Mark flaky tests (#813)

* mark flaky tests using pytest.mark * split integration test runs by flaky and non-flaky * fix the posargs syntax for tox * force flaky tests to run in series, fix python version getting truncated * revert the integration test job name to match the existing names * separate flaky tests as a separate matrix of jobs * recombine the integration tests * allow integration tests to run fully in parallel for non-flaky tests * remove windows from integration test matrix since we weren't originally testing on windows anyway * add windows back in * register the custom marker with pytest * revert the combination of integration tests * configure a pytest return of 5 to be 0 * fix test order to avoid unnecessary change * mark more flaky tests * mark more flaky tests * pin windows images to 2019 to avoid datadog traceport failures * pinning windows images to 2019 didn't solve datadog issue, updating pin to the most recent version * turn off telemetry warnings (and telemetry) * mark more tests as flaky * mark more tests as flaky * incorporate feedback * incorporate feedback
dbt-labs · May 10, 2024 · b57890e · b57890e
1 parent 80a484b
commit b57890e
Show file tree

Hide file tree

Showing 11 changed files with 151 additions and 167 deletions.
diff --git a/.github/scripts/integration-test-matrix.js b/.github/scripts/integration-test-matrix.js
diff --git a/.github/workflows/integration.yml b/.github/workflows/integration.yml
@@ -53,85 +53,32 @@ defaults:
     shell: bash
 
 jobs:
-  # generate test metadata about what files changed and the testing matrix to use
-  test-metadata:
+  test:
+    name: redshift / python ${{ matrix.python-version }} / ${{ matrix.os }}
+
     # run if not a PR from a forked repository or has a label to mark as safe to test
     if: >-
       github.event_name != 'pull_request_target' ||
       github.event.pull_request.head.repo.full_name == github.repository ||
       contains(github.event.pull_request.labels.*.name, 'ok to test')
-    runs-on: ubuntu-latest
-
-    outputs:
-      matrix: ${{ steps.generate-matrix.outputs.result }}
-
-    steps:
-      - name: Check out the repository (non-PR)
-        if: github.event_name != 'pull_request_target'
-        uses: actions/checkout@v3
-        with:
-          persist-credentials: false
-
-      - name: Check out the repository (PR)
-        if: github.event_name == 'pull_request_target'
-        uses: actions/checkout@v3
-        with:
-          persist-credentials: false
-          ref: ${{ github.event.pull_request.head.sha }}
-
-      - name: Check if relevant files changed
-        if: github.event_name == 'pull_request_target'
-        # https://github.com/marketplace/actions/paths-changes-filter
-        # For each filter, it sets output variable named by the filter to the text:
-        #  'true' - if any of changed files matches any of filter rules
-        #  'false' - if none of changed files matches any of filter rules
-        # also, returns:
-        #  `changes` - JSON array with names of all filters matching any of the changed files
-        uses: dorny/paths-filter@v2
-        id: get-changes
-        with:
-          token: ${{ secrets.GITHUB_TOKEN }}
-          filters: |
-            redshift:
-              - 'dbt/**'
-              - 'tests/**'
-              - 'dev-requirements.txt'
-
-      - name: Generate integration test matrix
-        id: generate-matrix
-        uses: actions/github-script@v6
-        env:
-          CHANGES: ${{ steps.get-changes.outputs.changes }}
-        with:
-          script: |
-            const script = require('./.github/scripts/integration-test-matrix.js')
-            const matrix = script({ context })
-            console.log(matrix)
-            return matrix
-  test:
-    name: ${{ matrix.adapter }} / python ${{ matrix.python-version }} / ${{ matrix.os }}
-
-    # run if not a PR from a forked repository or has a label to mark as safe to test
-    # also checks that the matrix generated is not empty
-    if: >-
-      needs.test-metadata.outputs.matrix &&
-      fromJSON( needs.test-metadata.outputs.matrix ).include[0] &&
-      (
-        github.event_name != 'pull_request_target' ||
-        github.event.pull_request.head.repo.full_name == github.repository ||
-        contains(github.event.pull_request.labels.*.name, 'ok to test')
-      )
     runs-on: ${{ matrix.os }}
 
-    needs: test-metadata
-
     strategy:
       fail-fast: false
       max-parallel: 3
-      matrix: ${{ fromJSON(needs.test-metadata.outputs.matrix) }}
+      matrix:
+        os: [ubuntu-22.04, macos12, windows-2022]
+        python-version: ["3.8"]
+        include:
+          - os: ubuntu-22.04
+            python-version: "3.9"
+          - os: ubuntu-22.04
+            python-version: "3.10"
+          - os: ubuntu-22.04
+            python-version: "3.11"
 
     env:
-      TOXENV: integration-${{ matrix.adapter }}
+      TOXENV: integration-redshift
       PYTEST_ADDOPTS: "-v --color=yes -n4 --csv integration_results.csv"
       DBT_INVOCATION_ENV: github-actions
       DD_CIVISIBILITY_AGENTLESS_ENABLED: true
@@ -202,7 +149,6 @@ jobs:
           AWS_REGION: ${{ vars.REDSHIFT_TEST_REGION }}
 
       - name: Run tox (redshift)
-        if: matrix.adapter == 'redshift'
         env:
           REDSHIFT_TEST_DBNAME: ${{ secrets.REDSHIFT_TEST_DBNAME }}
           REDSHIFT_TEST_PASS: ${{ secrets.REDSHIFT_TEST_PASS }}
@@ -218,7 +164,7 @@ jobs:
           DBT_TEST_USER_1: dbt_test_user_1
           DBT_TEST_USER_2: dbt_test_user_2
           DBT_TEST_USER_3: dbt_test_user_3
-        run: tox -- --ddtrace
+        run: tox -- -m "not flaky" --ddtrace
 
       - uses: actions/upload-artifact@v3
         if: always()
@@ -235,11 +181,93 @@ jobs:
       - uses: actions/upload-artifact@v3
         if: always()
         with:
-          name: integration_results_${{ matrix.python-version }}_${{ matrix.os }}_${{ matrix.adapter }}-${{ steps.date.outputs.date }}.csv
+          name: integration_results_${{ matrix.python-version }}_${{ matrix.os }}_redshift-${{ steps.date.outputs.date }}.csv
           path: integration_results.csv
 
+  test-flaky:
+    name: redshift / python ${{ matrix.python-version }} / ubuntu-22.04 - flaky
+
+    # run this after the norm integration tests to avoid collisions
+    needs: test
+
+    # run if not a PR from a forked repository or has a label to mark as safe to test
+    if: >-
+      github.event_name != 'pull_request_target' ||
+      github.event.pull_request.head.repo.full_name == github.repository ||
+      contains(github.event.pull_request.labels.*.name, 'ok to test')
+    runs-on: ubuntu-22.04
+
+    strategy:
+      fail-fast: false
+      max-parallel: 1
+      matrix:
+        python-version: ["3.8", "3.9", "3.10", "3.11"]
+
+    env:
+      TOXENV: integration-redshift
+      PYTEST_ADDOPTS: "-v --color=yes -n1 --csv integration_results.csv"
+      DBT_INVOCATION_ENV: github-actions
+      DD_CIVISIBILITY_AGENTLESS_ENABLED: true
+      DD_INSTRUMENTATION_TELEMETRY_ENABLED: false
+      DD_API_KEY: ${{ secrets.DATADOG_API_KEY }}
+      DD_SITE: datadoghq.com
+      DD_ENV: ci
+      DD_SERVICE: ${{ github.event.repository.name }}
+
+    steps:
+      - name: Check out the repository
+        if: github.event_name != 'pull_request_target'
+        uses: actions/checkout@v3
+        with:
+          persist-credentials: false
+
+      # explicity checkout the branch for the PR,
+      # this is necessary for the `pull_request_target` event
+      - name: Check out the repository (PR)
+        if: github.event_name == 'pull_request_target'
+        uses: actions/checkout@v3
+        with:
+          persist-credentials: false
+          ref: ${{ github.event.pull_request.head.sha }}
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v4
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install python dependencies
+        run: |
+          python -m pip install  --user --upgrade pip
+          python -m pip install tox
+          python -m pip --version
+          tox --version
+
+      - name: Update dev_requirements.txt
+        if: inputs.dbt-core-branch != ''
+        run: |
+          pip install bumpversion
+          ./.github/scripts/update_dbt_core_branch.sh ${{ inputs.dbt-core-branch }}
+
+      - name: Run tox (redshift)
+        env:
+          REDSHIFT_TEST_DBNAME: ${{ secrets.REDSHIFT_TEST_DBNAME }}
+          REDSHIFT_TEST_PASS: ${{ secrets.REDSHIFT_TEST_PASS }}
+          REDSHIFT_TEST_USER: ${{ secrets.REDSHIFT_TEST_USER }}
+          REDSHIFT_TEST_PORT: ${{ secrets.REDSHIFT_TEST_PORT }}
+          REDSHIFT_TEST_HOST: ${{ secrets.REDSHIFT_TEST_HOST }}
+          REDSHIFT_TEST_REGION: ${{ vars.REDSHIFT_TEST_REGION }}
+          REDSHIFT_TEST_CLUSTER_ID: ${{ vars.REDSHIFT_TEST_CLUSTER_ID }}
+          REDSHIFT_TEST_IAM_USER_PROFILE: ${{ vars.REDSHIFT_TEST_IAM_USER_PROFILE }}
+          REDSHIFT_TEST_IAM_USER_ACCESS_KEY_ID: ${{ vars.REDSHIFT_TEST_IAM_USER_ACCESS_KEY_ID }}
+          REDSHIFT_TEST_IAM_USER_SECRET_ACCESS_KEY: ${{ secrets.REDSHIFT_TEST_IAM_USER_SECRET_ACCESS_KEY }}
+          REDSHIFT_TEST_IAM_ROLE_PROFILE: ${{ vars.REDSHIFT_TEST_IAM_ROLE_PROFILE }}
+          DBT_TEST_USER_1: dbt_test_user_1
+          DBT_TEST_USER_2: dbt_test_user_2
+          DBT_TEST_USER_3: dbt_test_user_3
+        run: tox -- -m flaky --ddtrace
+
   require-label-comment:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-22.04
 
     needs: test
 
@@ -262,7 +290,7 @@ jobs:
           check_for_duplicate_msg: true
 
   post-failure:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-22.04
     needs: test
     if: ${{ failure() }}
 

diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
@@ -173,7 +173,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-latest, macos-12, windows-latest]
+        os: [ubuntu-22.04, macos-12, windows-2022]
         python-version: ['3.8', '3.9', '3.10', '3.11']
 
     steps:

diff --git a/pytest.ini b/pytest.ini
@@ -7,3 +7,5 @@ env_files =
 testpaths =
     tests/unit
     tests/functional
+markers =
+    flaky: marks tests as flaky so they run one at a time (de-select with '-m "not flaky"')
diff --git a/tests/conftest.py b/tests/conftest.py
@@ -1 +1,12 @@
 pytest_plugins = ["dbt.tests.fixtures.project"]
+
+
+def pytest_sessionfinish(session, exitstatus):
+    """
+    Configures pytest to treat a scenario with no tests as passing
+
+    pytest returns a code 5 when it collects no tests in an effort to warn when tests are expected but not collected
+    We don't want this when running tox because some combinations of markers and test segments return nothing
+    """
+    if exitstatus == 5:
+        session.exitstatus = 0
diff --git a/tests/functional/adapter/catalog_tests/test_get_catalog.py b/tests/functional/adapter/catalog_tests/test_get_catalog.py
@@ -97,6 +97,7 @@ def my_information_schema(self, adapter, my_schema):
             identifier="INFORMATION_SCHEMA",
         ).information_schema()
 
+    @pytest.mark.flaky
     def test_get_one_catalog_by_relations(
         self,
         adapter,
@@ -120,6 +121,7 @@ def test_get_one_catalog_by_relations(
         # note the underlying table is missing as it's not in `my_relations`
         assert len(catalog) == 12
 
+    @pytest.mark.flaky
     def test_get_one_catalog_by_schemas(
         self,
         adapter,

diff --git a/tests/functional/adapter/catalog_tests/test_relation_types.py b/tests/functional/adapter/catalog_tests/test_relation_types.py
@@ -24,6 +24,7 @@ def docs(self, project):
         run_dbt(["run"])
         yield run_dbt(["docs", "generate"])
 
+    @pytest.mark.flaky
     @pytest.mark.parametrize(
         "node_name,relation_type",
         [