Merge pull request #43 from semi-technologies/gh-42-Add-support-for-e…

…xternal-inference-services

Add support for image module and external inference services

weaviate/Chart.yaml

@@ -7,10 +7,10 @@ type: application
 # This is the chart version. This version number should be incremented each time you make changes
 # to the chart and its templates, including the app version.
 # Versions are expected to follow Semantic Versioning (https://semver.org/)
-version: 13.1.0
+version: 13.2.0
 
 
 # This is the version number of the application being deployed. This version number should be
 # incremented each time you make changes to the application. Versions are not expected to
 # follow Semantic Versioning. They should reflect the version the application is using.
-appVersion: v1.3.0
+appVersion: v1.4.0

weaviate/templates/_helpers.tpl

@@ -1,18 +1,38 @@
 {{/* Generate the enabled modules config. This can be done a lot nicer once we drop Helm v2 support */}}
 {{ define "enabled_modules" }}
-  {{- if index .Values "modules" "text2vec-contextionary" "enabled" -}}
-    {{- if index .Values "modules" "text2vec-transformers" "enabled" -}}
+  {{- if or (index .Values "modules" "text2vec-contextionary" "enabled") (index .Values "modules" "text2vec-contextionary" "inferenceUrl") -}}
+    {{- if or (index .Values "modules" "text2vec-transformers" "enabled") (index .Values "modules" "text2vec-transformers" "inferenceUrl") -}}
       {{ fail "cannot have two text2vec-* modules on at the same time" -}}
     {{- end -}}
-    {{- if index .Values "modules" "qna-transformers" "enabled" -}}
-      text2vec-contextionary,qna-transformers
+    {{- if or (index .Values "modules" "qna-transformers" "enabled") (index .Values "modules" "qna-transformers" "inferenceUrl") -}}
+      {{- if or (index .Values "modules" "img2vec-keras" "enabled") (index .Values "modules" "img2vec-keras" "inferenceUrl") -}}
+        text2vec-contextionary,qna-transformers,img2vec-keras
+      {{- else -}}
+        text2vec-contextionary,qna-transformers
+      {{- end -}}
+    {{- else if or (index .Values "modules" "img2vec-keras" "enabled") (index .Values "modules" "img2vec-keras" "inferenceUrl") -}}
+      {{- if or (index .Values "modules" "qna-transformers" "enabled") (index .Values "modules" "qna-transformers" "inferenceUrl") -}}
+        text2vec-contextionary,qna-transformers,img2vec-keras
+      {{- else -}}
+        text2vec-contextionary,img2vec-keras
+      {{- end -}}
     {{- else -}}
       text2vec-contextionary
     {{- end -}}
   {{- else -}}
-    {{- if index .Values "modules" "text2vec-transformers" "enabled" -}}
-      {{- if index .Values "modules" "qna-transformers" "enabled" -}}
-        text2vec-transformers,qna-transformers
+    {{- if or (index .Values "modules" "text2vec-transformers" "enabled") (index .Values "modules" "text2vec-transformers" "inferenceUrl") -}}
+      {{- if or (index .Values "modules" "qna-transformers" "enabled") (index .Values "modules" "qna-transformers" "inferenceUrl") -}}
+        {{- if or (index .Values "modules" "img2vec-keras" "enabled") (index .Values "modules" "img2vec-keras" "inferenceUrl") -}}
+          text2vec-transformers,qna-transformers,img2vec-keras
+        {{- else -}}
+          text2vec-transformers,qna-transformers
+        {{- end -}}
+      {{- else if or (index .Values "modules" "img2vec-keras" "enabled") (index .Values "modules" "img2vec-keras" "inferenceUrl") -}}
+        {{- if or (index .Values "modules" "qna-transformers" "enabled") (index .Values "modules" "qna-transformers" "inferenceUrl") -}}
+          text2vec-transformers,qna-transformers,img2vec-keras
+        {{- else -}}
+          text2vec-transformers,img2vec-keras
+        {{- end -}}
       {{- else -}}
         text2vec-transformers
       {{- end -}}

weaviate/templates/img2vecKerasDeployment.yaml

@@ -0,0 +1,45 @@
+{{ if index .Values "modules" "img2vec-keras" "enabled" }}
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: {{ index .Values "modules" "img2vec-keras" "fullnameOverride" }}
+  labels:
+    name: {{ index .Values "modules" "img2vec-keras" "fullnameOverride" }}
+    app: {{ index .Values "modules" "img2vec-keras" "fullnameOverride" }}
+spec:
+  replicas: {{ index .Values "modules" "img2vec-keras" "replicas" }}
+  selector:
+    matchLabels:
+      app: {{ index .Values "modules" "img2vec-keras" "fullnameOverride" }}
+  template:
+    metadata:
+      labels:
+        app: {{ index .Values "modules" "img2vec-keras" "fullnameOverride" }}
+    spec:
+      containers:
+      - name: {{ index .Values "modules" "img2vec-keras" "fullnameOverride" }}
+        image: {{ index .Values "modules" "img2vec-keras" "registry" }}/{{ index .Values "modules" "img2vec-keras" "repo" }}:{{ index .Values "modules" "img2vec-keras" "tag" }}
+        env:
+          - name: ENABLE_CUDA
+            value: "{{ if index .Values "modules" "img2vec-keras" "envconfig" "enable_cuda" }}1{{ else }}0{{ end }}"
+          {{ if index .Values "modules" "img2vec-keras" "envconfig" "enable_cuda" }}
+          - name: NVIDIA_VISIBLE_DEVICES
+            value: {{ index .Values "modules" "img2vec-keras" "envconfig" "nvidia_visible_devices"}}
+          - name: LD_LIBRARY_PATH
+            value: {{ index .Values "modules" "img2vec-keras" "envconfig" "ld_library_path"}}
+          {{ end }}
+        resources:
+{{ index .Values "modules" "img2vec-keras" "resources" | toYaml | indent 10 }}
+        livenessProbe:
+          httpGet:
+            path: /.well-known/live
+            port: 8080
+          initialDelaySeconds: 120
+          periodSeconds: 3
+        readinessProbe:
+          httpGet:
+            path: /.well-known/ready
+            port: 8080
+          initialDelaySeconds: 3
+          periodSeconds: 3
+{{ end }}

weaviate/templates/img2vecKerasService.yaml

@@ -0,0 +1,14 @@
+{{ if index .Values "modules" "img2vec-keras" "enabled" }}
+apiVersion: v1
+kind: Service
+metadata:
+  name: {{ index .Values "modules" "img2vec-keras" "fullnameOverride" }}
+spec:
+  type: ClusterIP
+  selector:
+    app: {{ index .Values "modules" "img2vec-keras" "fullnameOverride" }}
+  ports:
+    - protocol: TCP
+      port: 8080
+      targetPort: 8080
+{{ end }}

weaviate/templates/weaviateStatefulset.yaml

@@ -33,10 +33,23 @@ spec:
           {{ if index .Values "modules" "text2vec-transformers" "enabled" }}
           - name: TRANSFORMERS_INFERENCE_API
             value: http://{{ index .Values "modules" "text2vec-transformers" "fullnameOverride" }}.{{ .Release.Namespace }}.svc.cluster.local:8080
+          {{ else if index .Values "modules" "text2vec-transformers" "inferenceUrl" }}
+          - name: TRANSFORMERS_INFERENCE_API
+            value: {{ index .Values "modules" "text2vec-transformers" "inferenceUrl" }}
           {{ end }}
           {{ if index .Values "modules" "qna-transformers" "enabled" }}
           - name: QNA_INFERENCE_API
             value: http://{{ index .Values "modules" "qna-transformers" "fullnameOverride" }}.{{ .Release.Namespace }}.svc.cluster.local:8080
+          {{ else if index .Values "modules" "qna-transformers" "inferenceUrl" }}
+          - name: QNA_INFERENCE_API
+            value: {{ index .Values "modules" "qna-transformers" "inferenceUrl" }}
+          {{ end }}
+          {{ if index .Values "modules" "img2vec-keras" "enabled" }}
+          - name: IMAGE_INFERENCE_API
+            value: http://{{ index .Values "modules" "img2vec-keras" "fullnameOverride" }}.{{ .Release.Namespace }}.svc.cluster.local:8080
+          {{ else if index .Values "modules" "img2vec-keras" "inferenceUrl" }}
+          - name: IMAGE_INFERENCE_API
+            value: {{ index .Values "modules" "img2vec-keras" "inferenceUrl" }}
           {{ end }}
         ports:
           - containerPort: 8080

weaviate/values.yaml

@@ -158,6 +158,39 @@ modules:
         # enable if running with CUDA support
         # nvidia.com/gpu: 1
 
+  # The img2vec-keras module uses neural networks, such as BERT,
+  # DistilBERT, to find an aswer in text to a given question
+  img2vec-keras:
+    enabled: false
+    tag: resnet50-61dcbf8
+    repo: semitechnologies/img2vec-keras
+    registry: docker.io
+    replicas: 1
+    fullnameOverride: img2vec-keras
+    envconfig:
+      # enable for CUDA support. Your K8s cluster needs to be configured
+      # accordingly and you need to explicilty set GPU requests & limits below
+      enable_cuda: false
+
+      # only used when cuda is enabled
+      nvidia_visible_devices: all
+
+      # only used when cuda is enabled
+      ld_library_path: /usr/local/nvidia/lib64
+
+    resources:
+      requests:
+        cpu: '1000m'
+        memory: '3000Mi'
+
+        # enable if running with CUDA support
+        # nvidia.com/gpu: 1
+      limits:
+        cpu: '1000m'
+        memory: '5000Mi'
+
+        # enable if running with CUDA support
+        # nvidia.com/gpu: 1
 
   # by choosing the default vectorizer module, you can tell Weaviate to always
   # use this module as the vectorizer if nothing else is specified. Can be