Support docker file

[darionyaphet]

No description provided.

[yixinglu]

I suggest we should split 3 nebula services to 3 docker images and follow the recommendation: one service per container.

You maybe think there are many same common dependencies for 3 nebula services. We can create a base image described in this document. And 3 nebula services are FROM the base image, so there's only one base image layer stored in user‘s disk.

For easy usage, we can use docker-compose to deploy our nebula services with a docker-compose.yaml file.

Best practices for writing Dockerfiles is an excellent reference document.

[yixinglu]

We should extract a NEBULA_VERSION env variable to avoid updating the package name each release.

[darionyaphet]

Good point

[jude-zhu]

each 'RUN' command will create a new layer

[yixinglu]

refere to #416

[darionyaphet]

refere to #416

Thanks for you comment. It's really a good suggestion. Actually, we use docker as a demo to show how to use Nebula.

This PR should refer to Build docker image from Dockerfile .

Not Using Kubernetes to manage docker containers .

[yixinglu]

@darionyaphet
I think 3 nebula services could use a same docker image at this time. later we can build them separately to single image. The Dockerfile is like the following contents:

FROM centos:7.6.1810

MAINTAINER darion.wang "darion.wang@vesoft.com"

ARG NEBULA_VERSION

RUN yum -y install wget openssh-clients openssh-server net-tools

RUN wget -O nebula.el7-5.x86_64.rpm \
  https://github.com/vesoft-inc/nebula/releases/download/v${NEBULA_VERSION}/nebula-${NEBULA_VERSION}.el7-5.x86_64.rpm && \
  rpm -ivh nebula.el7-5.x86_64.rpm && \
  rm nebula.el7-5.x86_64.rpm && \
  mkdir -p /usr/local/nebula/logs

WORKDIR /usr/local/nebula

we build nebula-graph image each release by

$ docker build -t vesoft/nebula-graph:latest -f Dockerfile --build-arg NEBULA_VERSION=1.0.0-beta .

Given the following docker-compose.yaml, users can start nebula services with one shell command:

$ docker-compose up -d

# docker-compose.yaml
version: '3.2'
services:
  metad:
    image: vesoft/nebula-graph:latest
    expose:
      - 45500
    entrypoint:
      - ./bin/nebula-metad
      - --flagfile=./etc/nebula-metad.conf
      - --daemonize=false
      - --meta_server_addrs=172.28.1.1:45500
      - --local_ip=172.28.1.1
      - --port=45500
    networks:
      nebula-net:
        ipv4_address: 172.28.1.1

  graphd:
    image: vesoft/nebula-graph:latest
    ports:
      - 6699:3699
    entrypoint:
      - ./bin/nebula-graphd
      - --flagfile=./etc/nebula-graphd.conf
      - --daemonize=false
      - --meta_server_addrs=172.28.1.1:45500
      - --port=3699
    networks:
      nebula-net:
        ipv4_address: 172.28.1.2

  storaged:
    image: vesoft/nebula-graph:latest
    expose:
      - 44500
    entrypoint:
      - ./bin/nebula-storaged
      - --flagfile=./etc/nebula-storaged.conf
      - --daemonize=false
      - --meta_server_addrs=172.28.1.1:45500
      - --local_ip=172.28.1.3
      - --port=44500
    networks:
      nebula-net:
        ipv4_address: 172.28.1.3

networks:
  nebula-net:
    ipam:
      driver: default
      config:
        - subnet: 172.28.0.0/16

We can improve scripts/nebula.service bash script and let it accept some parameters from outside to simplify above docker-compose entrypoint option.