initial docker-compose setup

docker/README.md

@@ -0,0 +1,23 @@
+# Docker files for scio
+
+Run analyze, tika-server and api:
+
+```bash
+sudo docker network create scio
+cd docker
+./run.sh
+```
+
+### 
+Remove all data from docker instance
+
+```bash
+docker-compose down --volumes
+```
+
+### Example queries that should have hits on sample data:
+
+```
+docker exec -it docker-app-1 /bin/bash
+http_proxy= poetry run script/populate_sample_queries.sh $METASEARCH_APIKEY
+```

docker/beanstalkd/Dockerfile

@@ -0,0 +1,6 @@
+FROM docker.io/debian:buster
+
+RUN apt update
+RUN apt -y install beanstalkd
+
+ENTRYPOINT ["/usr/bin/beanstalkd"]

docker/docker-compose.yml

@@ -0,0 +1,98 @@
+---
+version: '3.8'
+
+networks:
+  scio:
+    external: true
+    name: scio
+
+x-proxy: &proxy
+  http_proxy: http://10.0.1.11:3128
+  https_proxy: http://10.0.1.11:3128
+  no_proxy: 
+
+x-build-args: &build-args
+  args:
+    <<: *proxy
+
+x-common: &common
+  networks:
+    - scio
+  logging:
+    options:
+      max-file: 5
+      max-size: 500m
+
+x-meta-scio: &scio
+  build: 
+    dockerfile: docker/scio/Dockerfile
+    context: ..
+    <<: *build-args
+  image: scio:dev
+  depends_on:
+    - elasticsearch
+    - beanstalkd
+  environment:
+      BEANSTALK: beanstalkd
+      ELASTICSEARCH: elasticsearch
+      http_proxy: 
+      https_proxy: 
+  volumes:
+    - cache-data:/home/scio/.cache/scio
+  deploy:
+    restart_policy:
+      condition: on-failure
+
+services:
+  scio-tika-server:
+    <<: *common
+    <<: *scio
+    command: .local/bin/scio-tika-server
+
+  scio-analyze:
+    <<: *common
+    <<: *scio
+    command: .local/bin/scio-analyze
+
+  scio-api:
+    <<: *common
+    <<: *scio
+    command: .local/bin/scio-api
+    ports:
+      - 3000:3000
+        #- 127.0.0.1:3000:3000
+
+  beanstalkd:
+    <<: *common
+    build: 
+      context: beanstalkd/
+      <<: *build-args
+    image: beanstalkd:dev
+    command: -l 0.0.0.0 -z 262140 -V
+    deploy:
+      restart_policy:
+        condition: on-failure
+
+  elasticsearch:
+    <<: *common
+    image: docker.elastic.co/elasticsearch/elasticsearch:7.17.3
+    environment:
+      - node.name=elasticsearch
+      - cluster.name=scio
+      - discovery.type=single-node
+      - bootstrap.memory_lock=true
+      - xpack.security.enabled=false
+      - ES_SETTING_INGEST_GEOIP_DOWNLOADER_ENABLED=false
+      - "ES_JAVA_OPTS=-Xms2048m -Xmx2048m"
+    ulimits:
+      memlock:
+        soft: -1
+        hard: -1
+    volumes:
+      - elasticsearch-data:/usr/share/elasticsearch/data
+
+volumes:
+  elasticsearch-data:
+    driver: local
+  cache-data:
+    driver: local

docker/resources/pip.conf

@@ -0,0 +1,7 @@
+[global]
+index-url = http://artifactory.mnemonic.no/artifactory/api/pypi/virt-pypi/simple/
+trusted-host = pypi.python.org artifactory.mnemonic.no
+# proxy = http://intern-proxy.mnemonic.no:3128
+timeout = 120
+cert = /etc/ssl/certs
+no-cache-dir = false

docker/run.sh

@@ -0,0 +1,6 @@
+#!/usr/bin/env bash
+
+BASE=$(dirname $0)
+
+cd $BASE
+docker-compose up -d --build --remove-orphans

docker/scio/Dockerfile

@@ -0,0 +1,27 @@
+FROM docker.io/debian:buster
+
+RUN apt update
+RUN apt -y install curl inetutils-ping telnet python3 python3-setuptools python3-pip
+
+RUN useradd -ms /bin/bash scio
+USER scio
+
+# All subsequent commands will be relative to /home/scio
+WORKDIR /home/scio
+
+RUN mkdir -p .config/pip scio/dist
+
+COPY /docker/resources/pip.conf .config/pip
+COPY /act scio/act
+COPY /etc scio/etc
+COPY /setup.py scio/setup.py
+COPY /README.md scio/README.md
+
+RUN mkdir -p /home/scio/.cache/scio
+RUN chown scio:scio /home/scio/.cache/scio
+
+RUN cd scio && \
+    rm -f dist/*gz && \
+    python3 setup.py sdist && \
+    pip3 install dist/*gz
+