Merge branch 'kubeflow:master' into specify-cpu-arch

.github/workflows/darts-cifar10-e2e-test.yaml

@@ -0,0 +1,33 @@
+name: E2E Test with darts-cnn-cifar10
+on:
+  - pull_request
+
+jobs:
+  e2e:
+    runs-on: ubuntu-20.04
+    timeout-minutes: 120
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v2
+
+      - name: Setup Test Env
+        uses: ./.github/workflows/template-setup-e2e-test
+        with:
+          kubernetes-version: ${{ matrix.kubernetes-version }}
+
+      - name: Run e2e test with ${{ matrix.experiments }} experiments
+        uses: ./.github/workflows/template-e2e-test
+        with:
+          experiments: ${{ matrix.experiments }}
+          # Comma Delimited
+          trial-images: darts-cnn-cifar10-cpu
+
+    strategy:
+      fail-fast: false
+      matrix:
+        # Detail: https://hub.docker.com/r/kindest/node
+        # TODO (tenzen-y): We need to consider running tests on more kubernetes versions.
+        # kubernetes-version: ["v1.20.15", "v1.21.12", "v1.22.9", "v1.23.6", "v1.24.1"]
+        kubernetes-version: ["v1.21.12", "v1.22.9", "v1.23.6"]
+        # Comma Delimited
+        experiments: ["darts-cpu"]

.github/workflows/enas-cifar10-e2e-test.yaml

@@ -0,0 +1,33 @@
+name: E2E Test with enas-cnn-cifar10
+on:
+  - pull_request
+
+jobs:
+  e2e:
+    runs-on: ubuntu-20.04
+    timeout-minutes: 120
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v2
+
+      - name: Setup Test Env
+        uses: ./.github/workflows/template-setup-e2e-test
+        with:
+          kubernetes-version: ${{ matrix.kubernetes-version }}
+
+      - name: Run e2e test with ${{ matrix.experiments }} experiments
+        uses: ./.github/workflows/template-e2e-test
+        with:
+          experiments: ${{ matrix.experiments }}
+          # Comma Delimited
+          trial-images: enas-cnn-cifar10-cpu
+
+    strategy:
+      fail-fast: false
+      matrix:
+        # Detail: https://hub.docker.com/r/kindest/node
+        # TODO (tenzen-y): We need to consider running tests on more kubernetes versions.
+        # kubernetes-version: ["v1.20.15", "v1.21.12", "v1.22.9", "v1.23.6", "v1.24.1"]
+        kubernetes-version: ["v1.21.12", "v1.22.9", "v1.23.6"]
+        # Comma Delimited
+        experiments: ["enas-cpu"]

.github/workflows/katib-ui-e2e-test.yaml

@@ -0,0 +1,30 @@
+name: E2E Test for katib-ui
+on:
+  - pull_request
+
+jobs:
+  e2e:
+    runs-on: ubuntu-20.04
+    timeout-minutes: 120
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v2
+
+      - name: Setup Test Env
+        uses: ./.github/workflows/template-setup-e2e-test
+        with:
+          kubernetes-version: ${{ matrix.kubernetes-version }}
+
+      - name: Set Up KinD Cluster
+        run: ./test/e2e/v1beta1/scripts/gh-actions/setup-kind.sh true
+
+      - name: Start Katib
+        run: ./test/e2e/v1beta1/scripts/gh-actions/setup-katib.sh true false
+
+    strategy:
+      fail-fast: false
+      matrix:
+        # Detail: https://hub.docker.com/r/kindest/node
+        # TODO (tenzen-y): We need to consider running tests on more kubernetes versions.
+        # kubernetes-version: ["v1.20.15", "v1.21.12", "v1.22.9", "v1.23.6", "v1.24.1"]
+        kubernetes-version: ["v1.21.12", "v1.22.9", "v1.23.6"]

.github/workflows/mxnet-mnist-e2e-test.yaml

@@ -0,0 +1,39 @@
+name: E2E Test with mxnet-mnist
+on:
+  - pull_request
+
+jobs:
+  e2e:
+    runs-on: ubuntu-20.04
+    timeout-minutes: 120
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v2
+
+      - name: Setup Test Env
+        uses: ./.github/workflows/template-setup-e2e-test
+        with:
+          kubernetes-version: ${{ matrix.kubernetes-version }}
+
+      - name: Run e2e test with ${{ matrix.experiments }} experiments
+        uses: ./.github/workflows/template-e2e-test
+        with:
+          experiments: ${{ matrix.experiments }}
+          # Comma Delimited
+          trial-images: mxnet-mnist
+
+    strategy:
+      fail-fast: false
+      matrix:
+        # Detail: https://hub.docker.com/r/kindest/node
+        # TODO (tenzen-y): We need to consider running tests on more kubernetes versions.
+        # kubernetes-version: ["v1.20.15", "v1.21.12", "v1.22.9", "v1.23.6", "v1.24.1"]
+        kubernetes-version: ["v1.21.12", "v1.22.9", "v1.23.6"]
+        # Comma Delimited
+        experiments:
+          # suggestion-hyperopt
+          - "random,tpe,never-resume"
+          - "median-stop,from-volume-resume"
+          # others
+          - "grid,bayesian-optimization,tpe"
+          - "multivariate-tpe,cma-es,hyperband"

.github/workflows/publish-algorithm-images.yaml

@@ -26,6 +26,7 @@ jobs:
           dockerfile: ${{ matrix.dockerfile }}
 
     strategy:
+      fail-fast: false
       matrix:
         include:
           - component-name: suggestion-hyperopt

.github/workflows/publish-core-images.yaml

@@ -26,6 +26,7 @@ jobs:
           dockerfile: ${{ matrix.dockerfile }}
 
     strategy:
+      fail-fast: false
       matrix:
         include:
           - component-name: katib-controller

.github/workflows/publish-trial-images.yaml

@@ -26,6 +26,7 @@ jobs:
           dockerfile: ${{ matrix.dockerfile }}
 
     strategy:
+      fail-fast: false
       matrix:
         include:
           - trial-name: mxnet-mnist
@@ -38,5 +39,7 @@ jobs:
             dockerfile: examples/v1beta1/trial-images/enas-cnn-cifar10/Dockerfile.gpu
           - trial-name: enas-cnn-cifar10-cpu
             dockerfile: examples/v1beta1/trial-images/enas-cnn-cifar10/Dockerfile.cpu
-          - trial-name: darts-cnn-cifar10
-            dockerfile: examples/v1beta1/trial-images/darts-cnn-cifar10/Dockerfile
+          - trial-name: darts-cnn-cifar10-cpu
+            dockerfile: examples/v1beta1/trial-images/darts-cnn-cifar10/Dockerfile.cpu
+          - trial-name: darts-cnn-cifar10-gpu
+            dockerfile: examples/v1beta1/trial-images/darts-cnn-cifar10/Dockerfile.gpu

.github/workflows/pytorch-mnist-e2e-test.yaml

@@ -0,0 +1,36 @@
+name: E2E Test with pytorch-mnist
+on:
+  - pull_request
+
+jobs:
+  e2e:
+    runs-on: ubuntu-20.04
+    timeout-minutes: 120
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v2
+
+      - name: Setup Test Env
+        uses: ./.github/workflows/template-setup-e2e-test
+        with:
+          kubernetes-version: ${{ matrix.kubernetes-version }}
+
+      - name: Run e2e test with ${{ matrix.experiments }} experiments
+        uses: ./.github/workflows/template-e2e-test
+        with:
+          experiments: ${{ matrix.experiments }}
+          training-operator: true
+          # Comma Delimited
+          trial-images: pytorch-mnist
+
+    strategy:
+      fail-fast: false
+      matrix:
+        # Detail: https://hub.docker.com/r/kindest/node
+        # TODO (tenzen-y): We need to consider running tests on more kubernetes versions.
+        # kubernetes-version: ["v1.20.15", "v1.21.12", "v1.22.9", "v1.23.6", "v1.24.1"]
+        kubernetes-version: ["v1.21.12", "v1.22.9", "v1.23.6"]
+        # Comma Delimited
+        experiments:
+          - "file-metrics-collector,pytorchjob-mnist"
+          - "median-stop-with-json-format,file-metrics-collector-with-json-format"

.github/workflows/template-e2e-test/action.yaml

@@ -0,0 +1,35 @@
+# Template for e2e tests.
+
+inputs:
+  cluster_name:
+    required: false
+    type: string
+    default: katib-e2e-cluster
+  experiments:
+    required: true
+    type: string
+  training-operator:
+    required: false
+    type: boolean
+  trial-images:
+    required: true
+    type: string
+  katib-ui:
+    required: true
+    type: boolean
+    default: false
+
+runs:
+  using: composite
+  steps:
+    - name: Set Up KinD Cluster
+      shell: bash
+      run: ./test/e2e/v1beta1/scripts/gh-actions/setup-kind.sh ${{ inputs.katib-ui }} ${{ inputs.trial-images }} ${{ inputs.cluster_name }} ${{ inputs.experiments }}
+
+    - name: Set Up Katib
+      shell: bash
+      run: ./test/e2e/v1beta1/scripts/gh-actions/setup-katib.sh ${{ inputs.katib-ui }} ${{ inputs.training-operator }}
+
+    - name: Run E2E Experiment
+      shell: bash
+      run: ./test/e2e/v1beta1/scripts/gh-actions/run-e2e-experiment.sh ${{ inputs.experiments }}

.github/workflows/template-setup-e2e-test/action.yaml

@@ -0,0 +1,29 @@
+# Template for e2e tests.
+
+inputs:
+  kubernetes-version:
+    required: true
+    type: string
+  cluster_name:
+    required: false
+    type: string
+    default: katib-e2e-cluster
+
+runs:
+  using: composite
+  steps:
+    - name: Set Up KinD Cluster
+      uses: helm/kind-action@v1.2.0
+      with:
+        version: v0.13.0
+        node_image: kindest/node:${{ inputs.kubernetes-version }}
+        cluster_name: ${{ inputs.cluster_name }}
+        wait: 120s
+
+    - name: Set Up Docker Buildx
+      uses: docker/setup-buildx-action@v1
+
+    - name: Set Up Go env
+      uses: actions/setup-go@v2
+      with:
+        go-version: 1.17.10

.github/workflows/test-go.yaml

@@ -23,7 +23,7 @@ jobs:
       - name: Setup Go
         uses: actions/setup-go@v2
         with:
-          go-version: 1.17.1
+          go-version: 1.17.10
 
       # Verify that go.mod and go.sum is synchronized
       - name: Check Go modules

.github/workflows/tf-mnist-with-summaries-e2e-test.yaml

@@ -0,0 +1,34 @@
+name: E2E Test with tf-mnist-with-summaries
+on:
+  - pull_request
+
+jobs:
+  e2e:
+    runs-on: ubuntu-20.04
+    timeout-minutes: 120
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v2
+
+      - name: Setup Test Env
+        uses: ./.github/workflows/template-setup-e2e-test
+        with:
+          kubernetes-version: ${{ matrix.kubernetes-version }}
+
+      - name: Run e2e test with ${{ matrix.experiments }} experiments
+        uses: ./.github/workflows/template-e2e-test
+        with:
+          experiments: ${{ matrix.experiments }}
+          training-operator: true
+          # Comma Delimited
+          trial-images: tf-mnist-with-summaries
+
+    strategy:
+      fail-fast: false
+      matrix:
+        # Detail: https://hub.docker.com/r/kindest/node
+        # TODO (tenzen-y): We need to consider running tests on more kubernetes versions.
+        # kubernetes-version: ["v1.20.15", "v1.21.12", "v1.22.9", "v1.23.6", "v1.24.1"]
+        kubernetes-version: ["v1.21.12", "v1.22.9", "v1.23.6"]
+        # Comma Delimited
+        experiments: ["tfjob-mnist-with-summaries"]

README.md

@@ -146,7 +146,7 @@ Follow the next steps to install Katib standalone.
 
 This is the minimal requirements to install Katib:
 
-- Kubernetes >= 1.17
+- Kubernetes >= 1.21
 - `kubectl` >= 1.21
 
 ## Latest Version

cmd/earlystopping/medianstop/v1beta1/Dockerfile

@@ -1,4 +1,4 @@
-FROM python:3.9
+FROM python:3.9-slim
 
 ENV TARGET_DIR /opt/katib
 ENV EARLY_STOPPING_DIR cmd/earlystopping/medianstop/v1beta1

cmd/metricscollector/v1beta1/tfevent-metricscollector/Dockerfile

@@ -1,4 +1,4 @@
-FROM python:3.9
+FROM python:3.9-slim
 
 ENV TARGET_DIR /opt/katib
 ENV METRICS_COLLECTOR_DIR cmd/metricscollector/v1beta1/tfevent-metricscollector

cmd/suggestion/chocolate/v1beta1/Dockerfile

@@ -1,15 +1,5 @@
-FROM python:3.9
-
-ENV TARGET_DIR /opt/katib
-ENV SUGGESTION_DIR cmd/suggestion/chocolate/v1beta1
+FROM alpine:3.15 AS downloader
 ENV GRPC_HEALTH_PROBE_VERSION v0.4.11
-
-RUN if [ "$(uname -m)" = "ppc64le" ] || [ "$(uname -m)" = "aarch64" ]; then \
-    apt-get -y update && \
-    apt-get -y install gfortran libopenblas-dev liblapack-dev && \
-    apt-get clean && \
-    rm -rf /var/lib/apt/lists/*; \
-    fi
 RUN if [ "$(uname -m)" = "ppc64le" ]; then \
     wget -qO/bin/grpc_health_probe https://github.com/grpc-ecosystem/grpc-health-probe/releases/download/${GRPC_HEALTH_PROBE_VERSION}/grpc_health_probe-linux-ppc64le; \
     elif [ "$(uname -m)" = "aarch64" ]; then \
@@ -19,8 +9,22 @@ RUN if [ "$(uname -m)" = "ppc64le" ]; then \
     fi && \
     chmod +x /bin/grpc_health_probe
 
+FROM python:3.9-slim
+ENV TARGET_DIR /opt/katib
+ENV SUGGESTION_DIR cmd/suggestion/chocolate/v1beta1
+
+RUN apt-get -y update && \
+    apt-get -y install git && \
+    if [ "$(uname -m)" = "ppc64le" ] || [ "$(uname -m)" = "aarch64" ]; then \
+    apt-get -y install gfortran libopenblas-dev liblapack-dev; \
+    fi && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists/*
+
 ADD ./pkg/ ${TARGET_DIR}/pkg/
 ADD ./${SUGGESTION_DIR}/ ${TARGET_DIR}/${SUGGESTION_DIR}/
+COPY --from=downloader /bin/grpc_health_probe /bin/grpc_health_probe
+
 WORKDIR  ${TARGET_DIR}/${SUGGESTION_DIR}
 RUN pip install --no-cache-dir -r requirements.txt