Docker Swarm is the official docker cluster management solution and it is natively integrated with Docker Engine.
You can find detailed instructions and consideration on running Orion Context Broker on Docker Swarm at this link.
Here we provide only a quick overview.
Alternatively, you can find a general discussion on Orion HA deployment here.
To deploy on your Linux-based system a Docker Swarm test cluster, you can use miniswarm.
The only pre-requisite are:
# As root
$ curl -sSL https://raw.githubusercontent.com/aelsabbahy/miniswarm/master/miniswarm -o /usr/local/bin/miniswarm
$ chmod +rx /usr/local/bin/miniswarm
# 1 manager 2 workers
$ miniswarm start 3
The above command will create a Docker Swarm cluster of 3 nodes, including 1 master and 2 workers nodes.
$ eval $(docker-machine env ms-manager0)
Now your Docker client will be connected to the manager node of the cluster you just created.
You can check that by running
$ docker machine ls
NAME ACTIVE DRIVER STATE URL SWARM DOCKER ERRORS
ms-manager0 * virtualbox Running tcp://192.168.99.101:2376 v18.02.0-ce
ms-worker0 - virtualbox Running tcp://192.168.99.102:2376 v18.02.0-ce
ms-worker1 - virtualbox Running tcp://192.168.99.100:2376 v18.02.0-ce
The node with *
is the node to which your docker client will connect.
To deploy Orion in High Availability, you need first to deploy a MongoDB ReplicaSet, and then deploy on top of it your Orion Context Broker.
Details on how to deploy a MongoDB ReplicaSet in Docker Swarm are available here.
-
Create an overlay network
$ docker network create --opt encrypted -d overlay backend ncb90nkwpiofoof757te09xmt
-
Create a docker-compose-mongo.yml file (or reuse the scripts in the repository that offers additional functionalities such as healthchecks):
version: '3.3' services: mongo: image: mongo:6.0 entrypoint: [ "/usr/bin/mongod", "--replSet", "rs", "--journal", "--smallfiles", "--bind_ip", "0.0.0.0"] volumes: - mongodata:/data/db networks: - backend deploy: mode: global restart_policy: condition: on-failure update_config: # should you update the mongo cluster this configuration ensure # that nodes are updated one by one ensuring that the mongo service # remains available to other services. the delay ensure that # when the new mongo is deployed it has enough time to be connected # to the cluster parallelism: 1 delay: 1m30s # this service contains the logic to manage the mongo db # replicaset consistency controller: image: martel/mongo-replica-ctrl:latest volumes: - /var/run/docker.sock:/var/run/docker.sock environment: - OVERLAY_NETWORK_NAME=backend - MONGO_SERVICE_NAME=mongo_mongo - REPLICASET_NAME=rs - MONGO_PORT=27017 # - DEBUG=1 #uncomment to debug the script entrypoint: python /src/replica_ctrl.py networks: - backend depends_on: - "mongo" deploy: mode: replicated replicas: 1 placement: constraints: [node.role==manager] restart_policy: condition: on-failure volumes: # External true ensures that the volume is not re-created if already present mongodata: external: true networks: backend: external: name: backend
-
Deploy the MongoDB ReplicaSet
$ docker stack deploy -c docker-compose-mongo.yml mongo Creating service mongo_mongo Creating service mongo_controller
-
Check that the deployment worked out correctly (you need to wait the services deployment to be completed):
$ docker service logs -f mongo_controller mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:Waiting mongo service (and tasks) (mongo_mongo) to start mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:Mongo service is up and running mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:No previous valid configuration, starting replicaset from scratch mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:replSetInitiate: {'ok': 1.0} mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:Primary is: 10.0.0.8 mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:Primary is: 10.0.0.8 mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:Primary is: 10.0.0.8 mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:Primary is: 10.0.0.8 mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:Primary is: 10.0.0.8 mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:Primary is: 10.0.0.8 mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:Primary is: 10.0.0.8 mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:Primary is: 10.0.0.8 mongo_controller.1.jx664h87o3ft@ms-manager0 | INFO:__main__:Primary is: 10.0.0.8
-
Create a docker-compose-orion.yml file (or reuse the scripts in the repository that offers additional functionalities such as healthchecks and use separate networks for frontend and backend):
version: '3' services: orion: image: fiware/orion:latest ports: - "1026:1026" command: -logLevel DEBUG -dbURI mongodb://mongo_mongo/?replicaSet=rs&connectTimeoutMS=10000 deploy: replicas: 2 networks: - backend networks: backend: driver: overlay external: true
-
Deploy the Orion Context Broker cluster
$ docker stack deploy -c docker-compose-orion.yml orion Creating service orion_orion
-
Check that the deployment worked out correctly (you need to wait the services deployment to be completed):
$ curl $(docker-machine ip ms-manager0):1026/version -s -S { "orion" : { "version" : "2.4.0-next", "uptime" : "0 d, 0 h, 0 m, 27 s", "git_hash" : "f2a3d436b2b507c5fd1611492ad7fad386901952", "compile_time" : "Thu Oct 29 15:56:28 UTC 2020", "compiled_by" : "root", "compiled_in" : "4d72f1940cd1", "release_date" : "Thu Oct 29 15:56:28 UTC 2020", "doc" : "https://fiware-orion.rtfd.io/", "libversions": ... } }
-
Scale up orion:
$ docker service scale orion_orion=3 orion_orion scaled to 3 overall progress: 2 out of 3 tasks 1/3: running [==================================================>] 2/3: preparing [=================================> ] 3/3: running [==================================================>]