this by project segregation). The stage will be some valid string that will be appended to the stack name at deploy time that will give it a unique name. This will allow two developers for example to deploy the same stack to the same AWS account under two different stage names.

All resources are tagged with the name of the stack they were deployed for:
x-nitric-stack: ${stackname}:${stageName}.

We deploy a top level resource group that allows that aggregates stack resource groups under a single group, with a name of the stack. We could also flatten deployed resources under this group as well.

We also deploy sub resource groups that belong to the parent resource group that compose a single resource type e.g. collections
${stackName}:${stageName}-collections. This would allow all collections to be directly accessed without having to interrogate the root group.

We include and environment variable for deployed container services in AWS that contains the name of the stack (stackname + stage) to be used when querying for local resources.