Files
GenAIExamples/.github/workflows/_helm-e2e.yml
2025-05-12 17:41:57 +08:00

253 lines
9.3 KiB
YAML

# Copyright (C) 2024 Intel Corporation
# SPDX-License-Identifier: Apache-2.0
name: Helm Chart E2e Test For Call
permissions:
contents: read
on:
workflow_call:
inputs:
example:
default: "chatqna"
required: true
type: string
description: "example to test, chatqna or common/asr"
hardware:
default: "xeon"
required: true
type: string
dockerhub:
default: "false"
required: false
type: string
description: "Set to true if you want to use released docker images at dockerhub. By default using internal docker registry."
mode:
default: "CD"
description: "Whether the test range is CI, CD or CICD"
required: false
type: string
tag:
default: "latest"
required: false
type: string
version:
default: "0-latest"
required: false
type: string
jobs:
get-test-case:
runs-on: ubuntu-latest
outputs:
value_files: ${{ steps.get-test-files.outputs.value_files }}
CHECKOUT_REF: ${{ steps.get-checkout-ref.outputs.CHECKOUT_REF }}
steps:
- name: Get checkout ref
id: get-checkout-ref
run: |
if [ "${{ github.event_name }}" == "pull_request" ] || [ "${{ github.event_name }}" == "pull_request_target" ]; then
CHECKOUT_REF=refs/pull/${{ github.event.number }}/merge
else
CHECKOUT_REF=${{ github.ref }}
fi
echo "CHECKOUT_REF=${CHECKOUT_REF}" >> $GITHUB_OUTPUT
echo "checkout ref ${CHECKOUT_REF}"
- name: Checkout Repo
uses: actions/checkout@v4
with:
ref: ${{ steps.get-checkout-ref.outputs.CHECKOUT_REF }}
fetch-depth: 0
- name: Get test Services
id: get-test-files
run: |
set -x
if [ "${{ inputs.mode }}" = "CI" ]; then
base_commit=${{ github.event.pull_request.base.sha }}
merged_commit=$(git log -1 --format='%H')
values_files=$(git diff --name-only ${base_commit} ${merged_commit} | \
grep "${{ inputs.example }}/kubernetes/helm" | \
grep "values.yaml" |\
sort -u)
echo $values_files
elif [ "${{ inputs.mode }}" = "CD" ]; then
values_files=$(ls ${{ inputs.example }}/kubernetes/helm/*values.yaml || true)
fi
value_files="["
for file in ${values_files}; do
if [ -f "$file" ]; then
filename=$(basename "$file")
if [[ "$filename" == *"gaudi"* ]]; then
if [[ "${{ inputs.hardware }}" == "gaudi" ]]; then
value_files="${value_files}\"${filename}\","
fi
elif [[ "$filename" == *"rocm"* ]]; then
if [[ "${{ inputs.hardware }}" == "rocm" ]]; then
value_files="${value_files}\"${filename}\","
fi
elif [[ "$filename" == *"nv"* ]]; then
continue
else
if [[ "${{ inputs.hardware }}" == "xeon" ]]; then
value_files="${value_files}\"${filename}\","
fi
fi
fi
done
value_files="${value_files%,}]"
echo "value_files=${value_files}"
echo "value_files=${value_files}" >> $GITHUB_OUTPUT
helm-test:
needs: [get-test-case]
if: ${{ needs.get-test-case.outputs.value_files != '[]' }}
strategy:
matrix:
value_file: ${{ fromJSON(needs.get-test-case.outputs.value_files) }}
fail-fast: false
runs-on: k8s-${{ inputs.hardware }}
continue-on-error: true
steps:
- name: Clean Up Working Directory
run: |
echo "value_file=${{ matrix.value_file }}"
sudo rm -rf ${{github.workspace}}/*
- name: Get checkout ref
id: get-checkout-ref
run: |
if [ "${{ github.event_name }}" == "pull_request" ] || [ "${{ github.event_name }}" == "pull_request_target" ]; then
CHECKOUT_REF=refs/pull/${{ github.event.number }}/merge
else
CHECKOUT_REF=${{ github.ref }}
fi
echo "CHECKOUT_REF=${CHECKOUT_REF}" >> $GITHUB_OUTPUT
echo "checkout ref ${CHECKOUT_REF}"
- name: Checkout Repo
uses: actions/checkout@v4
with:
ref: ${{ steps.get-checkout-ref.outputs.CHECKOUT_REF }}
fetch-depth: 0
- name: Set variables
env:
example: ${{ inputs.example }}
run: |
if [[ ! "$example" =~ ^[a-zA-Z0-9]{1,20}$ ]] || [[ "$example" =~ \.\. ]] || [[ "$example" == -* || "$example" == *- ]]; then
echo "Error: Invalid input - only lowercase alphanumeric and internal hyphens allowed"
exit 1
fi
# SAFE_PREFIX="kb-"
CHART_NAME="${SAFE_PREFIX}$(echo "$example" | tr '[:upper:]' '[:lower:]')"
RAND_SUFFIX=$(openssl rand -hex 2 | tr -dc 'a-f0-9')
cat <<EOF >> $GITHUB_ENV
CHART_NAME=${CHART_NAME}
RELEASE_NAME=${CHART_NAME}-$(date +%s)
NAMESPACE=ns-${CHART_NAME}-${RAND_SUFFIX}
ROLLOUT_TIMEOUT_SECONDS=600s
TEST_TIMEOUT_SECONDS=600s
KUBECTL_TIMEOUT_SECONDS=60s
should_cleanup=false
skip_validate=false
CHART_FOLDER=${example}/kubernetes/helm
EOF
echo "Generated safe variables:" >> $GITHUB_STEP_SUMMARY
echo "- CHART_NAME: ${CHART_NAME}" >> $GITHUB_STEP_SUMMARY
- name: Helm install
id: install
env:
GOOGLE_CSE_ID: ${{ secrets.GOOGLE_CSE_ID }}
GOOGLE_API_KEY: ${{ secrets.GOOGLE_API_KEY }}
HUGGINGFACEHUB_API_TOKEN: ${{ secrets.HUGGINGFACEHUB_API_TOKEN }}
HFTOKEN: ${{ secrets.HUGGINGFACEHUB_API_TOKEN }}
value_file: ${{ matrix.value_file }}
run: |
set -xe
echo "should_cleanup=true" >> $GITHUB_ENV
if [[ ! -f ${{ github.workspace }}/${{ env.CHART_FOLDER }}/${value_file} ]]; then
echo "No value file found, exiting test!"
echo "skip_validate=true" >> $GITHUB_ENV
echo "should_cleanup=false" >> $GITHUB_ENV
exit 0
fi
for img in `helm template -n $NAMESPACE $RELEASE_NAME oci://ghcr.io/opea-project/charts/${CHART_NAME} -f ${{ inputs.example }}/kubernetes/helm/${value_file} --version ${{ inputs.version }} | grep 'image:' | grep 'opea/' | awk '{print $2}' | xargs`;
do
# increase helm install wait for for vllm-gaudi case
if [[ $img == *"vllm-gaudi"* ]]; then
ROLLOUT_TIMEOUT_SECONDS=900s
fi
done
if ! helm install \
--create-namespace \
--namespace $NAMESPACE \
$RELEASE_NAME \
oci://ghcr.io/opea-project/charts/${CHART_NAME} \
--set global.HUGGINGFACEHUB_API_TOKEN=${HFTOKEN} \
--set global.modelUseHostPath=/data2/hf_model \
--set GOOGLE_API_KEY=${{ env.GOOGLE_API_KEY}} \
--set GOOGLE_CSE_ID=${{ env.GOOGLE_CSE_ID}} \
--set web-retriever.GOOGLE_API_KEY=${{ env.GOOGLE_API_KEY}} \
--set web-retriever.GOOGLE_CSE_ID=${{ env.GOOGLE_CSE_ID}} \
-f ${{ inputs.example }}/kubernetes/helm/${value_file} \
--version ${{ inputs.version }} \
--wait --timeout "$ROLLOUT_TIMEOUT_SECONDS"; then
echo "Failed to install chart ${{ inputs.example }}"
echo "skip_validate=true" >> $GITHUB_ENV
.github/workflows/scripts/k8s-utils.sh dump_pods_status $NAMESPACE
exit 1
fi
- name: Validate e2e test
if: always()
run: |
set -xe
if $skip_validate; then
echo "Skip validate"
else
LOG_PATH=/home/$(whoami)/helm-logs
chart=${{ env.CHART_NAME }}
helm test -n $NAMESPACE $RELEASE_NAME --logs --timeout "$TEST_TIMEOUT_SECONDS" | tee ${LOG_PATH}/charts-${chart}.log
exit_code=$?
if [ $exit_code -ne 0 ]; then
echo "Chart ${chart} test failed, please check the logs in ${LOG_PATH}!"
exit 1
fi
echo "Checking response results, make sure the output is reasonable. "
teststatus=false
if [[ -f $LOG_PATH/charts-${chart}.log ]] && \
[[ $(grep -c "^Phase:.*Failed" $LOG_PATH/charts-${chart}.log) != 0 ]]; then
teststatus=false
${{ github.workspace }}/.github/workflows/scripts/k8s-utils.sh dump_all_pod_logs $NAMESPACE
else
teststatus=true
fi
if [ $teststatus == false ]; then
echo "Response check failed, please check the logs in artifacts!"
exit 1
else
echo "Response check succeeded!"
exit 0
fi
fi
- name: Helm uninstall
if: always()
run: |
if $should_cleanup; then
helm uninstall $RELEASE_NAME --namespace $NAMESPACE
if ! kubectl delete ns $NAMESPACE --timeout=$KUBECTL_TIMEOUT_SECONDS; then
kubectl delete pods --namespace $NAMESPACE --force --grace-period=0 --all
kubectl delete ns $NAMESPACE --force --grace-period=0 --timeout=$KUBECTL_TIMEOUT_SECONDS
fi
fi