Move constraints generation to separate job in CI workflow (apache#34990

)

We generate constraints for preview right after images are build, in
order to be able to see the output and to diagnose which dependencies
have been updated later in the process.

So far we were doing it in CI workflow and When constraints generation
failed however (for example because pip backtracking takes a log ot
time), it faied build image workflow and did not allow tests to
complete.

This PR extracts constraint generation to CI separate job which
does not block tests from running in case constraints generation
fails or times out.

It also moves out the steps from the composite action which allows
to better see which step failed and allows to see the constraints
used in each job more easily.

.github/actions/build-ci-images/action.yml

@@ -34,30 +34,6 @@ runs:
     - name: "Build & Push AMD64 CI images ${{ env.IMAGE_TAG }} ${{ env.PYTHON_VERSIONS }}"
       shell: bash
       run: breeze ci-image build --push --tag-as-latest --run-in-parallel --upgrade-on-failure
-    - name: "Generate source constraints"
-      shell: bash
-      run: >
-        breeze release-management generate-constraints --run-in-parallel
-        --airflow-constraints-mode constraints-source-providers
-#    - name: "Generate PyPI constraints"
-#      shell: bash
-#      run: >
-#        breeze release-management generate-constraints --run-in-parallel
-#        --airflow-constraints-mode constraints
-#      if: ${{ inputs.build-provider-packages != 'true' }}
-    - name: "Print dependency upgrade summary"
-      shell: bash
-      run: |
-        for PYTHON_VERSION in ${{ env.PYTHON_VERSIONS }}; do
-          echo "Summarizing Python $PYTHON_VERSION"
-          cat "files/constraints-${PYTHON_VERSION}/*.md" >> $GITHUB_STEP_SUMMARY || true
-        done
-    - name: "Upload constraint artifacts"
-      uses: actions/upload-artifact@v3
-      with:
-        name: constraints
-        path: ./files/constraints-*/constraints-*.txt
-        retention-days: 7
     - name: "Fix ownership"
       shell: bash
       run: breeze ci fix-ownership

.github/workflows/ci.yml

@@ -326,6 +326,61 @@ jobs:
           DEBUG_RESOURCES: ${{needs.build-info.outputs.debug-resources}}
           BUILD_TIMEOUT_MINUTES: 70
 
+  preview-constraints:
+    permissions:
+      contents: read
+    timeout-minutes: 20
+    continue-on-error: true
+    name: >
+      Preview constraints
+      ${{needs.build-info.outputs.all-python-versions-list-as-string}}
+    runs-on: "${{needs.build-info.outputs.runs-on}}"
+    needs: [build-info, build-ci-images]
+    if: needs.build-info.outputs.upgrade-to-newer-dependencies != 'false'
+    env:
+      RUNS_ON: "${{ needs.build-info.outputs.runs-on }}"
+      PYTHON_VERSIONS: ${{needs.build-info.outputs.all-python-versions-list-as-string}}
+      DEBUG_RESOURCES: ${{needs.build-info.outputs.debug-resources}}
+    steps:
+      - name: Cleanup repo
+        run: docker run -v "${GITHUB_WORKSPACE}:/workspace" -u 0:0 bash -c "rm -rf /workspace/*"
+      - name: "Checkout ${{ github.ref }} ( ${{ github.sha }} )"
+        uses: actions/checkout@v3
+        with:
+          persist-credentials: false
+          submodules: recursive
+      - name: "Install Breeze"
+        uses: ./.github/actions/breeze
+      - name: Pull CI images ${{ env.PYTHON_VERSIONS }}:${{ env.IMAGE_TAG }}
+        run: breeze ci-image pull --run-in-parallel --tag-as-latest
+      - name: "Source constraints"
+        shell: bash
+        run: >
+          breeze release-management generate-constraints --run-in-parallel
+          --airflow-constraints-mode constraints-source-providers || true
+      - name: "PyPI constraints"
+        shell: bash
+        timeout-minutes: 25
+        run: >
+          breeze release-management generate-constraints --run-in-parallel
+          --airflow-constraints-mode constraints || true
+      - name: "Dependency upgrade summary"
+        shell: bash
+        run: |
+          for PYTHON_VERSION in ${{ env.PYTHON_VERSIONS }}; do
+            echo "Summarizing Python $PYTHON_VERSION"
+            cat "files/constraints-${PYTHON_VERSION}"/*.md >> $GITHUB_STEP_SUMMARY || true
+          done
+      - name: "Upload constraint artifacts"
+        uses: actions/upload-artifact@v3
+        with:
+          name: constraints
+          path: ./files/constraints-*/constraints-*.txt
+          retention-days: 7
+      - name: "Fix ownership"
+        run: breeze ci fix-ownership
+        if: always()
+
   build-prod-images:
     timeout-minutes: 80
     name: >
@@ -1662,6 +1717,7 @@ jobs:
     env:
       RUNS_ON: "${{needs.build-info.outputs.runs-on}}"
       DEBUG_RESOURCES: ${{needs.build-info.outputs.debug-resources}}
+      PYTHON_VERSIONS: ${{ needs.build-info.outputs.python-versions-list-as-string }}
     if: needs.build-info.outputs.upgrade-to-newer-dependencies != 'false'
     steps:
       - name: Cleanup repo
@@ -1677,16 +1733,18 @@ jobs:
         run: breeze ci-image pull --run-in-parallel --tag-as-latest
         env:
           PYTHON_VERSIONS: ${{ needs.build-info.outputs.python-versions-list-as-string }}
-      - name: "Generate constraints"
+      - name: "Generate source constraints"
+        run: |
+          breeze release-management generate-constraints \
+              --run-in-parallel --airflow-constraints-mode constraints-source-providers
+      - name: "Generate no-providers constraints"
         run: |
-          breeze release-management generate-constraints --run-in-parallel \
-              --airflow-constraints-mode constraints-source-providers
           breeze release-management generate-constraints \
               --run-in-parallel --airflow-constraints-mode constraints-no-providers
-#          breeze release-management generate-constraints \
-#              --run-in-parallel --airflow-constraints-mode constraints
-        env:
-          PYTHON_VERSIONS: ${{ needs.build-info.outputs.python-versions-list-as-string }}
+      - name: "Generate PyPi constraints"
+        run: |
+          breeze release-management generate-constraints \
+              --run-in-parallel --airflow-constraints-mode constraints
       - name: "Set constraints branch name"
         id: constraints-branch
         run: ./scripts/ci/constraints/ci_branch_constraints.sh >> ${GITHUB_OUTPUT}

Dockerfile.ci

@@ -1435,8 +1435,11 @@ RUN echo "Airflow version: ${AIRFLOW_VERSION}"
 # Those are additional constraints that are needed for some extras but we do not want to
 # force them on the main Airflow package. Currently we need no extra limits as PIP 23.1+ has much better
 # dependency resolution and we do not need to limit the versions of the dependencies
-# aiobotocore is limited temporarily until it stops backtracking pip
-ARG EAGER_UPGRADE_ADDITIONAL_REQUIREMENTS=""
+#
+# Without grpcio-status limit, pip gets into very long backtracking
+# We should attempt to remove it in the future
+#
+ARG EAGER_UPGRADE_ADDITIONAL_REQUIREMENTS="grpcio-status>=1.55.0"
 ARG UPGRADE_TO_NEWER_DEPENDENCIES="false"
 ARG VERSION_SUFFIX_FOR_PYPI=""
 

scripts/in_container/_in_container_utils.sh

@@ -333,9 +333,10 @@ function install_all_providers_from_pypi_with_eager_upgrade() {
     # Installing it with Airflow makes sure that the version of package that matches current
     # Airflow requirements will be used.
     # shellcheck disable=SC2086
+    set -x
     pip install ".[${NO_PROVIDERS_EXTRAS}]" "${packages_to_install[@]}" ${EAGER_UPGRADE_ADDITIONAL_REQUIREMENTS=} \
         --upgrade --upgrade-strategy eager
-
+    set +x
 }
 
 function install_all_provider_packages_from_wheels() {

scripts/in_container/run_generate_constraints.sh

@@ -152,5 +152,6 @@ echo
 echo "Constraints error markdown generated in ${CONSTRAINTS_MARKDOWN_DIFF}"
 echo
 
+ls "${CONSTRAINTS_MARKDOWN_DIFF}"
 
 exit 0

setup.py

@@ -1042,4 +1042,4 @@ def include_provider_namespace_packages_when_installing_from_sources() -> None:
 
 
 if __name__ == "__main__":
-    do_setup()  # comment
+    do_setup()  # comment to trigger upgrade to newer dependencies