Camunda Workflow uses Spyke to communicate with the Camunda REST API. It executes External Service Tasks using a Ruby class corresponding to each Camunda external task.
The Camunda model process definition key is the module name of your implementation classes.
An external task is created with a Ruby class name for the id. And the process definition key should be set as the topic name.
Tasks are fetched, locked and then queued. We expect classes (ActiveJob) to implement each external task. So, according to the above screenshots, the poller and queuer will expect a class called SomeProcess::SomeTask
to be implemented in your app.
Worker classes should inherit from CamundaJob
. It is generated with rails g camunda:install
.
perform
is implemented on Camunda::ExternalTaskJob
. It calls bpmn_perform
with variables from Camunda and returns the results back to Camunda.
class CamundaJob < ApplicationJob
# If using Sidekiq, change to include Sidekiq::Worker instead of inheriting from ApplicationJob
include Camunda::ExternalTaskJob
end
class SomeProcess::SomeTask < CamundaJob
def bpmn_perform(variables)
do_something(variables[:foo])
# A hash returned will become variables in the Camunda BPMN process instance
{ foo: 'bar', foo2: { json: "str" }, array_var: ["str"] }
end
end
If your implementation throws an exception, it is caught and then sent to Camunda with a stack trace.
Camunda supports throwing bpmn exceptions on a service task to communicate logic errors and not underlying code errors. These expected errors are thrown with
class SomeProcess::SomeTask < CamundaJob
def bpmn_perform(variables)
result = do_something(variables[:foo])
if result == :expected
{ foo: 'bar', foo2: { json: "str" }, array_var: ["str"] }
else
raise Camunda::BpmnError.new error_code: 'bpmn-error', message: "Special BPMN error", variables: { bpmn: 'error' }
end
end
end
rails generate camunda:spring_boot
Creates a skeleton Java Spring Boot app, which also contains the minimal files to run unit tests on a BPMN file. The Spring boot app can be used to start a Camunda instance with a REST api and also be deployed to PCF by generating a Spring Boot jar and pushing it.
rails generate camunda:install
Creates app/jobs/camunda_job.rb
. A class that inherits from ApplicationJob and includes ExternalTaskJob
. It can be changed to include Sidekiq::Worker
instead.
All of the BPMN worker classes will inherit from this class
rails generate camunda:bpmn_classes
Parses the BPMN file and creates task classes according to the ID of the process file and the ID of each task. It checks each task and only creates it if the topic name is the same as the process ID. This allows one to have some tasks be handled outside the Rails app. It confirms that the ID's are valid Ruby constant names.
Java 7 and Apache Maven are requirements to run the Camunda server. We are using the Spring distribution. The Camunda application has a pom.xml
, which Maven uses to install the required dependencies.
Start the application:
cd bpmn/java_app
mvn spring-boot:run
# Or use the included rake task:
# Start the Camunda spring boot app with `mvn spring-boot:run`
rake camunda:run
If you create Java based unit tests for your bpmn file they can be run with an included rake task as well.
cd bpmn/java_app
mvn clean test
# Or use the included rake task:
# Runs spring boot test suite with `mvn clean test`
rake camunda:test
Camunda-workflow defaults to an in-memory, h2 database engine. If you rather use a Postgres database engine, comment out the h2 database engine settings in the pom.xml
file located in bpmn/java_app
. Default settings for using Postgres are available in the pom.xml
file.
You will need to create a Postgres database on localhost called camunda
.
The default engine route prefix for the provided Java Spring Boot app is rest
. If you choose to download and use the Camunda distribution, the engine prefix is rest-engine
. Camunda-workflow is configured to use rest
by default.
To override the default engine route prefix, you need to add an initializer file in your rails app.
# filename initializers/camunda.rb
Camunda::Workflow.configure do |config|
config.engine_route_prefix = 'rest-engine'
end
Authentication can be enabled in the Camunda Java Spring Boot app by setting an environment variable CAMUNDA_AUTH
to true
or false
or by setting the camunda.authentication
variable located in the application.properties
(bpmn/java_app/src/main/resources) file to true
.
When HTTP Basic Auth is enabled, it's required that a user with the appropriate permissions is setup in Camunda. Otherwise, the request will return as 401 unauthorized
. Users are set up within the admin dashboard of Camunda and used to authenticate by passing an Authorization header during requests to the REST API. Below is how to configure the camunda_user
and camunda_password
to be used in the header request to authenticate using HTTP Basic Auth.
# filename initializers/camunda.rb
Camunda::Workflow.configure do |config|
config.camunda_user = ENV['CAMUNDA_USER']
config.camunda_password = ENV['CAMUNDA_PASSWORD']
end
mvn package spring-boot:repackage
The jar is in target/camunda-bpm-springboot.jar
cf push app_name -p target/camunda-bpm-springboot.jar
It will fail to start. Create a postgres database as a service in PCF and bind it to the application. The Springboot application is configured for Postgres and will then be able to start.
gem 'camunda-workflow'
Uses a default name, etc. Below outlines how to deploy a process using the included sample.bpmn file created by the generator. Alternatively you can deploy using Camunda Modeler
Camunda::Deployment.create file_names: ['bpmn/diagrams/sample.bpmn']
start_response = Camunda::ProcessDefinition.start_by_key'CamundaWorkflow', variables: { x: 'abcd' }, businessKey: 'WorkflowBusinessKey'
Camunda cannot handle snake case variables, all snake_case variables are serialized to camelCase before a request is sent to the REST api. Variables returned back from the Camunda API will be deserialized back to snake_case.
{ my_variable: "xyz" }
will be converted to:
{ myVariable: "xyz" }
Camunda::ProcessInstance.destroy_existing start_response.id
The poller runs an infinite loop with long polling to fetch tasks, queue, and run them. The topic is the process definition key, as shown in the screenshot example from the Camunda Modeler.
Below will run the poller to fetch, lock, and run a task for the example process definition located in the starting a process
detailed above.
Poller will need to run in a separate process or thread and needs to be running constantly in order to poll Camunda and queue jobs.
Camunda::Poller.fetch_and_queue %w[CamundaWorkflow]
We have had success with running a long running thread in a Rails app using Rufus Scheduler. Something like:
rufus_scheduler.in('10.seconds') do
Camunda::Poller.fetch_and_queue %w[Topics]
end
tasks = Camunda::ExternalTask.fetch_and_lock %w[CamundaWorkflow]
tasks.each(&:run_now)
# Or you can query Camunda::Task with other parameters like assignee task.complete!(var1: 'value')
Camunda::Task.find_by_business_key_and_task_definition_key!(instance_business_key, task_key).complete!
RSpec helpers validate your application to make sure it has a class for every External task in a given BPMN file.
require 'camunda/matchers'
RSpec.describe "BPMN Diagrams" do
describe Camunda::BpmnXML.new(File.open("bpmn/diagrams/YourFile.bpmn")) do
it { is_expected.to have_module('YourModule') }
it { is_expected.to have_topics(%w[YourModule]) }
it { is_expected.to have_defined_classes }
end
end
See CONTRIBUTING for additional information.
This project is in the worldwide public domain. As stated in CONTRIBUTING:
This project is in the public domain within the United States, and copyright and related rights in the work worldwide are waived through the CC0 1.0 Universal public domain dedication.
All contributions to this project will be released under the CC0 dedication. By submitting a pull request, you are agreeing to comply with this waiver of copyright interest.