test(helm): add reproducible upgrade smoke coverage

Signed-off-by: XiaoSeS <87064762+XiaoSeS@users.noreply.github.com>

.github/workflows/pr-helm-chart.yml

@@ -180,3 +180,28 @@ jobs:
         with:
           entrypoint: '/kubeconform'
           args: "-strict -summary -output text -schema-location default -schema-location https://raw.githubusercontent.com/datreeio/CRDs-catalog/main/{{.Group}}/{{.ResourceKind}}_{{.ResourceAPIVersion}}.json charts/skillhub/rendered.yaml"
+
+  install-upgrade:
+    name: Install and Upgrade Smoke
+    if: ${{ github.event_name != 'pull_request' || !github.event.pull_request.draft }}
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Check out repository
+        uses: actions/checkout@v4
+        with:
+          persist-credentials: false
+
+      - name: Set up Helm
+        uses: azure/setup-helm@v4
+        with:
+          version: v3.19.0
+
+      - name: Create Kubernetes cluster
+        uses: helm/kind-action@ef37e7f390d99f746eb8b610417061a60e82a6cc # v1.14.0
+        with:
+          cluster_name: skillhub-helm-smoke
+          wait: 120s
+
+      - name: Run install and upgrade smoke
+        run: bash charts/skillhub/tests/install-upgrade-smoke.sh

README.md

@@ -343,6 +343,7 @@ Basic Kubernetes manifests are available under [`deploy/k8s/`](./deploy/k8s):
 - `backend-deployment.yaml`
 - `frontend-deployment.yaml`
 - `services.yaml`
+- `ingress.yaml`
 
 For a configurable deployment with bundled PostgreSQL and Redis dependencies,
 use the Helm chart under [`charts/skillhub/`](./charts/skillhub):
@@ -357,7 +358,6 @@ helm upgrade --install skillhub ./charts/skillhub \
 
 See the [Helm chart guide](./charts/skillhub/README.md) for required secrets,
 Ingress/TLS, external data services, storage migration, and upgrade constraints.
-- `ingress.yaml`
 
 Apply them after creating your own secret:
 

charts/skillhub/README.md

@@ -54,6 +54,16 @@ helm -n skillhub upgrade -i skillhub ./charts/skillhub \
 > `ingress.tls[]`。旧的 `ingress.host`、`ingress.tls.enabled` 与
 > `ingress.tls.secretName` 不再接受，升级前必须改成本文 Ingress 示例中的数组结构。
 
+合并或发布前，可在一个空的测试 Kubernetes 集群中运行可重复的安装/升级 smoke：
+
+```bash
+bash charts/skillhub/tests/install-upgrade-smoke.sh
+```
+
+脚本验证 `install -> Ready -> HTTP health -> upgrade -> Ready`，并确认 Secret 数据、
+PVC UID 与绑定 PV 在升级前后保持不变。默认清理自己创建的 namespace；设置
+`KEEP_HELM_SMOKE=true` 可保留现场用于排查。
+
 ### 高可用模式
 
 ```bash
@@ -134,6 +144,11 @@ redis:
 可预测密码，而是在任何随机密码缺失时终止渲染并指出具体配置项。敏感值应放在
 受保护的 values、External Secrets、Sealed Secrets 或密钥注入插件中。
 
+内置 PostgreSQL、Redis、Sentinel 及 metrics exporter 镜像默认使用不可变 digest，
+避免 Bitnami 子 Chart 的 `latest` 默认值造成不可复现的安装和回滚。覆盖私有镜像
+仓库或 tag 时，必须同时把对应的 `image.digest` 设为空，或改成私有仓库中该镜像
+的真实 digest；digest 非空时会优先于 tag。
+
 ## 配置参考
 
 ### 副本数配置
@@ -207,27 +222,32 @@ postgresql:
     registry: registry.example.com
     repository: library/postgresql
     tag: 18.4.0
+    digest: ""
   metrics:
     image:
       registry: registry.example.com
       repository: library/postgres-exporter
       tag: 0.20.1
+      digest: ""
 
 redis:
   image:
     registry: registry.example.com
     repository: library/redis
     tag: 8.8.0
+    digest: ""
   sentinel:
     image:
       registry: registry.example.com
       repository: library/redis-sentinel
       tag: 8.8.0
+      digest: ""
   metrics:
     image:
       registry: registry.example.com
       repository: library/redis-exporter
       tag: 1.86.0
+      digest: ""
 ```
 
 `global.security.allowInsecureImages` 是 Bitnami 对自定义镜像仓库和镜像名称的校验

charts/skillhub/templates/validate.yaml

@@ -1,4 +1,5 @@
 {{- /* Cross-field validation that JSON Schema cannot express reliably. */ -}}
+{{- $absoluteHttpUrlPattern := "^https?://(\\[[0-9A-Fa-f:.]+\\]|[A-Za-z0-9]([A-Za-z0-9.-]*[A-Za-z0-9])?)(:[0-9]{1,5})?([/?#][^[:space:]]*)?$" -}}
 {{- if not .Values.server.enabled -}}
 {{- fail "server.enabled=false is unsupported because the bundled web component requires the SkillHub server" -}}
 {{- end -}}
@@ -17,7 +18,7 @@
 {{- end -}}
 {{- range $host := .Values.ingress.hosts -}}
 {{- range $path := $host.paths -}}
-{{- if regexMatch "^/(api|oauth2|login/oauth2|\\.well-known)(?:/|$)" $path.path -}}
+{{- if regexMatch "^/(api|oauth2|login/oauth2|\\.well-known)(/|$)" $path.path -}}
 {{- fail "ingress.hosts[].paths reserves /api, /oauth2, /login/oauth2 and /.well-known for the SkillHub server" -}}
 {{- end -}}
 {{- end -}}
@@ -61,10 +62,10 @@
 {{- if and (not .Values.redis.enabled) .Values.externalRedis.sentinel.enabled (not .Values.externalRedis.sentinel.nodes) -}}
 {{- fail "external Redis Sentinel requires at least one externalRedis.sentinel.nodes entry" -}}
 {{- end -}}
-{{- if and .Values.s3.endpoint (not (regexMatch "^https?://.+" .Values.s3.endpoint)) -}}
+{{- if and .Values.s3.endpoint (not (regexMatch $absoluteHttpUrlPattern .Values.s3.endpoint)) -}}
 {{- fail "s3.endpoint must be an absolute HTTP(S) URL" -}}
 {{- end -}}
-{{- if and .Values.s3.publicEndpoint (not (regexMatch "^https?://.+" .Values.s3.publicEndpoint)) -}}
+{{- if and .Values.s3.publicEndpoint (not (regexMatch $absoluteHttpUrlPattern .Values.s3.publicEndpoint)) -}}
 {{- fail "s3.publicEndpoint must be an absolute HTTP(S) URL" -}}
 {{- end -}}
 

charts/skillhub/tests/configuration-contracts.sh

@@ -27,6 +27,13 @@ render verify "$CHART_DIR" >"$TMP_DIR/default.yaml"
 grep -Fq 'name: POSTGRESQL_MAX_CONNECTIONS' "$TMP_DIR/default.yaml"
 grep -Fq 'value: "verify-postgresql"' "$TMP_DIR/default.yaml"
 grep -Fq 'value: "verify-redis-master"' "$TMP_DIR/default.yaml"
+grep -Fq 'bitnami/postgresql@sha256:52e4eae10959a7ef4810f2d0c9ce656e9337090fd2427d8d15d2e1335ba4f8fb' "$TMP_DIR/default.yaml"
+grep -Fq 'bitnami/postgres-exporter@sha256:fbd5fbc3f350c793efef777b0710f17019e8dd7812fc7c7396b13aa5130f3bd9' "$TMP_DIR/default.yaml"
+grep -Fq 'bitnami/redis@sha256:5ff436b8a0dbe3808f03c7b682865cd30ddb1f08c1a8d17970ebf475f5689722' "$TMP_DIR/default.yaml"
+grep -Fq 'bitnami/redis-exporter@sha256:0468a9a14ebd96953f967f9f3e8673c3eb1ad1dd54d379fef15c3f3ee714c99e' "$TMP_DIR/default.yaml"
+if grep -Eq 'image:.*:latest([@"[:space:]]|$)' "$TMP_DIR/default.yaml"; then
+  fail "default workloads must not use mutable latest image tags"
+fi
 grep -Fq 'fsGroup: 101' "$TMP_DIR/default.yaml"
 grep -Fq 'fsGroupChangePolicy: OnRootMismatch' "$TMP_DIR/default.yaml"
 grep -Fq 'type: Recreate' "$TMP_DIR/default.yaml"
@@ -104,6 +111,13 @@ grep -Fq 'name: SPRING_DATA_REDIS_PASSWORD' "$TMP_DIR/sentinel.yaml"
 grep -Fq 'name: SPRING_DATA_REDIS_SENTINEL_PASSWORD' "$TMP_DIR/sentinel.yaml"
 grep -A1 -F 'name: SKILLHUB_REDIS_SENTINEL_CHECK_SENTINELS_LIST' "$TMP_DIR/sentinel.yaml" \
   | grep -Fq 'value: "false"'
+render sentinel-full "$CHART_DIR" \
+  --set redis.architecture=replication \
+  --set redis.sentinel.enabled=true >"$TMP_DIR/sentinel-full.yaml"
+grep -Fq 'bitnami/redis-sentinel@sha256:667bfef3e22ce9910bdcf304b1b7c659cfc15390a2d473187eb3caac2537d0e6' "$TMP_DIR/sentinel-full.yaml"
+if grep -Eq 'image:.*:latest([@"[:space:]]|$)' "$TMP_DIR/sentinel-full.yaml"; then
+  fail "Sentinel workloads must not use mutable latest image tags"
+fi
 
 render external-sentinel "$CHART_DIR" \
   --set postgresql.enabled=false \
@@ -221,8 +235,13 @@ assert_rejected old-ingress-tls-object --set ingress.tls.enabled=true
 assert_rejected reserved-oauth-ingress-path \
   --set ingress.enabled=true \
   --set-json 'ingress.hosts=[{"host":"skills.example.com","paths":[{"path":"/oauth2","pathType":"Prefix"}]}]'
+assert_rejected reserved-oauth-ingress-child-path \
+  --set ingress.enabled=true \
+  --set-json 'ingress.hosts=[{"host":"skills.example.com","paths":[{"path":"/login/oauth2/code/github","pathType":"Prefix"}]}]'
 assert_rejected invalid-s3-endpoint --set s3.endpoint=s3.amazonaws.com
 assert_rejected invalid-s3-public-endpoint --set s3.publicEndpoint=cdn.example.com
+assert_rejected invalid-s3-empty-authority --set-string 's3.endpoint=https://?'
+assert_rejected invalid-s3-whitespace-authority --set-string 's3.publicEndpoint=https:// '
 assert_rejected empty-ingress-hosts --set-json 'ingress.hosts=[]'
 assert_rejected cert-manager-without-tls \
   --set ingress.enabled=true \

charts/skillhub/tests/install-upgrade-smoke.sh

@@ -0,0 +1,169 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+CHART_DIR=$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)
+TEST_VALUES="$CHART_DIR/tests/test-values.yaml"
+NAMESPACE=${HELM_SMOKE_NAMESPACE:-skillhub-helm-smoke}
+RELEASE=${HELM_SMOKE_RELEASE:-skillhub-smoke}
+TIMEOUT=${HELM_SMOKE_TIMEOUT:-15m}
+KEEP_ENVIRONMENT=${KEEP_HELM_SMOKE:-false}
+TMP_DIR=$(mktemp -d)
+PORT_FORWARD_PID=""
+OWNS_NAMESPACE=false
+
+fail() {
+  echo "ERROR: $*" >&2
+  exit 1
+}
+
+for command in helm kubectl curl jq sha256sum; do
+  command -v "$command" >/dev/null 2>&1 || fail "$command is required"
+done
+
+if kubectl get namespace "$NAMESPACE" >/dev/null 2>&1; then
+  fail "namespace $NAMESPACE already exists; choose an unused HELM_SMOKE_NAMESPACE"
+fi
+OWNS_NAMESPACE=true
+
+stop_port_forward() {
+  if [[ -n "$PORT_FORWARD_PID" ]]; then
+    kill "$PORT_FORWARD_PID" >/dev/null 2>&1 || true
+    wait "$PORT_FORWARD_PID" >/dev/null 2>&1 || true
+    PORT_FORWARD_PID=""
+  fi
+}
+
+cleanup() {
+  local exit_code=$?
+  trap - EXIT
+  stop_port_forward
+
+  if (( exit_code != 0 )) && kubectl get namespace "$NAMESPACE" >/dev/null 2>&1; then
+    echo "Helm smoke failed; collecting non-secret diagnostics" >&2
+    helm status "$RELEASE" --namespace "$NAMESPACE" >&2 || true
+    kubectl get pods,pvc,deployments,statefulsets --namespace "$NAMESPACE" -o wide >&2 || true
+    kubectl get events --namespace "$NAMESPACE" --sort-by=.lastTimestamp >&2 || true
+  fi
+
+  if [[ "$KEEP_ENVIRONMENT" != "true" && "$OWNS_NAMESPACE" == "true" ]]; then
+    helm uninstall "$RELEASE" --namespace "$NAMESPACE" --wait >/dev/null 2>&1 || true
+    kubectl delete namespace "$NAMESPACE" --wait --timeout=5m >/dev/null 2>&1 || true
+  fi
+
+  rm -rf "$TMP_DIR"
+  exit "$exit_code"
+}
+trap cleanup EXIT
+
+probe_service() {
+  local service=$1
+  local service_port=$2
+  local local_port=$3
+  local path=$4
+  local log_file="$TMP_DIR/${service}.port-forward.log"
+
+  stop_port_forward
+  kubectl port-forward \
+    --namespace "$NAMESPACE" \
+    "service/$service" \
+    "$local_port:$service_port" >"$log_file" 2>&1 &
+  PORT_FORWARD_PID=$!
+
+  for _ in $(seq 1 60); do
+    if curl --fail --silent --show-error "http://127.0.0.1:$local_port$path" >/dev/null; then
+      stop_port_forward
+      return 0
+    fi
+    if ! kill -0 "$PORT_FORWARD_PID" >/dev/null 2>&1; then
+      break
+    fi
+    sleep 1
+  done
+
+  cat "$log_file" >&2
+  fail "$service$path did not become healthy"
+}
+
+snapshot_secrets() {
+  local output=$1
+  : >"$output"
+  for secret in "$RELEASE-secret" "$RELEASE-postgresql" "$RELEASE-redis"; do
+    printf '%s ' "$secret" >>"$output"
+    kubectl get secret "$secret" --namespace "$NAMESPACE" -o json \
+      | jq -cS '.data' \
+      | sha256sum \
+      | awk '{print $1}' >>"$output"
+  done
+}
+
+snapshot_pvcs() {
+  local output=$1
+  kubectl get pvc --namespace "$NAMESPACE" -o json \
+    | jq -r '.items[] | [.metadata.name, .metadata.uid, .spec.volumeName] | @tsv' \
+    | sort >"$output"
+  [[ -s "$output" ]] || fail "Helm install did not create any PVCs"
+}
+
+assert_ready_and_healthy() {
+  kubectl wait pod \
+    --namespace "$NAMESPACE" \
+    --all \
+    --for=condition=Ready \
+    --timeout="$TIMEOUT"
+
+  probe_service "$RELEASE-server" 8080 18081 /actuator/health
+  probe_service "$RELEASE-web" 80 18080 /nginx-health
+  probe_service "$RELEASE-web" 80 18080 /api/v1/namespaces
+  probe_service "$RELEASE-scanner" 8000 18082 /health
+
+  local restarts
+  restarts=$(kubectl get pods --namespace "$NAMESPACE" -o json \
+    | jq '[.items[].status.containerStatuses[]?.restartCount] | add // 0')
+  [[ "$restarts" == "0" ]] || fail "workloads restarted $restarts time(s)"
+}
+
+helm dependency build "$CHART_DIR"
+
+helm install "$RELEASE" "$CHART_DIR" \
+  --namespace "$NAMESPACE" \
+  --create-namespace \
+  --values "$TEST_VALUES" \
+  --set-string fullnameOverride="$RELEASE" \
+  --set-string publicBaseUrl=http://skillhub-smoke.local \
+  --wait \
+  --timeout "$TIMEOUT"
+
+assert_ready_and_healthy
+snapshot_secrets "$TMP_DIR/secrets-before"
+snapshot_pvcs "$TMP_DIR/pvcs-before"
+revision_before=$(helm history "$RELEASE" --namespace "$NAMESPACE" -o json \
+  | jq -r '.[-1].revision')
+
+helm upgrade "$RELEASE" "$CHART_DIR" \
+  --namespace "$NAMESPACE" \
+  --reuse-values \
+  --set-string publicBaseUrl=https://skillhub-smoke.local \
+  --set-string server.podAnnotations.helm-smoke-revision=revision-2 \
+  --wait \
+  --timeout "$TIMEOUT"
+
+assert_ready_and_healthy
+snapshot_secrets "$TMP_DIR/secrets-after"
+snapshot_pvcs "$TMP_DIR/pvcs-after"
+revision_after=$(helm history "$RELEASE" --namespace "$NAMESPACE" -o json \
+  | jq -r '.[-1].revision')
+
+(( revision_after == revision_before + 1 )) \
+  || fail "Helm revision did not advance exactly once"
+cmp "$TMP_DIR/secrets-before" "$TMP_DIR/secrets-after" \
+  || fail "application or dependency Secret data changed during upgrade"
+cmp "$TMP_DIR/pvcs-before" "$TMP_DIR/pvcs-after" \
+  || fail "PVC identity or bound volume changed during upgrade"
+
+public_base_url=$(kubectl get configmap "$RELEASE-config" \
+  --namespace "$NAMESPACE" \
+  -o json | jq -r '.data["public-base-url"]')
+[[ "$public_base_url" == "https://skillhub-smoke.local" ]] \
+  || fail "publicBaseUrl was not applied by the upgrade"
+
+echo "Helm install/upgrade smoke passed"

charts/skillhub/values.schema.json

@@ -114,8 +114,8 @@
       "properties": {
         "enabled": { "type": "boolean" },
         "bucket": { "type": "string", "minLength": 1 },
-        "endpoint": { "type": "string", "pattern": "^(|https?://.+)$" },
-        "publicEndpoint": { "type": "string", "pattern": "^(|https?://.+)$" },
+        "endpoint": { "type": "string", "pattern": "^(|https?://(\\[[0-9A-Fa-f:.]+\\]|[A-Za-z0-9]([A-Za-z0-9.-]*[A-Za-z0-9])?)(:[0-9]{1,5})?([/?#]\\S*)?)$" },
+        "publicEndpoint": { "type": "string", "pattern": "^(|https?://(\\[[0-9A-Fa-f:.]+\\]|[A-Za-z0-9]([A-Za-z0-9.-]*[A-Za-z0-9])?)(:[0-9]{1,5})?([/?#]\\S*)?)$" },
         "region": { "type": "string", "minLength": 1 },
         "forcePathStyle": { "type": "boolean" },
         "disableChunkedEncoding": { "type": "boolean" },

charts/skillhub/values.yaml

@@ -105,6 +105,11 @@ postgresql:
 
   architecture: standalone
 
+  # Bitnami's current chart defaults to a mutable latest tag. Pin the image
+  # digest so installs and rollbacks remain reproducible.
+  image:
+    digest: "sha256:52e4eae10959a7ef4810f2d0c9ce656e9337090fd2427d8d15d2e1335ba4f8fb"
+
   auth:
     postgresPassword: ""
     database: skillhub
@@ -165,6 +170,8 @@ postgresql:
 
   metrics:
     enabled: true
+    image:
+      digest: "sha256:fbd5fbc3f350c793efef777b0710f17019e8dd7812fc7c7396b13aa5130f3bd9"
     serviceMonitor:
       enabled: false
 
@@ -184,6 +191,10 @@ redis:
 
   architecture: standalone
 
+  # Keep the bundled Redis runtime immutable for repeatable upgrades.
+  image:
+    digest: "sha256:5ff436b8a0dbe3808f03c7b682865cd30ddb1f08c1a8d17970ebf475f5689722"
+
   auth:
     enabled: true
     password: ""
@@ -228,6 +239,8 @@ redis:
   sentinel:
     enabled: false
     masterSet: mymaster
+    image:
+      digest: "sha256:667bfef3e22ce9910bdcf304b1b7c659cfc15390a2d473187eb3caac2537d0e6"
     service:
       ports:
         sentinel: 26379
@@ -236,6 +249,8 @@ redis:
 
   metrics:
     enabled: true
+    image:
+      digest: "sha256:0468a9a14ebd96953f967f9f3e8673c3eb1ad1dd54d379fef15c3f3ee714c99e"
     serviceMonitor:
       enabled: false