Update training packaging pipeline's docker files (#20853)

### Description Similar to #20786 . The last PR was able to update all pipelines and all docker files. This is a follow-up to that PR. ### Motivation and Context 1. To extract the common part as a reusable build infra among different ONNX Runtime projects. 2. Avoid hitting docker hub's limit: 429 Too Many Requests - Server message: toomanyrequests: You have reached your pull rate limit. You may increase the limit by authenticating and upgrading: https://www.docker.com/increase-rate-limit
microsoft · May 31, 2024 · 67bc943 · 67bc943
1 parent 00589f5
commit 67bc943
Show file tree

Hide file tree

Showing 42 changed files with 229 additions and 958 deletions.
diff --git a/tools/ci_build/github/azure-pipelines/bigmodels-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/bigmodels-ci-pipeline.yml
@@ -40,12 +40,6 @@ parameters:
 
 resources:
   repositories:
-  - repository: manylinux
-    type: Github
-    endpoint: Microsoft
-    name: pypa/manylinux
-    ref: 5eda9aded5462201e6310105728d33016e637ea7
-
   - repository: LLaMa2Onnx
     type: Github
     endpoint: Microsoft
@@ -55,7 +49,7 @@ resources:
 variables:
   - template: templates/common-variables.yml
   - name: docker_base_image
-    value: nvidia/cuda:11.8.0-cudnn8-devel-ubi8
+    value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20240530.3
   - name: linux_trt_version
     value: 10.0.1.6-1.cuda11.8
   - name: Repository
@@ -113,7 +107,7 @@ stages:
       inputs:
         script: |
           mkdir -p $HOME/.onnx
-          docker run -e CFLAGS="-Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fstack-protector-strong -fstack-clash-protection -fcf-protection -O3 -Wl,--strip-all" -e CXXFLAGS="-Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fstack-protector-strong -fstack-clash-protection -fcf-protection -O3 -Wl,--strip-all" --rm \
+          docker run -e --rm \
             --volume /data/onnx:/data/onnx:ro \
             --volume $(Build.SourcesDirectory):/onnxruntime_src \
             --volume $(Build.BinariesDirectory):/build \

diff --git a/tools/ci_build/github/azure-pipelines/c-api-noopenmp-packaging-pipelines.yml b/tools/ci_build/github/azure-pipelines/c-api-noopenmp-packaging-pipelines.yml
@@ -93,11 +93,6 @@ variables:
     value: nvidia/cuda:11.8.0-cudnn8-devel-ubi8
   ${{ if eq(parameters.CudaVersion, '12.2') }}:
     value: nvidia/cuda:12.2.2-cudnn8-devel-ubi8
-- name: linux_trt_version
-  ${{ if eq(parameters.CudaVersion, '11.8') }}:
-    value: 10.0.1.6-1.cuda11.8
-  ${{ if eq(parameters.CudaVersion, '12.2') }}:
-    value: 10.0.1.6-1.cuda12.4
 - name: win_trt_version
   ${{ if eq(parameters.CudaVersion, '11.8') }}:
     value: 11.8
@@ -184,7 +179,6 @@ stages:
       DoCompliance: ${{ parameters.DoCompliance }}
       CudaVersion: ${{ parameters.CudaVersion }}
       docker_base_image: ${{ variables.docker_base_image }}
-      linux_trt_version: ${{ variables.linux_trt_version }}
       RunOnnxRuntimeTests: ${{ parameters.RunOnnxRuntimeTests }}
       UseIncreasedTimeoutForTests: ${{ parameters.UseIncreasedTimeoutForTests }}
       win_trt_home: ${{ variables.win_trt_home }}
@@ -228,6 +222,7 @@ stages:
           --build-arg PREPEND_PATH=/opt/rh/gcc-toolset-12/root/usr/bin:
           --build-arg LD_LIBRARY_PATH_ARG=/opt/rh/gcc-toolset-12/root/usr/lib64:/opt/rh/gcc-toolset-12/root/usr/lib:/opt/rh/gcc-toolset-12/root/usr/lib64/dyninst:/opt/rh/gcc-toolset-12/root/usr/lib/dyninst:/usr/local/lib64:/usr/local/lib
         Repository: onnxruntimetrainingrocmbuild-rocm$(RocmVersion)
+        CheckOutManyLinux: true
 
     - template: templates/set-version-number-variables-step.yml
 

diff --git a/tools/ci_build/github/azure-pipelines/cuda-packaging-pipeline.yml b/tools/ci_build/github/azure-pipelines/cuda-packaging-pipeline.yml
@@ -66,11 +66,6 @@ variables:
       value: nvidia/cuda:11.8.0-cudnn8-devel-ubi8
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
       value: nvidia/cuda:12.2.2-cudnn8-devel-ubi8
-  - name: linux_trt_version
-    ${{ if eq(parameters.CudaVersion, '11.8') }}:
-      value: 10.0.1.6-1.cuda11.8
-    ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: 10.0.1.6-1.cuda12.4
   - name: win_trt_home
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
       value: $(Agent.TempDirectory)\TensorRT-10.0.1.6.Windows10.x86_64.cuda-11.8
@@ -120,7 +115,6 @@ stages:
       DoCompliance: ${{ parameters.DoCompliance }}
       CudaVersion: ${{ parameters.CudaVersion }}
       docker_base_image: ${{ variables.docker_base_image }}
-      linux_trt_version: ${{ variables.linux_trt_version }}
       RunOnnxRuntimeTests: ${{ parameters.RunOnnxRuntimeTests }}
       UseIncreasedTimeoutForTests: ${{ parameters.UseIncreasedTimeoutForTests }}
       win_trt_home: ${{ variables.win_trt_home }}

diff --git a/tools/ci_build/github/azure-pipelines/linux-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/linux-ci-pipeline.yml
@@ -1,3 +1,4 @@
+
 ##### start trigger Don't edit it manually, Please do edit set-trigger-rules.py ####
 trigger:
   branches:
@@ -27,13 +28,6 @@ pr:
     - 'onnxruntime/core/providers/js'
 #### end trigger ####
 
-resources:
-  repositories:
-  - repository: manylinux # The name used to reference this repository in the checkout step
-    type: Github
-    endpoint: Microsoft
-    name: pypa/manylinux
-    ref: 5eda9aded5462201e6310105728d33016e637ea7
 stages:
 - stage: x64
   dependsOn: []

diff --git a/tools/ci_build/github/azure-pipelines/linux-cpu-minimal-build-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/linux-cpu-minimal-build-ci-pipeline.yml
@@ -46,14 +46,6 @@ pr:
     - BUILD.md
     - 'js/web'
     - 'onnxruntime/core/providers/js'
-resources:
-  repositories:
-  - repository: manylinux
-    type: Github
-    endpoint: Microsoft
-    name: pypa/manylinux
-    ref: 5eda9aded5462201e6310105728d33016e637ea7
-
 jobs:
 - job: Linux_CPU_Minimal_Build_E2E
   timeoutInMinutes: 120

diff --git a/tools/ci_build/github/azure-pipelines/linux-dnnl-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/linux-dnnl-ci-pipeline.yml
@@ -27,14 +27,6 @@ pr:
     - 'onnxruntime/core/providers/js'
 #### end trigger ####
 
-resources:
-  repositories:
-  - repository: manylinux
-    type: Github
-    endpoint: Microsoft
-    name: pypa/manylinux
-    ref: 5eda9aded5462201e6310105728d33016e637ea7
-
 jobs:
 - job: Linux_py_Wheels
   timeoutInMinutes: 180

diff --git a/tools/ci_build/github/azure-pipelines/linux-gpu-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/linux-gpu-ci-pipeline.yml
@@ -45,26 +45,12 @@ parameters:
     type: string
     default: '0'
 
-resources:
-  repositories:
-  - repository: manylinux
-    type: Github
-    endpoint: Microsoft
-    name: pypa/manylinux
-    ref: 5eda9aded5462201e6310105728d33016e637ea7
-
 variables:
   - name: docker_base_image
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
-      value: nvidia/cuda:11.8.0-cudnn8-devel-ubi8
-    ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: nvidia/cuda:12.2.2-cudnn8-devel-ubi8
-
-  - name: linux_trt_version
-    ${{ if eq(parameters.CudaVersion, '11.8') }}:
-      value: 10.0.1.6-1.cuda11.8
+      value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20240530.3
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: 10.0.1.6-1.cuda12.4
+      value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_2_x64_ubi8_gcc12:20240530.3
 
   - name: Repository
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
@@ -97,12 +83,7 @@ stages:
       parameters:
         Dockerfile: tools/ci_build/github/linux/docker/Dockerfile.manylinux2_28_cuda
         Context: tools/ci_build/github/linux/docker
-        DockerBuildArgs: "
-        --network=host
-        --build-arg BASEIMAGE=$(docker_base_image)
-        --build-arg TRT_VERSION=$(linux_trt_version)
-        --build-arg BUILD_UID=$( id -u )
-        "
+        DockerBuildArgs: "--build-arg BASEIMAGE=$(docker_base_image) --build-arg BUILD_UID=$( id -u )"
         Repository: $(Repository)
 
     - task: Cache@2
@@ -123,7 +104,7 @@ stages:
     - script: |
         set -e -x
         mkdir -p $HOME/.onnx
-        docker run -e CFLAGS="-Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fstack-protector-strong -fstack-clash-protection -fcf-protection -O3 -Wl,--strip-all" -e CXXFLAGS="-Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fstack-protector-strong -fstack-clash-protection -fcf-protection -O3 -Wl,--strip-all" --rm \
+        docker run --rm \
           --volume /data/onnx:/data/onnx:ro \
           --volume $(Build.SourcesDirectory):/onnxruntime_src \
           --volume $(Build.BinariesDirectory):/build \
@@ -133,39 +114,12 @@ stages:
           -e ALLOW_RELEASED_ONNX_OPSET_ONLY=0 \
           -e NIGHTLY_BUILD \
           -e BUILD_BUILDNUMBER \
-          -e CCACHE_DIR=/cache \
-          $(Repository) \
-          /bin/bash -c "
-            set -ex; \
-            env; \
-            ccache -s; \
-            /opt/python/cp38-cp38/bin/python3 /onnxruntime_src/tools/ci_build/build.py \
-              --build_dir /build --cmake_generator Ninja \
-              --config Release --update --build \
-              --skip_submodule_sync \
-              --build_shared_lib \
-              --parallel --use_binskim_compliant_compile_flags \
-              --build_wheel \
-              --enable_onnx_tests --use_cuda --cuda_version=${{parameters.CudaVersion}} --cuda_home=/usr/local/cuda-${{parameters.CudaVersion}} --cudnn_home=/usr/local/cuda-${{parameters.CudaVersion}} \
-              --enable_cuda_profiling --enable_cuda_nhwc_ops \
-              --enable_pybind --build_java \
-              --use_cache \
-              --cmake_extra_defines  CMAKE_CUDA_ARCHITECTURES=75 \
-              --cmake_extra_defines onnxruntime_BUILD_UNIT_TESTS=ON \
-              --cmake_extra_defines  onnxruntime_ENABLE_CUDA_EP_INTERNAL_TESTS=ON; \
-                ccache -sv; \
-                ccache -z"
+          -e CCACHE_DIR=/cache -w /onnxruntime_src \
+          $(Repository) tools/ci_build/github/linux/build_cuda_ci.sh
       workingDirectory: $(Build.SourcesDirectory)
       displayName: Build Onnxruntime
 
-    - task: CmdLine@2
-      inputs:
-        script: |
-          rm -rf $(Build.BinariesDirectory)/Release/onnxruntime $(Build.BinariesDirectory)/Release/pybind11
-          rm -f $(Build.BinariesDirectory)/Release/models
-          find $(Build.BinariesDirectory)/Release/_deps -mindepth 1 ! -regex '^$(Build.BinariesDirectory)/Release/_deps/onnx-src\(/.*\)?' -delete
-          cd $(Build.BinariesDirectory)/Release
-          find -executable -type f > $(Build.BinariesDirectory)/Release/perms.txt
+    - script: $(Build.SourcesDirectory)/tools/ci_build/github/linux/delete_unused_files_before_upload.sh
 
     - task: PublishPipelineArtifact@0
       displayName: 'Publish Pipeline Artifact'
@@ -203,12 +157,7 @@ stages:
       parameters:
         Dockerfile: tools/ci_build/github/linux/docker/Dockerfile.manylinux2_28_cuda
         Context: tools/ci_build/github/linux/docker
-        DockerBuildArgs: "
-        --network=host
-        --build-arg BASEIMAGE=$(docker_base_image)
-        --build-arg TRT_VERSION=$(linux_trt_version)
-        --build-arg BUILD_UID=$( id -u )
-        "
+        DockerBuildArgs: "--build-arg BASEIMAGE=$(docker_base_image) --build-arg BUILD_UID=$( id -u )"
         Repository: $(Repository)
 
     - task: CmdLine@2

diff --git a/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/linux-gpu-tensorrt-ci-pipeline.yml
@@ -34,29 +34,19 @@ parameters:
     values:
       - 11.8
       - 12.2
-      - 12.4
-resources:
-  repositories:
-  - repository: manylinux
-    type: Github
-    endpoint: Microsoft
-    name: pypa/manylinux
-    ref: 5eda9aded5462201e6310105728d33016e637ea7
+
 variables:
   - name: docker_base_image
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
-      value: nvidia/cuda:11.8.0-cudnn8-devel-ubi8
+      value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda11_x64_almalinux8_gcc11:20240530.3
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
-      value: nvidia/cuda:12.2.2-cudnn8-devel-ubi8
-    ${{ if eq(parameters.CudaVersion, '12.4') }}:
-      value: nvidia/cuda:12.4.1-cudnn-devel-ubi8
+      value: onnxruntimebuildcache.azurecr.io/internal/azureml/onnxruntime/build/cuda12_2_x64_ubi8_gcc12:20240530.3
   - name: linux_trt_version
     ${{ if eq(parameters.CudaVersion, '11.8') }}:
       value: 10.0.1.6-1.cuda11.8
     ${{ if eq(parameters.CudaVersion, '12.2') }}:
       value: 10.0.1.6-1.cuda12.4
-    ${{ if eq(parameters.CudaVersion, '12.4') }}:
-      value: 10.0.1.6-1.cuda12.4
+
 jobs:
 - job: Linux_Build
   timeoutInMinutes: 180
@@ -99,7 +89,7 @@ jobs:
         - task: CmdLine@2
           inputs:
             script: |
-              docker run --gpus all -e CFLAGS="-Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fstack-protector-strong -fstack-clash-protection -fcf-protection -O3 -Wl,--strip-all" -e CXXFLAGS="-Wp,-D_FORTIFY_SOURCE=2 -Wp,-D_GLIBCXX_ASSERTIONS -fstack-protector-strong -fstack-clash-protection -fcf-protection -O3 -Wl,--strip-all" --rm \
+              docker run --gpus all --rm \
                   --volume /data/onnx:/data/onnx:ro \
                   --volume $(Build.SourcesDirectory):/onnxruntime_src \
                   --volume $(Build.BinariesDirectory):/build \
@@ -109,26 +99,8 @@ jobs:
                   -e ALLOW_RELEASED_ONNX_OPSET_ONLY=0 \
                   -e NIGHTLY_BUILD \
                   -e BUILD_BUILDNUMBER \
-                  -e CCACHE_DIR=/cache \
-                  onnxruntimetensorrt86gpubuild \
-                  /bin/bash -c "
-                    set -ex; \
-                    ccache -s; \
-                      /opt/python/cp38-cp38/bin/python3 /onnxruntime_src/tools/ci_build/build.py \
-                      --build_dir /build --cmake_generator Ninja \
-                      --config Release \
-                      --skip_submodule_sync \
-                      --build_shared_lib \
-                      --parallel --use_binskim_compliant_compile_flags \
-                      --build_wheel \
-                      --enable_onnx_tests \
-                      --use_cuda --cuda_home=/usr/local/cuda-${{ parameters.CudaVersion }} --cudnn_home=/usr/lib64/ \
-                      --enable_pybind --build_java \
-                      --use_tensorrt --tensorrt_home /usr \
-                      --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=75 \
-                      --use_cache; \
-                     ccache -sv; \
-                     ccache -z"
+                  -e CCACHE_DIR=/cache -w /onnxruntime_src \
+                  onnxruntimetensorrt86gpubuild tools/ci_build/github/linux/build_tensorrt_ci.sh
             workingDirectory: $(Build.SourcesDirectory)
 
   - template: templates/explicitly-defined-final-tasks.yml
diff --git a/tools/ci_build/github/azure-pipelines/orttraining-pai-ci-pipeline.yml b/tools/ci_build/github/azure-pipelines/orttraining-pai-ci-pipeline.yml
@@ -62,6 +62,7 @@ jobs:
         --build-arg PREPEND_PATH=/opt/rh/gcc-toolset-12/root/usr/bin:
         --build-arg LD_LIBRARY_PATH_ARG=/opt/rh/gcc-toolset-12/root/usr/lib64:/opt/rh/gcc-toolset-12/root/usr/lib:/opt/rh/gcc-toolset-12/root/usr/lib64/dyninst:/opt/rh/gcc-toolset-12/root/usr/lib/dyninst:/usr/local/lib64:/usr/local/lib
       Repository: onnxruntimetrainingrocm-cibuild-rocm$(RocmVersion)-manylinux-build
+      CheckOutManyLinux: true
 
   - task: Cache@2
     inputs: