[ETL-654] Clean up before integration test run

.github/workflows/README.md

@@ -7,12 +7,12 @@ Recover ETL has four github workflows:
 - workflows/codeql-analysis.yml
 - workflows/cleanup.yaml
 
-| Workflow name                     | Scenario it's run |
-| :-------------------------------- |:----------------  |
-| upload-and-deploy                 | on-push from feature branch, feature branch merged into main |
-| upload-and-deploy-to-prod-main    | whenever a new tag is created |
-| codeql-analysis                   | on-push from feature branch, feature branch merged into main
-| cleanup                           | feature branch deleted |
+| Workflow name                  | Scenario it's run                                            |
+|:-------------------------------|:-------------------------------------------------------------|
+| upload-and-deploy              | on-push from feature branch, feature branch merged into main |
+| upload-and-deploy-to-prod-main | whenever a new tag is created                                |
+| codeql-analysis                | on-push from feature branch, feature branch merged into main |
+| cleanup                        | feature branch deleted                                       |
 
 ## upload-and-deploy
 
@@ -45,6 +45,16 @@ With the current way when the `test_json_to_parquet.py` run, sometimes the glue
 
 ### sceptre-deploy-develop
 
+### integration-test-develop-cleanup
+
+This is responsible for cleaning up any data locations that are used by integration
+tests. This is used after `sceptre-deploy-develop`, but before
+`integration-test-develop`. Cleans these locations:
+
+* `s3://recover-dev-input-data/$GITHUB_REF_NAME/`
+* `s3://recover-dev-intermediate-data/$GITHUB_REF_NAME/json/`
+
+
 ### integration-test-develop
 
 This builds the S3 to JSON lambda and triggers it with the pilot data so that the Recover ETL Glue Workflow will start running and processing the pilot data. **Note** that this will run with every push to the feature branch so it would be good to wait until one run of the Glue workflow finishes running as we cannot have more than 1 concurrent Glue Workflow run.
@@ -57,6 +67,16 @@ This integration test means that you have to wait until the glue workflow has fi
 
 Here that we are **NOT** configuring a S3 event notification configuration for our `prod/staging` space because we plan to submit data to `staging` "manually" after merging a PR into main and triggering the GitHub workflow.
 
+### integration-test-staging-cleanup
+
+This is responsible for cleaning up any data locations that are used by integration
+tests during the staging run. This is used after `sceptre-deploy-staging`, but before
+`integration-test-staging`. Cleans these locations:
+
+* `s3://recover-dev-input-data/staging/`
+* `s3://recover-dev-intermediate-data/staging/json/`
+
+
 ## upload-and-deploy-to-prod-main
 
 This runs **ONLY** when we create a new tag.

.github/workflows/upload-and-deploy.yaml

@@ -9,8 +9,10 @@ env:
   NAMESPACE: main
   PYTHON_VERSION: 3.9
   DEV_INPUT_BUCKET: recover-dev-input-data
+  DEV_INTERMEDIATE_BUCKET: recover-dev-intermediate-data
   DEV_PROCESSED_BUCKET: recover-dev-processed-data
   PROD_INPUT_BUCKET: recover-input-data
+  PROD_INTERMEDIATE_BUCKET: recover-intermediate-data
   INTEGRATION_TEST_NUM_EXPORTS: 28
 
 jobs:
@@ -255,11 +257,44 @@ jobs:
           Payload: '{"RequestType": "Create"}'
           LogType: Tail
 
+  integration-test-develop-cleanup:
+    name: Cleanup non-main branch data before integration tests
+    runs-on: ubuntu-latest
+    needs: sceptre-deploy-develop
+    if: github.ref_name != 'main'
+    environment: develop
+    # These permissions are needed to interact with GitHub's OIDC Token endpoint.
+    permissions:
+      id-token: write
+      contents: read
+    steps:
+      - name: Setup code, pipenv, aws
+        uses: Sage-Bionetworks/action-pipenv-aws-setup@v3
+        with:
+          role_to_assume: ${{ vars.AWS_CREDENTIALS_IAM_ROLE }}
+          role_session_name: GitHubActions-${{ github.repository_owner }}-${{ github.event.repository.name }}-${{ github.run_id }}
+          python_version: ${{ env.PYTHON_VERSION }}
+
+      - name: Set namespace for non-default branch
+        run: echo "NAMESPACE=$GITHUB_REF_NAME" >> $GITHUB_ENV
+
+      - name: Clean input data bucket
+        run: >
+          python src/scripts/manage_artifacts/clean_staging.py
+          --bucket $DEV_INPUT_BUCKET
+          --bucket_prefix "${{ env.NAMESPACE }}/"
+
+      - name: Clean intermediate data bucket
+        run: >
+          python src/scripts/manage_artifacts/clean_staging.py
+          --bucket $DEV_INTERMEDIATE_BUCKET
+          --bucket_prefix "${{ env.NAMESPACE }}/json/"
+
 
   integration-test-develop:
     name: Triggers ETL workflow with S3 test files
     runs-on: ubuntu-latest
-    needs: sceptre-deploy-develop
+    needs: [sceptre-deploy-develop, integration-test-develop-cleanup]
     environment: develop
     # These permissions are needed to interact with GitHub's OIDC Token endpoint.
     permissions:
@@ -335,11 +370,40 @@ jobs:
       - name: Deploy sceptre stacks to staging on prod
         run: pipenv run sceptre --var "namespace=staging" launch prod --yes
 
+  integration-test-staging-cleanup:
+    name: Cleanup main branch staging data before integration tests
+    runs-on: ubuntu-latest
+    needs: sceptre-deploy-staging
+    if: github.ref_name == 'main'
+    environment: prod
+    # These permissions are needed to interact with GitHub's OIDC Token endpoint.
+    permissions:
+      id-token: write
+      contents: read
+    steps:
+      - name: Setup code, pipenv, aws
+        uses: Sage-Bionetworks/action-pipenv-aws-setup@v3
+        with:
+          role_to_assume: ${{ vars.AWS_CREDENTIALS_IAM_ROLE }}
+          role_session_name: GitHubActions-${{ github.repository_owner }}-${{ github.event.repository.name }}-${{ github.run_id }}
+          python_version: ${{ env.PYTHON_VERSION }}
+
+      - name: Clean input data bucket
+        run: >
+          python src/scripts/manage_artifacts/clean_staging.py
+          --bucket $PROD_INPUT_BUCKET
+          --bucket_prefix "staging/"
+
+      - name: Clean intermediate data bucket
+        run: >
+          python src/scripts/manage_artifacts/clean_staging.py
+          --bucket $PROD_INTERMEDIATE_BUCKET
+          --bucket_prefix "staging/json/"
 
   integration-test-staging:
     name: Triggers staging workflow with production data
     runs-on: ubuntu-latest
-    needs: sceptre-deploy-staging
+    needs: [sceptre-deploy-staging, integration-test-staging-cleanup]
     environment: prod
     # These permissions are needed to interact with GitHub's OIDC Token endpoint.
     permissions:

Pipfile

@@ -18,6 +18,7 @@ moto = "~=4.1"
 datacompy = "~=0.8"
 docker = "~=6.1"
 ecs_logging = "~=2.0"
+boto3 = "<2.0"
 # flask libraries required for moto_server
 flask = "~=2.0"
 flask-cors = "~=3.0"