Nebula-orchestartor，廣域網容器資源協調器

Nebula-orchestartor，廣域網容器資源協調器

Nebula-orchestartor能夠跨數據中心進行容器資源的管理。傳統的資源協調器例如YARN/Mesos/Kubernetes都是在數據中心內部使用，假設集羣內經過高速網絡鏈接，主要用於數據密集或計算密集的場合。對於IoT、邊緣計算、霧計算、多數據中心互聯等廣域網應用來講，面對傳輸延遲、低帶寬、數據同步、容錯恢復、服務路由、災難備份等各類狀況，須要開發新的解決方案。Nebula-orchestartor採用MongoDB進行數據存儲，rabbitMQ做爲數據通信，具備擴展到廣域網資源調度的潛力，不過目前還在早期發展階段。這裏經過一個簡單的實例，能夠體驗Nebula-orchestartor的操做模式和能力。

參見：

• 官網：https://nebula-orchestrator.github.io/
• 源碼：https://github.com/nebula-orchestrator
• 文檔：http://nebula.readthedocs.io/en/latest/getting-started/

快速入門教程

瞭解新技術的最好辦法是親自動手去作一作。下面的教程使用docker-compose快速設置一個 Nebula 集羣。要求docker & docker-compose預先安裝，推薦使用Docker for Mac的最新版本。

1）安裝基礎Docker鏡像

1. Note that some linux systems have epmd installed (Ubuntu included) which the following rabbitmq container requires so if your having problems with rabbit refusing to run try killing the host epmd and\or host rabbitmq-server for the duration of this tutorial
2. First get the docker-compose.yml of this tutorial and save it locally on your machine
3. Nebula is geared towards using a private registry but will also work with docker hub, to use your own user\pass (by default no auth is used) edit the docker-compose.yml by entering your docker hub user & password into "REGISTRY_AUTH_USER" & "REGISTRY_AUTH_PASSWORD" under the "worker" container, if your using a private registry change the "REGISTRY_HOST" variable as well to point to your private registry. from the directory where you saved docker-compose.yml at (same name is important) run docker-compose up -d (you might need to sudo su first if you didn't set your user to be part of the docker group), don't worry if you see the worker-manager & api-manager restarting, it's because the mongo & rabbit containers aren't configured yet so they fail to connect to it

2）配置MongoDB

1. 建立database的user & schema，命令以下：

   docker exec -it mongo mongo
   use nebula
   db.createUser(
      {
        user: "nebula",
        pwd: "nebula",
        roles: [ "readWrite" ]
      }
   )

2. 退出容器 (ctrl-d)。

3）配置rabbitMQ

1. 建立 rabbitMQ的 user 和 vhost，命令以下：

   docker exec -it rabbit sh
   rabbitmqctl add_vhost nebula
   rabbitmqctl add_user nebula nebula
   rabbitmqctl set_permissions -p nebula nebula ".*" ".*" ".*"

2. 退出容器 (ctrl-d)。

3. Either wait for the changes to catch (they are examined at the container boot only so if the DB's schemas are not yet created the container restarts) or restart the api-manager & worker-manager containers

4）建立Worker-Nginx

如今Nebula cluster已經運行起來了, lets use Curl to create an nginx app to fill the "example" APP_NAME app that the worker-manager has set to manage:

curl -X POST \
  http://127.0.0.1/api/apps/example \
  -H 'authorization: Basic bmVidWxhOm5lYnVsYQ==' \
  -H 'cache-control: no-cache' \
  -H 'content-type: application/json' \
  -d '{
  "starting_ports": [{"81":"80"}],
  "containers_per": {"server": 1},
  "env_vars": {},
  "docker_image" : "nginx",
  "running": true,
  "volumes": ["/tmp:/tmp/1", "/var/tmp/:/var/tmp/1:ro"],
  "networks": ["nebula"],
  "privileged": false,
  "devices": []
}'

由於須要下載Nginx容器，若是之前機器上沒有鏡像，須要耐心等待.......

可使用下面的命令查看進度(一直等待到結束通知)：

docker logs -f work-manager

5）訪問Worker服務

Either wait for the changes to catch (usually few seconds at most) or restart the worker-manager container, you now have your first nebula worker (try logging into 127.0.0.1:81 in your browser to see), because the network is internal in this tutorial you can only run more on the same machine (which kinda defeats the purpose) but after you deploy Nebula by following the install guide you can run as many workers as you need by having multiple servers running the same worker-manager container with the same envvars\config file.

視頻教程： asciinema