Skip to main content
Version: Current

Docker Setup Tutorial

You can start your journey with Nussknacker using fully-featured docker-based environment. The dependencies and manual steps were reduced to minimum to allows to start interacting with Nussknacker on your local machine.

A very concise Nussknacker terminology primer​

In Nussknacker, scenario describes a decision algorithm i.e. how data records should be processed and what actions should be performed. Scenarios are composed of nodes, which work with data records. Nodes can organize the flow, transform data carried in data records, bring additional data ("enrich"), call actions in external systems, etc.

Prerequisites​

  • Docker (20.10.14+) installed
  • unused ports: 8080, 8181
  • MacOS, Windows (10, 11), Linux

Running​

Download and run

Run in the console:

curl -o- https://raw.githubusercontent.com/TouK/nussknacker-quickstart/main/download-and-start.sh | bash

And then wait a little bit and open http://localhost:8080/ in your favorite browser and log in as admin:admin

You can interact with the solution using the following user interfaces:

What's in the Quickstart?​

The Quickstart bootstraps Nussknacker and its dependencies in docker containers.

Click to read about the exposed services details
  • Nussknacker Designer - Nu scenarios are developed and managed from it
  • Postgres - database for the Designer
  • Flink (Job Manager & Task Manager) - a powerful steaming processing framework
  • Kafka - broker for steaming data
  • Schema Registry - schema for the data in Kafka's topics
  • AKHQ - UI for Kafka & Schema Registry
  • Grafana & InfluxDB & Telegraf - for scenarios monitoring
  • Nginx - gateway for Designer and other UIs
  • Quickstart setup - automatically creates example scenarios and generates sample Streaming and Request-Response data, creates Kafka's topics and topics' schemas.

After logging in to the Nu Designer, in the scenarios list, you will find example scenarios. They are fully functional and working.

Streaming example​

With DetectLargeTransaction scenario we want to show you a Streaming mode

The scenario logic is pretty simple: it reads messages from Kafka's Transactions topic and the large ones (which amount is greater than 20), it sends them to audit (to Kafka's ProcessedTransactions topic).

Input data (transaction messages) are being generated by one of the Quickstart services. You can watch them in AKQH. The scenario is already deployed and is being executed by the Flink engine. You can verify that by checking the Flink Dashboard. The "large" transactions are sent to ProcessedTransactions topic.

As a next step we'd advise you to:

  1. see the metrics of the scenario - click the metrics button
  2. see how many messages have been processed so far, and by which nodes - click the counts button
  3. try to run Nu Tests to see how the messages are processed - click the generated and/or ad-hoc buttons (you may want to read more about the Test functionality first)
  4. try to modify the scenario - drag&drop and attach a new component from the Creator Panel and/or modify the existing scenario nodes
  5. save the modified scenario and deploy the changes - click the save and then deploy button (you can go back to previous steps to verify if your changes work as expected)
Docker quickstart streaming scenario example - Designer window view.

Streaming scenario example. UI panels placement may differ from the actual one.

Request-Response example​

The next example is LoanRequest scenario. It shows a Request-Response mode. The scenario logics describes a simple method of determining maximum of loan amount based on loan type (passed in a requestType field of the input request) and a city (also from the request data).

Input data (HTTP requests) are being generated by one of the Quickstart services. The scenario is already deployed and is being executed by the Lite engine.

As a next step we'd advise you to:

  1. see the metrics of the scenario - click the metrics button
  2. see how many requests have been processed so far, and by which nodes - click the counts button
  3. try to run Nu Tests to see how the requests are processed - click the generated and/or ad-hoc buttons (you may want to read more about the Test functionality first)
  4. you can send a request by your own, directly to the running scenario:
POST http://localhost:8181/scenario/loan

Body:

{"customerId": "client01", "requestedAmount": 500, "requestType": "mortgage", "location": { "city": "Warszawa", "street": "Obozowa" }}
  1. try to modify the scenario - drag&drop and attach a new component from the Creator Panel and/or modify the existing scenario nodes
  2. save the modified scenario and deploy the changes - click the save and then deploy button (you can go back to previous steps to verify if your changes work as expected)
Docker quickstart Request-Response scenario example - Designer window view.

Streaming scenario example. UI panels placement may differ from the actual one.

Cleaning up​

To stop the Quickstart environment and clean the resources, just run stop-and-clean.sh script (you will find in the downloaded Quickstart folder) and wait a few seconds.