Deploy/testing (#1)

* fixed pip install

* Update Dockerfile

* fixed installs

* dockerfile hotfix

* cache dir for pip

* Update Dockerfile

* pip dirs

* changed tmp dir

* free space action

* fixed reg credentials

* copy files inside the image

* fixed cache dir

* fixed issues

* cache purge

* pip update

* fixed build path

* fixed torch version

* fixed fairseq version

* dependency fix

* fixed dockerfile

* public registry push

* small bug fixes

* registry name fix

* fixed ingress tls

* fixed registry name

* fixed env position

.github/workflows/production-deploy.yml

@@ -26,17 +26,19 @@ jobs:
         with:
             aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
             aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
-            aws-region: eu-central-1
+            aws-region: us-east-1
 
       - name: Login to Amazon ECR
         id: login-ecr
-        uses: aws-actions/amazon-ecr-login@v1
-
+        uses: aws-actions/amazon-ecr-login@v2
+        with:
+          registry-type: public
+
       - name: build and deploy
         env:
           ENV: production
           SWARM_HOST: 156.54.237.240
-          REGISTRY: ${{ steps.login-ecr.outputs.registry }}
+          REGISTRY: public.ecr.aws/k2u7h0h2
           AWS_REGION: eu-central-1
           SWARM_PEM: ${{secrets.AI_PRODUCTION_PEM}}
           SHAREPOINT_ID: ${{secrets.SHAREPOINT_ID}}
@@ -48,4 +50,10 @@ jobs:
           ssh-keyscan -p 2222 -H $SWARM_HOST >> ~/.ssh/known_hosts
           echo "ssh key added"
           echo "running deploy.sh $ENV"
+          sudo rm -rf /usr/share/dotnet
+          sudo rm -rf /opt/ghc
+          sudo rm -rf "/usr/local/share/boost"
+          sudo rm -rf "$AGENT_TOOLSDIRECTORY"
+          echo "Free space:"
+          df -h
           sh deploy.sh $ENV

Dockerfile

@@ -1,7 +1,7 @@
 FROM nvidia/cuda:11.4.3-cudnn8-runtime-ubuntu20.04
 ENV PYTHON_VERSION=3.9
+# this to prevent interactive intallation of 'ffmpeg'
 ENV DEBIAN_FRONTEND=noninteractive
-
 RUN apt-get -qq update \
     && apt-get -qq install --no-install-recommends \
     python${PYTHON_VERSION} \
@@ -10,8 +10,9 @@ RUN apt-get -qq update \
     python${PYTHON_VERSION}-dev build-essential \
     locales \
     ffmpeg \
-    && rm -rf /var/lib/apt/lists/* && \
-    ln -s -f /usr/bin/python${PYTHON_VERSION} /usr/bin/python3 && \
+    wget \
+    && rm -rf /var/lib/apt/lists/*
+RUN ln -s -f /usr/bin/python${PYTHON_VERSION} /usr/bin/python3 && \
     ln -s -f /usr/bin/python${PYTHON_VERSION} /usr/bin/python && \
     ln -s -f /usr/bin/pip3 /usr/bin/pip
 RUN locale-gen en_US.UTF-8 && update-locale LANG=en_US.UTF-8
@@ -32,8 +33,8 @@ ENV VIRTUAL_ENV=/opt/env/helsinki
 RUN python -m venv $VIRTUAL_ENV
 ENV PATH="$VIRTUAL_ENV/bin:$ORIG_PATH"
 RUN pip install --no-cache-dir --upgrade pip
-RUN pip install torch==2.3.0 torchvision==0.18.0 torchaudio==2.3.0 transformers sentencepiece --index-url https://download.pytorch.org/whl/cu118
-
+RUN pip install --no-cache-dir torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
+RUN pip install --no-cache-dir transformers sentencepiece
 # Create app directory
 WORKDIR /FBK
 # Copy pipeline files
@@ -45,8 +46,9 @@ RUN mkdir -p /FBK/SHAS
 COPY SHAS/. /FBK/SHAS
 ENV SHAS_ROOT=/FBK/SHAS
 # Copy http server files
-RUN mkdir -p /FBK/server/data
 COPY CMD.httpserver_start.sh httpserver.py /FBK/server/
-COPY data/. /FBK/server/data
-# run the command
-ENTRYPOINT ["entrypoint.sh"]
+COPY data /FBK/server/data
+
+COPY entrypoint.sh /FBK/
+
+ENTRYPOINT bash /FBK/entrypoint.sh 

download_models.sh

@@ -10,4 +10,4 @@ if [ ! -d "models/FBK_data_$VERSION" ]; then
 fi
 echo "FBK_data_$VERSION downloaded"
 
-mv models/FBK_data_$VERSION/* root/.cache
+mv models/FBK_data_$VERSION/* /root/.cache

entrypoint.sh

@@ -1 +1,15 @@
-bash download_models.sh && bash ./server/CMD.httpserver_start.sh
+export VERSION=v1_2
+
+cd /FBK
+# download models if the directory doesn't contains the same version
+if [ ! -d "models" ]; then
+    mkdir models
+fi
+if [ ! -d "models/FBK_data_$VERSION" ]; then
+    wget -O models/FBK_data_$VERSION.tar.gz https://fbk.sharepoint.com/:u:/s/MTUnit/$SHAREPOINT_ID?download=1 && tar xvfz models/FBK_data_$VERSION.tar.gz && rm models/FBK_data_$VERSION.tar.gz && mv FBK_data/.cache/* /root/.cache
+fi
+echo "FBK_data_$VERSION downloaded"
+echo "Creating upload and out directories"
+mkdir /FBK/server/upload mkdir /FBK/server/out
+echo "Starting the server"
+bash /FBK/server/CMD.httpserver_start.sh

kubernetes/production-deployment.yaml

@@ -29,9 +29,20 @@ spec:
         app: europeana-service
     spec:
       runtimeClassName: nvidia
+      restartPolicy: Always
+
+      imagePullSecrets:
+        - name: europeana-regcred
       containers:
       - name: europeana-service
-        image: $IMAGE
+        env:
+        - name: SHAREPOINT_ID
+          valueFrom:
+            secretKeyRef:
+              name: europeana-service
+              key: SHAREPOINT_ID
+        imagePullPolicy: Always
+        image: "$IMAGE"
         ports:
         - containerPort: 8080
         volumeMounts:
@@ -76,6 +87,10 @@ metadata:
   namespace: ai-${ENV}
 spec:
   ingressClassName: contour
+  tls:
+  - hosts:
+    - europeana-service.translatedlabs.com
+    secretName: translatedlabs.com
   rules:
   - host: europeana-service.translatedlabs.com
     http: