Fix docker builds in CI (flexflow#774)

goliaro · Jun 16, 2023 · d3d914b · d3d914b
1 parent 0f13b90
commit d3d914b
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 8 deletions.
diff --git a/.github/workflows/docker-build.yml b/.github/workflows/docker-build.yml
@@ -8,10 +8,9 @@ on:
   push:
     branches:
       - "master"
-    paths:
-      - "docker/**"
-      - "!docker/README.md"
-      - ".github/workflows/docker-build.yml"
+  schedule:
+    # Run every week on Sunday at midnight PT (3am ET / 8am UTC) to keep the docker images updated
+    - cron: "0 8 * * 0"
   workflow_dispatch:
 
 # Cancel outdated workflows if they are still running
@@ -43,7 +42,7 @@ jobs:
           # On push to master, build for all compatible architectures, so that we can publish 
           # a pre-built general-purpose image. On all other cases, only build for one architecture
           # to save time.
-          if [[ ${{ github.event_name }} == 'push' && ${GITHUB_REF#refs/heads/} == "master" ]]; then
+          if [[ ( ${{ github.event_name }} == 'push' || ${{ github.event_name }} == 'schedule' ) && ${GITHUB_REF#refs/heads/} == "master" ]]; then
             export FF_CUDA_ARCH=all
           else
             export FF_CUDA_ARCH=70
@@ -55,13 +54,26 @@ jobs:
         run: docker run --entrypoint python flexflow-cuda:latest -c "import flexflow.core; exit()"
 
       - name: Publish Docker environment image (on push to master)
+        if: github.repository_owner == 'flexflow'
         env:
           FLEXFLOW_CONTAINER_TOKEN: ${{ secrets.FLEXFLOW_CONTAINER_TOKEN }}
           FF_GPU_BACKEND: ${{ matrix.gpu_backend }}
         run: |
-          if [[ ${{ github.event_name }} == 'push' && ${GITHUB_REF#refs/heads/} == "master" ]]; then
+          if [[ ( ${{ github.event_name }} == 'push' || ${{ github.event_name }} == 'schedule' ) && ${GITHUB_REF#refs/heads/} == "master" ]]; then
             ./docker/publish.sh "flexflow-environment-${FF_GPU_BACKEND}"
             ./docker/publish.sh "flexflow-${FF_GPU_BACKEND}"
           else
             echo "No need to update Docker containers in ghrc.io registry at this time."
           fi
+  
+  notify-slack:
+    name: Notify Slack in case of failure
+    runs-on: ubuntu-20.04
+    needs: docker-build
+    if: ${{ failure() && github.event_name == 'schedule' && github.repository_owner == 'flexflow' }}
+    steps:
+      - name: Send Slack message
+        env:
+          SLACK_WEBHOOK: ${{ secrets.SLACK_WEBHOOK }}
+        run: |
+          curl -X POST -H 'Content-type: application/json' --data "{\"text\":\"Weekly FlexFlow Docker images build failed! <https://github.com/flexflow/FlexFlow/actions/runs/$GITHUB_RUN_ID|(See here).> :x: \"}" $SLACK_WEBHOOK
diff --git a/docker/flexflow-environment/Dockerfile b/docker/flexflow-environment/Dockerfile
@@ -48,9 +48,10 @@ ENV CUDNN_DIR /usr/local/cuda
 ENV CUDA_DIR /usr/local/cuda
 
 # Install python packages and other dependencies
-RUN conda install -c conda-forge cmake make pillow cmake-build-extension pybind11 numpy pandas keras-preprocessing 
+RUN conda install -c conda-forge cmake make pillow cmake-build-extension pybind11 numpy pandas keras-preprocessing
 # Install CPU-only Pytorch and related dependencies
 RUN conda install pytorch torchvision torchaudio cpuonly -c pytorch
-RUN conda install -c conda-forge onnx tensorflow transformers sentencepiece
+RUN conda install -c conda-forge onnx transformers sentencepiece
+RUN pip3 install tensorflow
 
 ENTRYPOINT ["/bin/bash"]