Skip to main content

Research Notes on running integration tests in bamboo CI/CD

Deploying terraform modules via bamboo​

AWS resources can be deployed via Bamboo pipeline by first adding the following to a script and running that in a Bamboo task.

#configure aws
export AWS_ACCESS_KEY_ID=$bamboo_AWS_ACCESS_KEY_ID
export AWS_SECRET_ACCESS_KEY=$bamboo_AWS_SECRET_ACCESS_KEY
export AWS_DEFAULT_REGION=$bamboo_AWS_DEFAULT_REGION
note

In order to use a user-defined bamboo environment variable inside a script, $bamboo_ must be added as a prefix to the variable. For example, use $bamboo_AWS_DEFAULT_REGION in your script to use the variable AWS_DEFAULT_REGION.

Details of some of the initial work done are added in the ORCA-test-bamboo branch of cumulus-orca repo. The bamboo spec file has been modified to add two new stages named Deploy Dev RDS Stack stage which deploys the RDS cluster in sandbox and the Deploy Dev Cumulus and ORCA Stack stage which deploys the data persistence module as well as cumulus and orca modules. The stages in bamboo spec are shown below.

#deploy RDS cluster for integration test
- Deploy Dev RDS Stack:
    manual: true
    final: false
    jobs:
    - Deploy RDS cluster

#deploy orca and cumulus for integration test
- Deploy Dev Cumulus and ORCA Stack:
    manual: true
    final: false
    jobs:
    - Deploy cumulus and orca

The deployment-rds.sh script is an initial working script that was used to deploy the RDS cluster. The deployment-cumulus-orca.sh script was used to deploy the data-persistence-tf module. Some changes still need to be made in these scripts to make it more functional and handle errors. A successful bamboo build can be seen here. A snippet of the script to deploy RDS cluster via terraform is shown below.

# Deploy drds-cluster-tf via terraform
echo "Deploying rds-cluster-tf  module to $bamboo_DEPLOYMENT"
terraform apply \
  -auto-approve \
  -lock=false \
  -input=false \
  -var-file="terraform.tfvars" \
  -var "prefix=$bamboo_PREFIX" \
  -var "region=$bamboo_AWS_DEFAULT_REGION" \
  -var "subnets=[\"$bamboo_AWS_SUBNET_ID1\", \"$bamboo_AWS_SUBNET_ID2\"]" \
  -var "db_admin_username=$bamboo_DB_ADMIN_USERNAME" \
  -var "db_admin_password=$bamboo_DB_ADMIN_PASSWORD" \
  -var "vpc_id=$bamboo_VPC_ID" \
  -var "cluster_identifier=$bamboo_RDS_CLUSTER_ID" \
  -var "deletion_protection=false"\
  -var "provision_user_database=false"\
  -var "engine_version=10.14"\
  -var "permissions_boundary_arn=arn:aws:iam::$bamboo_AWS_ACCOUNT_ID:policy/$bamboo_ROLE_BOUNDARY"

Variables in terraform.tfvars can be overwritten by the -var field as seen above.

Some of the sensitive bamboo variables such as AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, have been encrypted using bamboo encryption service. While running the pipeline, those variables have to be replaced manually with the real values since bamboo does not automatically decrypt the values while running the pipeline. Make sure all sensitive variables are encrypted before pushing the changes to the repo.

The bamboo spec for running the two scripts under tasks is shown below.

#job for deploying cumulus and orca
Deploy RDS cluster:
  key: DRC
  other:
    clean-working-dir: true
    # Some plugin configurations are not supported by YAML Specs
  docker:
    image: amazonlinux:2
    volumes:
      ${bamboo.working.directory}: ${bamboo.working.directory}
      ${bamboo.tmp.directory}: ${bamboo.tmp.directory}
  tasks:
  - checkout:
      force-clean-build: 'true'
      description: Checkout Default Repository

  - script:
      interpreter: SHELL
      scripts:
      - |-
        #!/bin/bash
        chmod +x bin/deployment-rds.sh
        bin/deployment-rds.sh
#job for deploying cumulus and orca
Deploy cumulus and orca:
  key: DCO
  other:
    clean-working-dir: true
    # Some plugin configurations are not supported by YAML Specs
  docker:
    image: amazonlinux:2
    volumes:
      ${bamboo.working.directory}: ${bamboo.working.directory}
      ${bamboo.tmp.directory}: ${bamboo.tmp.directory}
  tasks:
  - checkout:
      force-clean-build: 'true'
      description: Checkout Default Repository

  - script:
      interpreter: SHELL
      scripts:
      - |-
        #!/bin/bash
        chmod +x bin/deployment-cumulus-orca.sh
        bin/deployment-cumulus-orca.sh

Future directions​

The first step will be to finish deploying all cumulus tf modules via Bamboo into sandbox account. Once that is completed, automation scripts for running the integrations tests need to be created. The original plan ORCA Integrator in bamboo will need to be updated with the same changes made in prototype-demo plan's feature/ORCA-test-bamboo branch so that the develop branch is updated. In addition, older resources should be deleted from AWS once tests are validated.

The details of the DockerFile as well as the scripts currently used by GHRC to deploy their repository to Bamboo CI/CD can be found here which will be useful to look into later on. Moreover, the environment variables in bamboo can be exported using env files to better manage terraform variables needed for deployment. See ORCA-393 for more information on this.

Some of the cards created to finish the task include:

References​