JarCasting
JarCasting JarCasting

RabbitMQ High Availability Client

RabbitMQ High Availability Client provides auto-reconnect functionality and call replay for RabbitMQ which takes the burden off of your client code.

License

License
Categories

Categories

AWS Container PaaS Providers CLI User Interface
GroupId

GroupId

com.jasonclawson
ArtifactId

ArtifactId

rabbitmq-ha-client
Last Version

Last Version

0.5.2
Release Date

Release Date
Type

Type

jar
Description

Description

RabbitMQ High Availability Client
RabbitMQ High Availability Client provides auto-reconnect functionality and call replay for RabbitMQ which takes the burden off of your client code.
Project URL

Project URL

https://github.com/jclawson/rabbitmq-ha-client/
Source Code Management

Source Code Management

https://github.com/jclawson/hazeltask/

Download rabbitmq-ha-client

Filename Size
rabbitmq-ha-client-0.5.2.pom
rabbitmq-ha-client-0.5.2.jar 17 KB
rabbitmq-ha-client-0.5.2-sources.jar 9 KB
rabbitmq-ha-client-0.5.2-javadoc.jar 59 KB
Browse

How to add to project

<!-- https://jarcasting.com/artifacts/com.jasonclawson/rabbitmq-ha-client/ -->
<dependency>
    <groupId>com.jasonclawson</groupId>
    <artifactId>rabbitmq-ha-client</artifactId>
    <version>0.5.2</version>
</dependency>
// https://jarcasting.com/artifacts/com.jasonclawson/rabbitmq-ha-client/
implementation 'com.jasonclawson:rabbitmq-ha-client:0.5.2'
// https://jarcasting.com/artifacts/com.jasonclawson/rabbitmq-ha-client/
implementation ("com.jasonclawson:rabbitmq-ha-client:0.5.2")
'com.jasonclawson:rabbitmq-ha-client:jar:0.5.2'
<dependency org="com.jasonclawson" name="rabbitmq-ha-client" rev="0.5.2">
  <artifact name="rabbitmq-ha-client" type="jar" />
</dependency>
@Grapes(
@Grab(group='com.jasonclawson', module='rabbitmq-ha-client', version='0.5.2')
)
libraryDependencies += "com.jasonclawson" % "rabbitmq-ha-client" % "0.5.2"
[com.jasonclawson/rabbitmq-ha-client "0.5.2"]

Dependencies

compile (3)

Group / Artifact Type Version
org.projectlombok : lombok jar 0.11.6
com.rabbitmq : amqp-client jar 3.1.4
org.slf4j : slf4j-api jar 1.7.5

Project Modules

There are no modules declared in this project.

Hazeltask

An advanced general purpose task distribution system based on Hazelcast with a familiar ExecutorService API.

Hazeltask was originally designed to provide large multi-tenent SaaS applications with an easy to use general purpose work distribution platform focusing on fair customer task load balancing. It has evolved from this mission to be much more powerful and configurable but still really easy to use!

Top Features

  1. Fault tolerant with task recovery
  2. Automatic node task rebalancing
  3. Configurable task prioritization
  4. ExecutorService API

FAQ

Why should I use this instead of Hazelcast's build in distributed ExecutorService
Put simply, if you are not using member-targeted DistributedTasks or MultiTasks, you should absolutely use this library. At the very least, it will give you failover recovery. No MemberLeftExceptions. And auto-balancing.
What can Hazeltask NOT do, that Hazelcast's ExecutorService CAN
The only thing Hazeltask doesn't do is handle instances of DistributedTask and MultiTask. This is because Hazeltask does the task distribution, and it only submits a task for execution on a single member. Its simply not part of the design.

Features, Expanded

This project encompasses an advanced distributed work library for Hazelcast.
It is modeled after the ExecutorService API but adds a lot of missing features the built in Hazelcast executor service doesn't provide such as:

  • Failover capabilities when nodes go down. No lost work!
  • Work distribution loadbalancing with customizable prioritizers (RoundRobin, Enum, and Fair provided... or create your own)
  • Push model is light on Hazelcast with few cluster operations and no cluster locking
  • No more MemberLeftExceptions when waiting on Futures! Work will be redistributed to another member and executed returning the result to your Future
  • Google Guava ListenableFuture support for easy callbacks and task chaining!

Example Use Cases:

Multi-tenent Task Queue

Let's say you have a multi-tenent system where users can submit hundreds of pages of scanned documents that you need to OCR. You want to guarantee that you will OCR all of the pages sent, and furthermore guarantee that one user cannot starve out another. For example, if user A submits 20,000 pages and then user B submits 1 page. You don't want user B to wait a really long time before their 1 page is worked on. You can use this library, group tasks by userId and use the RoundRobin router to solve this.

Priority Task Queue

The easiest way to implement this is by grouping tasks into priorities based on an enum like HIGH, MEDIUM, LOW. Hazeltask comes with built-in support for prioritizing tasks based on an enum.

Advanced Multi-tenent Priority Task Queue

Oh yes you can. Its easy to combine the examples above and provide per-customer HIGH, MEDIUM, and LOW prioritization. You can implement any kind of prioritization you can think of. And... priorities can CHANGE. This allows you to implement features like: "Bump customer 7 to the highest priority for the next hour".

Future Plans

  • Migrate to Hazelcast 3.0 to take advantage of SPI and custom datastructure implementation
  • Allow task retries if a task fails to execute due to an exception in user code. Retry after _ time option.

Change Log

2.0.0-SNAPSHOT

  • Refactored for Hazelcast 3.1 support
  • Depend on codahale metrics 3.0.1

Versions

Version
0.5.2
0.5.1
0.5.0