Refactoring web apps and add CI testing for chart and images

.github/workflows/build-push-artifacts.yml

@@ -1,7 +1,6 @@
 # Adapted from https://github.com/stackhpc/azimuth/blob/master/.github/workflows/build-push-artifacts.yaml
 name: Publish artifacts
 # Run the tasks on every push
-# TODO: Add path filtering to only run on relevant changes
 on: push
 jobs:
   # Job to run change detection
@@ -24,9 +23,10 @@ jobs:
         id: filter
         with:
           base: ${{ github.ref_name }}
+          # TODO: Make image filters more granular
           filters: |
             images:
-              - 'images/**'
+              - 'web-apps/**'
             chart:
               - 'chart/**'
 
@@ -39,8 +39,7 @@ jobs:
     strategy:
       matrix:
         include:
-          # - component: api
-          - component: ui
+          - component: chat-interface
     steps:
       - name: Check out the repository
         uses: actions/checkout@v4
@@ -56,23 +55,22 @@ jobs:
         id: image-meta
         uses: docker/metadata-action@v5
         with:
-          images: ghcr.io/stackhpc/azimuth-llm-${{ matrix.component }}-base
+          images: ghcr.io/stackhpc/azimuth-llm-${{ matrix.component }}
           # Produce the branch name or tag and the SHA as tags
           tags: |
             type=ref,event=branch
             type=ref,event=tag
             type=sha,prefix=
 
       - name: Build and push image
-        uses: stackhpc/github-actions/docker-multiarch-build-push@allow-continue-after-scan
+        uses: azimuth-cloud/github-actions/docker-multiarch-build-push@update-trivy-action
         with:
-          cache-key: ${{ matrix.component }}-base
-          context: ./images/${{ matrix.component }}-base
+          cache-key: ${{ matrix.component }}
+          context: ./web-apps/${{ matrix.component }}
           platforms: linux/amd64,linux/arm64
           push: true
           tags: ${{ steps.image-meta.outputs.tags }}
           labels: ${{ steps.image-meta.outputs.labels }}
-          fail_on_high_severity_cve: false
 
   # Job to build and publish Helm chart
   build_push_chart:
@@ -91,10 +89,10 @@ jobs:
 
       - name: Get SemVer version for current commit
         id: semver
-        uses: stackhpc/github-actions/semver@master
+        uses: azimuth-cloud/github-actions/semver@master
 
       - name: Publish Helm charts
-        uses: stackhpc/github-actions/helm-publish@master
+        uses: azimuth-cloud/github-actions/helm-publish@master
         with:
           token: ${{ secrets.GITHUB_TOKEN }}
           version: ${{ steps.semver.outputs.version }}

.github/workflows/test-pr.yml

@@ -0,0 +1,49 @@
+name: Test pull request
+on:
+  pull_request:
+    types:
+      - opened
+      - reopened
+      - ready_for_review
+      - synchronize
+concurrency:
+  group: ${{ github.workflow }}-${{ github.head_ref }}
+  cancel-in-progress: true
+jobs:
+  chart_validation:
+    runs-on: ubuntu-latest
+    env:
+      CLUSTER_NAME: chart-testing
+      RELEASE_NAME: ci-test
+    steps:
+      - name: Checkout the repo
+        uses: actions/checkout@v4
+
+      - name: Install Helm
+        uses: azure/setup-helm@v4
+
+      - name: Set up chart testing
+        uses: helm/chart-testing-action@v2
+
+      - name: Run chart linting
+        run: ct lint --config ct.yaml
+
+      - name: Run helm template with default values
+        run: helm template ci-test .
+        working-directory: chart
+
+      - name: Create Kind Cluster
+        uses: helm/kind-action@v1
+        with:
+          cluster_name: ${{ env.CLUSTER_NAME }}
+
+      - name: Add Helm repos for dependencies
+        run: |
+          helm repo add stakater https://stakater.github.io/stakater-charts
+
+      # https://github.com/helm/charts/blob/master/test/README.md#providing-custom-test-values
+      # Each chart/ci/*-values.yaml file will be treated as a separate test case with it's
+      # own helm install/test process.
+      - name: Run chart install and test
+        run: ct install --config ct.yaml
+

.gitignore

@@ -8,11 +8,8 @@ __pycache__/
 
 # Ignore local dev helpers
 test-values.y[a]ml
-chart/web-app/settings.yml
-gradio-client-test.py
 **venv*/
 
-
 # Helm chart stuff
 chart/Chart.lock
-chart/charts
+chart/charts

chart/.helmignore

@@ -25,12 +25,9 @@
 # Others
 README.md
 Dockerfile
-kubeconfig.yml
+*kubeconfig.y[a]ml
 venv/
 __pycache__/
 images/
-.hf-token.secret
-hu-poc/
-test-values.yaml
-web-app/settings.yml
-web-app/example-settings.yml
+*.secret
+ci/

chart/Chart.yaml

@@ -1,6 +1,9 @@
 apiVersion: v2
 name: azimuth-llm
 description: HuggingFace LLM model serving along with a simple web interface.
+maintainers:
+  - name: "Scott Davidson"
+    email: [email protected]
 
 # A chart can be either an 'application' or a 'library' chart.
 #
@@ -32,4 +35,4 @@ dependencies:
   - name: reloader
     version: 1.0.63
     repository: https://stakater.github.io/stakater-charts
-    condition: ui.enabled
+    condition: ui.enabled

chart/ci/web-apps-only-values.yaml

@@ -0,0 +1,6 @@
+api:
+  enabled: false
+ui:
+  service:
+    zenith:
+      enabled: false

chart/templates/api/config-map-grafana-dashboard-details.yml

@@ -1,4 +1,4 @@
-{{- if .Values.api.monitoring.enabled -}}
+{{- if and .Values.api.enabled .Values.api.monitoring.enabled -}}
 apiVersion: v1
 kind: ConfigMap
 metadata:

chart/templates/api/config-map-grafana-dashboard-summary.yml

@@ -1,4 +1,4 @@
-{{- if .Values.api.monitoring.enabled -}}
+{{- if and .Values.api.enabled .Values.api.monitoring.enabled -}}
 apiVersion: v1
 kind: ConfigMap
 metadata:

chart/templates/api/deployment.yml

@@ -1,3 +1,4 @@
+{{- if .Values.api.enabled -}}
 apiVersion: apps/v1
 kind: Deployment
 metadata:
@@ -64,3 +65,4 @@ spec:
           emptyDir:
             medium: Memory
             sizeLimit: 1Gi
+{{- end -}}

chart/templates/api/ingress.yml

@@ -1,4 +1,4 @@
-{{- if .Values.ingress.api.enabled -}}
+{{- if and .Values.api.enabled .Values.ingress.api.enabled -}}
 apiVersion: networking.k8s.io/v1
 kind: Ingress
 metadata:

chart/templates/api/service-monitor.yml

@@ -1,4 +1,4 @@
-{{- if .Values.api.monitoring.enabled -}}
+{{- if and .Values.api.enabled .Values.api.monitoring.enabled -}}
 apiVersion: monitoring.coreos.com/v1
 kind: ServiceMonitor
 metadata:

chart/templates/api/service.yml

@@ -1,3 +1,4 @@
+{{- if .Values.api.enabled -}}
 apiVersion: v1
 kind: Service
 metadata:
@@ -13,3 +14,4 @@ spec:
   type: {{ .Values.api.service.type }}
   selector:
     {{- include "azimuth-llm.api-selectorLabels" . | nindent 4 }}
+{{- end -}}

chart/templates/api/zenith-client.yml

@@ -1,4 +1,4 @@
-{{- if .Values.api.service.zenith.enabled -}}
+{{- if and .Values.api.enabled .Values.api.service.zenith.enabled -}}
 apiVersion: zenith.stackhpc.com/v1alpha1
 kind: Client
 metadata:
@@ -11,4 +11,4 @@ spec:
     serviceName: {{ .Values.api.service.name }}
   auth:
     skip: {{ .Values.api.service.zenith.skipAuth }}
-{{- end -}}
+{{- end -}}

chart/templates/api/zenith-reservation.yml

@@ -1,4 +1,4 @@
-{{- if .Values.api.service.zenith.enabled -}}
+{{- if and .Values.api.enabled .Values.api.service.zenith.enabled -}}
 apiVersion: zenith.stackhpc.com/v1alpha1
 kind: Reservation
 metadata:
@@ -13,4 +13,4 @@ metadata:
     {{- end }}
 spec:
   credentialSecretName: {{ .Release.Name }}-api-zenith-credential
-{{- end -}}
+{{- end -}}

chart/templates/test/end-to-end.yml

@@ -0,0 +1,27 @@
+{{- if and .Values.ui.enabled .Values.api.enabled -}}
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: gradio-client-response
+  annotations:
+    "helm.sh/hook": test
+spec:
+  template:
+    spec:
+      containers:
+      - name: gradio-client-test
+        image: ghcr.io/stackhpc/azimuth-llm-chat-interface:{{ .Values.ui.image.version }}
+        imagePullPolicy: IfNotPresent
+        command:
+        - python
+        - gradio-client-test.py
+        {{- if .Values.ingress.ui.enabled }}
+        - {{ .Values.ingress.host }}{{ .Values.ingress.ui.path }}
+        {{- else }}
+        - http://{{ .Values.ui.service.name }}.{{ .Release.Namespace }}.svc
+        {{- end }}
+      restartPolicy: Never
+  # Allow plenty of retries since downloading
+  # model weights can take a long time.
+  backoffLimit: 10
+{{- end -}}

chart/templates/test/web-app.yml

@@ -0,0 +1,26 @@
+{{- if and .Values.ui.enabled (not .Values.api.enabled) -}}
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: web-app-liveness
+  annotations:
+    "helm.sh/hook": test
+spec:
+  template:
+    spec:
+      containers:
+      - name: web-app-test
+        image: nicolaka/netshoot:v0.12
+        imagePullPolicy: IfNotPresent
+        command:
+        - nc
+        - -vz
+        {{- if .Values.ingress.ui.enabled }}
+        - {{ .Values.ingress.host | trimPrefix "http://" | trimPrefix "https://" }}{{ .Values.ingress.ui.path }}
+        {{- else }}
+        - {{ .Values.ui.service.name }}.{{ .Release.Namespace }}.svc
+        {{- end }}
+        - "80"
+      restartPolicy: Never
+  backoffLimit: 3
+{{- end -}}

chart/templates/ui/app-config-map.yml

@@ -6,7 +6,6 @@ metadata:
   labels:
     {{- include "azimuth-llm.labels" . | nindent 4 }}
 data:
-{{ (.Files.Glob "web-app/*").AsConfig | nindent 2 }}
-  settings.yml: |
+  overrides.yml: |
     {{- .Values.ui.appSettings | toYaml | nindent 4 }}
-{{- end -}}
+{{- end -}}

chart/templates/ui/deployment.yml

@@ -22,26 +22,25 @@ spec:
     spec:
       containers:
       - name: {{ .Release.Name }}-ui
-        image: {{ printf "%s:%s" .Values.ui.image.repository .Values.ui.image.version }}
+        {{- with .Values.ui.image }}
+        image: {{ printf "%s:%s" .repository .version }}
+        {{- if .imagePullPolicy }}
+        imagePullPolicy: {{ .imagePullPolicy }}
+        {{- end -}}
+        {{- end }}
         ports:
         - name: ui
-          containerPort: 7680
-        workingDir: /etc/web-app
+          containerPort: 7860
         volumeMounts:
         - name: app
           mountPath: /etc/web-app
-        command:
-        - python
-        args:
-        - {{ .Values.ui.entrypoint }}
-        - {{ .Values.huggingface.model }}
         env:
         - name: PYTHONUNBUFFERED
           value: "1"
         tty: true # Make stdout from python visible in k8s logs
         readinessProbe:
           tcpSocket:
-            port: 7680
+            port: 7860
           periodSeconds: 5
       volumes:
         - name: app

chart/values.yaml

@@ -30,6 +30,7 @@ huggingface:
   token:
 # Configuration for the backend model serving API
 api:
+  enabled: true
   # Container image config
   image:
     repository: vllm/vllm-openai
@@ -99,8 +100,9 @@ ui:
         - Arial
   # Container image config
   image:
-    repository: ghcr.io/stackhpc/azimuth-llm-ui-base
-    version: "0.3.0"
+    repository: ghcr.io/stackhpc/azimuth-llm-chat-interface
+    version: 87a0342
+    imagePullPolicy:
   # Service config
   service:
     name: web-app