ci: Use OIDC to log into Azure

This completely eliminates the Azure secret from the repo, following the below
guidance:

https://docs.github.com/en/actions/security-for-github-actions/security-hardening-your-deployments/configuring-openid-connect-in-azure

The federated identity is scoped to the `ci` environment, meaning:

 * I had to specify this environment in some YAMLs. I don't believe there's any
   downside to this.
 * As previously, the CI works seamlessly both from PRs and in the manual
   workflow.

I also deleted the tools/packaging/kata-deploy/action folder as it doesn't seem
to be used anymore, and it contains a reference to the secret.

Signed-off-by: Aurélien Bombo <abombo@microsoft.com>
This commit is contained in:
Aurélien Bombo
2025-06-05 19:43:56 -05:00
parent 31a8944da1
commit 9dd3807467
21 changed files with 37 additions and 392 deletions

View File

@@ -1,31 +0,0 @@
# Copyright (c) 2019 Intel Corporation
#
# SPDX-License-Identifier: Apache-2.0
FROM mcr.microsoft.com/azure-cli:2.9.1
LABEL com.github.actions.name="Test kata-deploy in an AKS cluster"
LABEL com.github.actions.description="Test kata-deploy in an AKS cluster"
# Default to latest validated AKS-engine version
ARG AKS_ENGINE_VER="v0.62.0"
ARG ARCH=amd64
ENV GITHUB_ACTION_NAME="Test kata-deploy in an AKS cluster"
# When run, we expect the caller (GitHub Action workflow) to provide the
# PKG_SHA environment variable
ENV PKG_SHA=HEAD
RUN curl -LO "https://dl.k8s.io/release/$(curl -s https://dl.k8s.io/release/stable.txt)/bin/linux/${ARCH}/kubectl" \
&& chmod +x ./kubectl \
&& mv ./kubectl /usr/local/bin/kubectl
RUN curl -LO "https://github.com/Azure/aks-engine/releases/download/${AKS_ENGINE_VER}/aks-engine-${AKS_ENGINE_VER}-linux-${ARCH}.tar.gz" \
&& tar xvf "aks-engine-${AKS_ENGINE_VER}-linux-${ARCH}.tar.gz" \
&& mv "aks-engine-${AKS_ENGINE_VER}-linux-${ARCH}/aks-engine" /usr/local/bin/aks-engine \
&& rm "aks-engine-${AKS_ENGINE_VER}-linux-${ARCH}.tar.gz"
COPY kubernetes-containerd.json /
COPY setup-aks.sh test-kata.sh entrypoint.sh /
ENTRYPOINT ["/entrypoint.sh"]

View File

@@ -1,13 +0,0 @@
# action.yml
name: 'kata-deploy-aks'
description: 'test Kata container image in AKS'
inputs:
packaging-sha:
description: 'SHA we are using for pulling packaging manifests'
required: true
default: ''
runs:
using: 'docker'
image: 'Dockerfile'
args:
- ${{ inputs.packaging-sha }}

View File

@@ -1,24 +0,0 @@
#!/usr/bin/env bash
#
# Copyright (c) 2018 Intel Corporation
#
# SPDX-License-Identifier: Apache-2.0
#
set -o errexit
set -o pipefail
set -o nounset
# This entrypoint expects an environment variable, PKG_SHA, to be
# within the container runtime. A default is provided in the Dockerfile,
# but we expect the caller to pass this into the container run (ie docker run -e PKG_SHA=foo ...)
echo "provided package reference: ${PKG_SHA}"
# Since this is the entrypoint for the container image, we know that the AKS and Kata setup/testing
# scripts are located at root.
source /setup-aks.sh
source /test-kata.sh
trap destroy_aks EXIT
setup_aks
test_kata

View File

@@ -1,43 +0,0 @@
{
"apiVersion": "vlabs",
"properties": {
"orchestratorProfile": {
"orchestratorType": "Kubernetes",
"orchestratorVersion": "1.20.5",
"kubernetesConfig": {
"containerRuntime": "containerd",
"useManagedIdentity": false
}
},
"masterProfile": {
"count": 1,
"dnsPrefix": "",
"vmSize": "Standard_D2_v2"
},
"agentPoolProfiles": [
{
"name": "agentpool",
"count": 1,
"vmSize": "Standard_D4s_v3",
"availabilityProfile": "AvailabilitySet"
}
],
"linuxProfile": {
"adminUsername": "azureuser",
"ssh": {
"publicKeys": [
{
"keyData": ""
}
]
}
},
"servicePrincipalProfile": {
"clientId": "",
"secret": ""
},
"linuxProfile": {
"runUnattendedUpgradesOnBootstrap": false
}
}
}

View File

@@ -1,51 +0,0 @@
#!/usr/bin/env bash
#
# Copyright (c) 2019 Intel Corporation
#
# SPDX-License-Identifier: Apache-2.0
#
set -o errexit
set -o pipefail
set -o nounset
export AZURE_HTTP_USER_AGENT="GITHUBACTIONS_${GITHUB_ACTION_NAME}_${GITHUB_REPOSITORY}"
LOCATION=${LOCATION:-westus2}
DNS_PREFIX=${DNS_PREFIX:-kata-deploy-${GITHUB_SHA:0:10}}
CLUSTER_CONFIG=${CLUSTER_CONFIG:-/kubernetes-containerd.json}
function die() {
msg="$*"
echo "ERROR: $msg" >&2
exit 1
}
function destroy_aks() {
set +x
export KUBECONFIG="${PWD}/_output/${DNS_PREFIX}/kubeconfig/kubeconfig.${LOCATION}.json"
az login --service-principal -u "$AZ_APPID" -p "$AZ_PASSWORD" --tenant "$AZ_TENANT_ID"
az group delete --name "$DNS_PREFIX" --yes --no-wait
az logout
}
function setup_aks() {
[[ -z "$AZ_APPID" ]] && die "no Azure service principal ID provided"
[[ -z "$AZ_PASSWORD" ]] && die "no Azure service principal secret provided"
[[ -z "$AZ_SUBSCRIPTION_ID" ]] && die "no Azure subscription ID provided"
[[ -z "$AZ_TENANT_ID" ]] && die "no Azure tenant ID provided"
aks-engine deploy --subscription-id "$AZ_SUBSCRIPTION_ID" \
--client-id "$AZ_APPID" --client-secret "$AZ_PASSWORD" \
--location "$LOCATION" --dns-prefix "$DNS_PREFIX" \
--api-model "$CLUSTER_CONFIG" --force-overwrite
export KUBECONFIG="${PWD}/_output/${DNS_PREFIX}/kubeconfig/kubeconfig.${LOCATION}.json"
# wait for the cluster to be settled:
kubectl wait --timeout=10m --for=condition=Ready --all nodes
# make sure coredns is up before moving forward:
kubectl wait --timeout=10m -n kube-system --for=condition=Available deployment/coredns
}

View File

@@ -1,162 +0,0 @@
#!/usr/bin/env bash
#
# Copyright (c) 2019 Intel Corporation
#
# SPDX-License-Identifier: Apache-2.0
#
set -o errexit
set -o pipefail
set -o nounset
function die() {
msg="$*"
echo "ERROR: $msg" >&2
exit 1
}
function waitForProcess() {
wait_time="$1"
cmd="$2"
sleep_time=5
echo "waiting for process $cmd"
while [ "$wait_time" -gt 0 ]; do
if eval "$cmd"; then
return 0
else
sleep "$sleep_time"
wait_time=$((wait_time-sleep_time))
fi
done
return 1
}
# waitForLabelRemoval will wait for the kata-runtime labels to removed until a given
# timeout expires
function waitForLabelRemoval() {
wait_time="$1"
sleep_time=5
echo "waiting for kata-runtime label to be removed"
while [[ "$wait_time" -gt 0 ]]; do
# if a node is found which matches node-select, the output will include a column for node name,
# NAME. Let's look for that
if [[ -z $(kubectl get nodes --selector katacontainers.io/kata-runtime 2>&1 | grep NAME) ]]
then
return 0
else
sleep "$sleep_time"
wait_time=$((wait_time-sleep_time))
fi
done
echo $(kubectl get pods,nodes --all-namespaces --show-labels)
echo "failed to cleanup"
return 1
}
function run_test() {
YAMLPATH="./tools/packaging/kata-deploy/"
echo "verify connectivity with a pod using Kata"
deployment=""
busybox_pod="test-nginx"
busybox_image="busybox"
cmd="kubectl get pods | grep $busybox_pod | grep Completed"
wait_time=120
configurations=("nginx-deployment-qemu" "nginx-deployment-clh" "nginx-deployment-dragonball")
for deployment in "${configurations[@]}"; do
# start the kata pod:
kubectl apply -f "$YAMLPATH/examples/${deployment}.yaml"
# in case the control plane is slow, give it a few seconds to accept the yaml, otherwise
# our 'wait' for deployment status will fail to find the deployment at all
sleep 3
kubectl wait --timeout=5m --for=condition=Available deployment/${deployment} || kubectl describe pods
kubectl expose deployment/${deployment}
# test pod connectivity:
kubectl run $busybox_pod --restart=Never --image="$busybox_image" -- wget --timeout=5 "$deployment"
waitForProcess "$wait_time" "$cmd"
kubectl logs "$busybox_pod" | grep "index.html"
kubectl describe pod "$busybox_pod"
# cleanup:
kubectl delete deployment "$deployment"
kubectl delete service "$deployment"
kubectl delete pod "$busybox_pod"
done
}
function test_kata() {
set -x
[[ -z "$PKG_SHA" ]] && die "no PKG_SHA provided"
YAMLPATH="./tools/packaging/kata-deploy/"
# This action could be called in two contexts:
# 1. Packaging workflows: testing in packaging repository, where we assume yaml/packaging
# bits under test are already part of teh action workspace.
# 2. From kata-containers: when creating a release, the appropriate packaging repository is
# not yet part of the workspace, and we will need to clone
if [[ ! -d $YAMLPATH ]]; then
[[ -d $YAMLPATH ]] || git clone https://github.com/kata-containers/kata-containers
cd kata-containers
git fetch
git checkout $PKG_SHA
fi
kubectl apply -f "$YAMLPATH/kata-rbac/base/kata-rbac.yaml"
# apply runtime classes:
kubectl apply -f "$YAMLPATH/runtimeclasses/kata-runtimeClasses.yaml"
kubectl get runtimeclasses
# update deployment daemonset to utilize the container under test:
sed -i "s#quay.io/kata-containers/kata-deploy:latest#quay.io/kata-containers/kata-deploy-ci:${PKG_SHA}#g" $YAMLPATH/kata-deploy/base/kata-deploy.yaml
sed -i "s#quay.io/kata-containers/kata-deploy:latest#quay.io/kata-containers/kata-deploy-ci:${PKG_SHA}#g" $YAMLPATH/kata-cleanup/base/kata-cleanup.yaml
cat $YAMLPATH/kata-deploy/base/kata-deploy.yaml
# deploy kata:
kubectl apply -f $YAMLPATH/kata-deploy/base/kata-deploy.yaml
# in case the control plane is slow, give it a few seconds to accept the yaml, otherwise
# our 'wait' for deployment status will fail to find the deployment at all. If it can't persist
# the daemonset to etcd in 30 seconds... then we'll fail.
sleep 30
# wait for kata-deploy to be up
kubectl -n kube-system wait --timeout=10m --for=condition=Ready -l name=kata-deploy pod
# show running pods, and labels of nodes
kubectl get pods,nodes --all-namespaces --show-labels
run_test
kubectl get pods,nodes --show-labels
# Remove Kata
kubectl delete -f $YAMLPATH/kata-deploy/base/kata-deploy.yaml
kubectl -n kube-system wait --timeout=10m --for=delete -l name=kata-deploy pod
kubectl get pods,nodes --show-labels
kubectl apply -f $YAMLPATH/kata-cleanup/base/kata-cleanup.yaml
# The cleanup daemonset will run a single time, since it will clear the node-label. Thus, its difficult to
# check the daemonset's status for completion. instead, let's wait until the kata-runtime labels are removed
# from all of the worker nodes. If this doesn't happen after 2 minutes, let's fail
timeout=120
waitForLabelRemoval $timeout
kubectl delete -f $YAMLPATH/kata-cleanup/base/kata-cleanup.yaml
set +x
}