Synopsis
Titanoboa is fully distributed, highly scalable and fault tolerant workflow orchestration platform. It employs hybrid iPaaS concepts and runs on the JVM. You can run it on-premises or in a cloud.
Due to its generic, distributed and easily extensible design, you can use titanoboa for wide variety of purposes.
You can use it:
- as a Service Bus (ESB)
- as a full-featured iPaaS / Integration Platform
- for Big Data processing
- for IT Automation
- for Batch Processing
- for Data Transformations / ETL
See also titanoboa.io and our wiki. Predefined workflow steps are here.
Motivation
Titanoboa has been created with aim to create a workflow platform, that would support all the usual features of complex workflow engines and majority of enterprise integration patterns including:
- sequential and/or parallel step execution
- configurable step retry upon error and advanced customizable error handling
- advanced branching and conditional flow
- potentially cyclic workflow graphs
- splitter and aggregator (aka map/reduce) patterns which allow for processing of larger data sets
- fully transactional nature that assures that all steps are executed even in case of a failover
- full extensibility and ability to rapidly develop and deploy new workflows during runtime
In addition to this titanoboa also strives to honor immutability and functional programming principles. This enables it to be fully distributed and to be highly available with no central master server or database. This also helps lift performance limitations and thus titanoboa can be suitable for not only batch processing, but also for performance critical workflows.
Titanoboa's in-browser GUI can be used not only to monitor workflow jobs and server configuration, but provides an in-build IDE with workflow visualisation, advanced data and properties editor, autcomplete functionality and a repl (even for java!) so as users can rapidly test-develop new workflows directly in there.
Titanoboa is designed for both java & clojure developers and we are striving to make it usable even for java developers with no prior clojure knowledge.
Predefined Steps
Though it is easy to rapidly develop new workflow steps directly from GUI, there is a number of ready-made steps and tasklets in this repository. Some of the steps included are:
- AWS (EC2, S3, SNS, SQS, SES)
- Http Client
- Smtp Client
- Sftp
- PDF Generation
and more.
Prerequisites
Java 8 or higher.
Installation
We suggest giving titanoboa's GUI a try as well since it is the best starting point!
Download titanoboa's distribution including its GUI from https://www.titanoboa.io/distributions/titanoboa-0.8.2_gui_non-commercial_use_only.zip
curl --remote-name https://www.titanoboa.io/distributions/titanoboa-0.8.2_gui_non-commercial_use_only.zip
Note: The GUI is free for non-commercial use only
Note: If you are intending on running titanoboa server on java 8 JRE, download a distribution for JRE instead:
curl --remote-name https://www.titanoboa.io/distributions/titanoboa-0.8.2_jre_gui_non-commercial_use_only.zip
Unzip the file:
unzip titanoboa-0.8.2_gui_non-commercial_use_only.zip
then execute the start script:
./start
In your console you should see bunch of log messages and ultimately you will see
INFO [main] - Started @2338ms
which means the server started successfully. By default both the server and the GUI will start on port 3000 and you can open http://localhost:3000 in your browser.
Congratulations! You have just started your titanoboa server!
You can go ahead and try to create a sample workflow.
Installing server without GUI
Download the latest release from https://www.titanoboa.io/distributions/titanoboa-0.8.2.zip .
curl --remote-name https://www.titanoboa.io/distributions/titanoboa-0.8.2.zip
Note: If you are intending on running titanoboa server on java 8 JRE, download a distribution for JRE instead:
curl --remote-name https://www.titanoboa.io/distributions/titanoboa-0.8.2_jre.zip
And then follow the instructions above. By default the server will start on port 3000.
Building from the repository
Titanoboa uses leiningen for dependency management, so if you don't have it download it from https://leiningen.org/ and follow its installation instructions.
Clone this repository
git clone https://github.com/mikub/titanoboa
run lein package:
lein package
in case that your target platform is Java 8 JRE use following profile:
lein with-profile java8jre package
this will generate titanoboa jar file in the build directory and will also download all libraries into lib folder.
If you want to use GUI you can clone the titanoboa-gui repository as well:
git clone https://github.com/mikub/titanoboa-gui
Note: The GUI is free for non-commercial use only
then merge GUI's public folder into the uberjar:
zip -r -g ./build/titanoboa.jar public
then run the start script:
./start
In your console you should see bunch of log messages and ultimately you will see
INFO [main] - Started @3238ms
which means the server started successfully. By default the server will start on port 3000.
Congratulations! You have just built & started your titanoboa server!
If you included GUI you can go ahead and try to create a sample workflow.
Server Configuration
Server configuration and external dependencies file can be specified by system properties boa.server.config.path and boa.server.dependencies.path:
java -Dboa.server.config.path=boa_server_config_local.clj -Dboa.server.dependencies.path=ext-dependencies.clj -cp "./build/titanoboa.jar:./lib/*" titanoboa.server
See Server configuration wiki for more details.
Getting Started
Before you start, it might be a good idea to get familiar with titanoboa's concepts & workflow design principles.
Develop & Test Workflows with titanoboa GUI
Titanoboa GUI is a good place to start devloping and testing workflows:
See an example in our wiki on how to create a sample workflow.
Develop & Test Workflows Locally in Your Clojure REPL
If you cannot use GUI and do not want to use REST API, you can as well just start REPL locally and play with titanoboa there. Either build titanoboa from repo or get it as leiningen or maven dependency:
[io.titanoboa/titanoboa "0.7.3"]
<dependency>
<groupId>io.titanoboa</groupId>
<artifactId>titanoboa</artifactId>
<version>0.7.3</version>
</dependency>
then fire off a repl and start:
Define a sample "Hello World!" workflow:
(ns local-system-test
(:require [titanoboa.system]
[titanoboa.processor]))
(defn hello-fn [p]
{:message (str "Hello " (or (:name p) "human") "!")
:return-code (nil? (:name p))})
(defn greet-fn [p]
{:message (str (:message p) " Nice to meet you!")})
(defn fill-in-blanks [p]
{:message (str (:message p) " What is your name?")})
(def job-def {:first-step "step1"
:name "test"
:properties {:name nil}
:steps [{:id "step1"
:type :custom
:supertype :tasklet
:next [[false "step2"] [true "step3"]]
:workload-fn 'local-system-test/hello-fn
:properties {}}
{:id "step2" :type :custom :supertype :tasklet
:workload-fn 'local-system-test/greet-fn
:next []
:properties {}}
{:id "step3" :type :custom :supertype :tasklet
:workload-fn 'local-system-test/fill-in-blanks
:next []
:properties {}}]})
Start a simple local system with workers that will process the job:
(def new-jobs-chan (clojure.core.async/chan (clojure.core.async/dropping-buffer 1024)))
(def jobs-chan (clojure.core.async/chan (clojure.core.async/dropping-buffer 1024)))
(def finished-jobs-chan (clojure.core.async/chan (clojure.core.async/dropping-buffer 1024)))
(titanoboa.system/start-system! :core-local
{:core-local {:system-def #'titanoboa.system.local/local-core-system
:worker-def #'titanoboa.system.local/local-worker-system
:worker-count 2}}
{:new-jobs-chan new-jobs-chan
:jobs-chan jobs-chan
:finished-jobs-chan finished-jobs-chan
:node-id "localhost"
:eviction-interval (* 1000 60 5)
:eviction-age (* 1000 60 10)
:jobs-repo-path "repo-test/"
:job-folder-path "job-folders/"})
INFO [nREPL-worker-0] - Starting system :core-local ...
INFO [nREPL-worker-0] - Starting CacheEvictionComponent...
INFO [CacheEvictionComponent thread 0] - Starting CacheEvictionComponent thread [ CacheEvictionComponent thread 0 ].
INFO [nREPL-worker-0] - Starting action processor pool...
INFO [nREPL-worker-0] - Starting to watch repo folder for changes: dev-resources/repo-test/
INFO [nREPL-worker-0] - System :core-local started
=> true
(titanoboa.system/start-workers! :core-local
{:core-local {:system-def #'titanoboa.system.local/local-core-system
:worker-def #'titanoboa.system.local/local-worker-system
:worker-count 2}})
INFO [nREPL-worker-1] - Starting 2 workers for system :core-local :
INFO [nREPL-worker-1] - Starting a worker for system :core-local ...
INFO [nREPL-worker-1] - Starting job worker....
INFO [nREPL-worker-1] - Starting a worker for system :core-local ...
INFO [nREPL-worker-1] - Starting job worker....
=> nil
Start the job:
(titanoboa.processor/run-job! :core-local
{:jobdef job-def
:properties {:name "World"}}
true)
INFO [nREPL-worker-2] - Submitting new job [] into new jobs channel...
INFO [async-thread-macro-2] - Initializing a new job; First step will be: [ step1 ]
INFO [async-thread-macro-2] - Retrieved job [ d673c759-4fc6-4af1-bdad-d1dfd0f50f22 ] from jobs channel; Starting step [ step1 ]
INFO [async-thread-macro-2] - Next step is step2 ; Submitting into jobs channel for next step's processing...
INFO [async-thread-macro-1] - Initializing a next step; next step [ step2 ] was found among steps as step [ step2 ]
INFO [async-thread-macro-2] - Acking main message for step step1 with thread stack nil
INFO [async-thread-macro-1] - Retrieved job [ d673c759-4fc6-4af1-bdad-d1dfd0f50f22 ] from jobs channel; Starting step [ step2 ]
INFO [async-thread-macro-1] - Looping through finalize-job! fn with thread-stack: []
INFO [async-thread-macro-1] - Acking main message for step step2 with thread stack nil
INFO [async-thread-macro-1] - Job d673c759-4fc6-4af1-bdad-d1dfd0f50f22 has finshed.
=>
{:properties {:name "World", :message "Hello World! Nice to meet you!"},
:step-start #inst"2018-11-23T07:35:34.218-00:00",
:step-retries {},
:tracking-id nil,
:step-state :completed,
:start #inst"2018-11-23T07:35:34.203-00:00",
:history [{:step-state :completed,
:start #inst"2018-11-23T07:35:34.210-00:00",
:duration 6,
:result false,
:node-id "localhost",
:id "step1",
:next-step "step2",
:exception nil,
:end #inst"2018-11-23T07:35:34.216-00:00",
:retry-count nil,
:thread-stack nil,
:message "Step [step1] finshed with result [false]\n"}
{:id "step2",
:step-state :running,
:start #inst"2018-11-23T07:35:34.218-00:00",
:node-id "localhost",
:retry-count 0}
{:step-state :completed,
:start #inst"2018-11-23T07:35:34.218-00:00",
:duration 1,
:result nil,
:node-id "localhost",
:id "step2",
:next-step nil,
:exception nil,
:end #inst"2018-11-23T07:35:34.219-00:00",
:retry-count nil,
:thread-stack nil,
:message "Step [step2] finshed with result []\n"}],
:duration 16,
:state :finished,
:jobid "d673c759-4fc6-4af1-bdad-d1dfd0f50f22",
:create-folder? true,
:node-id "localhost",
:next-step nil,
:end #inst"2018-11-23T07:35:34.219-00:00"}
When you are done testing you may want to stop the system:
(titanoboa.system/stop-all-systems!)
INFO [nREPL-worker-3] - Stopping all workers for system :core-local
INFO [nREPL-worker-3] - Stopping job worker gracefully; sending a stop signal to the worker via service bus....
INFO [nREPL-worker-3] - Stopping job worker gracefully; sending a stop signal to the worker via service bus....
INFO [nREPL-worker-3] - Stopping system :core-local ...
INFO [nREPL-worker-3] - Stopping action processor pool...
INFO [nREPL-worker-3] - Stopping CacheEvictionComponent thread [ CacheEvictionComponent thread 0 ]...
Developing custom workflow steps in Java
Titanoboa is also meant to be used by java developers who (apart from few concepts like EDN) do not need to be familiar with clojure. If you do not want to use clojure java interop to instantiate your objects and/or invoke your methods, you also have other options:
To create a custom workflow step, simply add a (maven) dependency on 
to your project.
and create a class that will implement io.titanoboa.java.IWorkloadFn interface:
public interface IWorkloadFn {
public Object invoke (Map properties);
}
If you then add your project (or the corresponding maven artifact) to titanoboa's external dependencies, you can use your class name in the workflow-fn field. The class will be automatically instantiated as a singleton bean (so it has to have a constructor with no argumet) and all subsequent references to it from any workflow-fn will invoke its invoke method:
:workload-fn io.titanoboa.java.SampleWorkloadImpl
or
:workload-fn 'io.titanoboa.java.SampleWorkloadImpl
or in GUI:
Java lambda support
To rapidly development and test new steps, you can also type a lambda function in the GUI and have titanoboa evaluate it during runtime:
License
Copyright © Miroslav Kubicek
Titanoboa is dual-licensed under either AGPL license or a Commercial license.
Can you improve this documentation? These fine people already did:
Miro Kubicek, mikub & UnknownEdit on GitHub
cljdoc is a website building & hosting documentation for Clojure/Script libraries
× close