In this tutorial we will going to setup Odysseus Studio with OdysseusNet capabilities and start it as a master node. Then worker nodes are started with docker-compose and will be dynamically added to the master node with OdysseusNet in Odysseus Studio. At next queries will be remotely executed on the worker nodes from the master node.
Setup Odysseus with Development with Odysseus, but change the following
Now you have installed the target platform, before you launch the product, we are going to add the OdysseusNet monolithic target to the normal monolithic target (see for a general tutorial Adding features to products) :
The OdysseusNet monolithic target also includes the OdysseusNet server target.
If successfully started you can close Odysseus Studio for now for changing OdysseusNet config under ODYSSEUS_HOME (OdysseusNet Configuration)
With Linux the ODYSSEUS_HOME (ODYSSEUS_HOME) is ~/.odysseus, there you can change the odysseusNet.config to
<?xml version="1.0" encoding="UTF-8"?> <!DOCTYPE properties SYSTEM "http://java.sun.com/dtd/properties.dtd"> <properties> <comment>OdysseusNet Property File edit only if you know what you are doing</comment> <entry key="net.networkoperatorgenerator">websocket</entry> <entry key="net.node.default.username">System</entry> <entry key="net.dd.local">true</entry> <entry key="net.node.name">OdysseusNode_1554</entry> <entry key="net.discoverer.name">BroadcastOdysseusNodeDiscoverer,IPListOdysseusNodeDiscoverer</entry> <entry key="net.querydistribute.allocation">querycount</entry> <entry key="net.source.lifetime">3600000</entry> <entry key="net.connect.selector.name">GroupSelectorComponent</entry> <entry key="net.nodemanager.storetype">filestore</entry> <entry key="net.nodemanager.filename">/home/$USER/.odysseus/store/nodes.store</entry> <entry key="net.dd.checkinterval">30000</entry> <entry key="net.remoteUpdate">false</entry> <entry key="net.node.communicator">rest</entry> <entry key="net.node.preserveid">false</entry> <entry key="net.querydistribute.minport">10000</entry> <entry key="net.autostart">true</entry> <entry key="net.querydistribute.partition">querycloud</entry> <entry key="net.querydistribute.randomport">false</entry> <entry key="net.discoverer.interval">5000</entry> <entry key="net.node.default.password">manager</entry> <entry key="net.node.group">OdysseusGroup</entry> <entry key="net.querydistribute.maxport">20000</entry> <entry key="net.logging.receive">false</entry> </properties> |
Replace $USER with your local user
Now we start Nexmark (a data stream simulation tool, see Getting Started with Nexmark) and Odysseus worker nodes with docker-compose, therefore create a docker-compose.yml in a directory (it will be named $CURRENT_DIRECTORY here for reference, in practice it could be e. g. the odysseusnet-repo) and with the following content (adopted from OdysseusNet Docker Compose Example):
version: '3.3' services: nexmark: image: odysseusol/nexmark ports: - 65440-65443:65440-65443 worker01: image: odysseusol/odysseusnet stdin_open: true ports: - 18881:8888 volumes: - ./worker01:/var/lib/odysseus worker02: image: odysseusol/odysseusnet stdin_open: true ports: - 18882:8888 volumes: - ./worker02:/var/lib/odysseus worker03: image: odysseusol/odysseusnet stdin_open: true ports: - 18883:8888 volumes: - ./worker03:/var/lib/odysseus |
As a workaround make sure the worker01-03 directories have the 777-permissions with:
sudo chmod 777 -R worker0* |
Otherwise the internal Odysseus worker node user cannot access the volume properly and fails.
All configuration files will be placed under the $CURRENT_DIRECTORY, e. g. there are the following files: logs odysseus.conf odysseusNet.conf reloadlog.store scheduling.conf store |
You can then start the docker stack with:
docker-compose up -d |
You can use the Odysseus Webstudio (WebStudio) for monitoring and interacting with the worker nodes. To install it, first clone the Webstudio-repo (https://git.swl.informatik.uni-oldenburg.de/projects/API_APPS/repos/webstudio/browse?at=refs%2Fheads%2Fdevelopment) and checkout the development-branch with
git clone https://git.swl.informatik.uni-oldenburg.de/scm/api_apps/webstudio.git git checkout -b development |
Then create and give the all permissions to everyone to the following directories as a workaround:
mkdir frontend odysseushome chmod 777 -R frontend odysseushome |
For starting the docker stack you can set the images accordingly in the docker-compose.yml file:
version: '3.3' services: frontend: image: odysseusol/webstudio_frontend #build: frontend ports: - "4200:80" environment: NODE_ENV: ${NODE_ENV} restart: always depends_on: - backend backend: image: odysseusol/webstudio_backend #build: backend ports: - "3000:3000" environment: NODE_ENV: ${NODE_ENV} MONGODB_URI: ${MONGODB_URI} SECRET_TOKEN: ${SECRET_TOKEN} DOWNLOAD_TOKEN: ${DOWNLOAD_TOKEN} PROJECTS_FOLDER: ${PROJECTS_FOLDER} restart: always depends_on: - database volumes: - backenddata:/home/node/app/projects - ./backend/logs:/home/node/app/logs database: image: mongo restart: always ports: - "27017:27017" volumes: - mongodbdata:/data/db odysseus: image: odysseusol/odysseus restart: always ports: - "8889:8888" volumes: - ./odysseushome:/var/lib/odysseus volumes: mongodbdata: backenddata: odysseushome: |
At last you can start the stack with:
docker-compose up -d |
Before adding the worker nodes to Odysseus Studio, stop the OdysseusNet feature in the nodes tab, it will be started automatically when adding the nodes dynamically with OdysseusScript.
Before you can install queries on the nodes, you have to add the worker nodes with:
#CONFIG DISTRIBUTE true #PARSER PQL #DEFINE HOST 192.168.188.36 #ADD_NODE (name=worker_1) (serverAddress=${HOST}) (serverPort=18881) (username=System) (password=manager) #ADD_NODE (name=worker_2) (serverAddress=${HOST}) (serverPort=18882) (username=System) (password=manager) #ADD_NODE (name=worker_3) (serverAddress=${HOST}) (serverPort=18883) (username=System) (password=manager) |
Run the following OdysseusScript query in Odysseus Studio.
#REMOTEQUERY (name=worker_1) #PARSER PQL #RUNQUERY nexmark:person := ACCESS({source='nexmark:person', wrapper='GenericPush', transport='TCPClient', protocol='SizeByteBuffer', dataHandler='Tuple', options=[ ['host', 'nexmark'], ['port', '65440'], ['ByteOrder', 'LittleEndian'] ], schema=[ ['timestamp', 'STARTTIMESTAMP'], ['id', 'INTEGER'], ['name', 'STRING'], ['email', 'STRING'], ['creditcard', 'STRING'], ['city', 'STRING'], ['state', 'STRING'] ] }) #RUNQUERY out = nexmark:person |
Remark, that the host is 'nexmark' instead of 'localhost', because the remote worker node 'worker_01' is started in the default docker network and access the nexmark by the DNS resolved service name 'nexmark' |
Here is described how you can see the remote query output.
You can the following query in Odysseus Studio to access the output (Reading Query Results via REST/Websocket):
#PARSER PQL #RUNQUERY in = ACCESS({ transport = 'WebsocketClient', wrapper = 'GenericPush', datahandler = 'Tuple', protocol = 'CSV', source = 'access', readMetadata = 'true', options = [ ['uri','ws://localhost:18881/queries/0/a2e1a574-6c72-4961-b304-111c01708445/0/CSV/fif8htets00ql9q71okbuu7qs9'] ], schema = [['timestamp','StartTimestamp'],['id','Integer'],['name','String'],['email','String'],['creditcard','String'],['city','String'],['state','String']] } ) |
You have to adjust the optioins-uri in the query with the rootOperators-websockets-CSV uri from http://localhost:18881/queries/0
You can then show the last x elements:
In Webstudio you can see the query output unter the queries tab: