BoxLang 🚀 A New JVM Dynamic Language Learn More...
RabbitMQ is an open-source message-broker software that originally implemented the Advanced Message Queuing Protocol and has since been extended with a plug-in architecture to support Streaming Text Oriented Messaging Protocol, MQ Telemetry Transport, and other protocols.
This library is a wrapper for CFML/ColdFusion apps to be able to interact with RabbitMQ via the Java SDK.
Apache License, Version 2.0.
Previous Adobe versions may work, I just haven't tested them yet.
Install into your modules folder using the box
cli to install
box install rabbitsdk
Due to a nasty classloading issue with Lucee, I had to abandon using Javaloader: https://luceeserver.atlassian.net/browse/LDEV-2296
As such, it is required for you to be responsible for loading the Rabbit jar into your application.
If this is a CFML web app, you can add this to your Application.cfc
this.javaSettings = {
loadPaths = directorylist( expandPath( '/modules/rabbitsdk/lib' ), true, 'array', '*jar' ),
loadColdFusionClassPath = true,
reloadOnChange = false
};
Or if you are using this module from the CLI, you can load the jars in a task runner or custom command in CommandBox prior to using the module like so:
classLoad( 'expandPath( '/rabbitsdk/lib' )' );
This module wraps and simplifies the java SDK. There are only a few CFCs for you to worry about, and while not 100% of the Java SDK functionality is exposed, all the major functions are here.
Here are the major CFCs you need to know about:
Create this only once and re-use over the life of your app. The CFC is marked as a singleton, so if you are using WireBox to access it, you don't need to manually cache it anywhere.
Simply inject it wherever you need and use it. It is very important to call the shutdown()
method to release the connection when your app is shutting down or reiniting.
wirebox.getInstance( 'RabbitClient@rabbitsdk' );
or
property name='rabbitClient' inject='RabbitClient@rabbitsdk';
Each RabbitClient
instance contains a single connection which is auto-created on first use if you have supplied the connection details in the module settings in /config/Coldbox.cfc
.
moduleSettings = {
// here we are pulling from environment variables, by
// but these can also just be hard-coded in
rabbitsdk = {
host : getSystemSetting( 'rabbit_host' ),
username : getSystemSetting( 'rabbit_username' ),
password : getSystemSetting( 'rabbit_password' ),
virtualHost : getSystemSetting( 'rabbit_virtualHost' ),
useSSL : getSystemSetting( 'rabbit_useSSL' ),
sslProtocol : getSystemSetting( 'rabbit_sslProtocol' )
}
};
You can also manually create your connection, which only needs to happen once:
rabbitClient.connect( host='localhost', port=5672, username='guest', password='guest', virtualHost='/');
Or, for SSL-enabled connections:
rabbitClient.connect( host='localhost', port=5671, username='guest', password='guest', virtualHost='/', useSSL = true, sslProtocol = 'TLSV1.2');
In ColdBox, the RabbitClient will register a prereinit
interceptor listener to shut itself down when the framework reinits. If you have another scenario such as a test case suite that simply clears the application scope
when it's finished, it will leave hanging connections which you can see in the web-based management UI.
If you want to have more than one connection, then create additional WireBox mappings for this CFC and create them individually.
// In your /config/Wirebox.cfc
map( 'PublishClient' ).to( 'rabbitsdk.models.RabbitClient' );
map( 'ConsumerClient' ).to( 'rabbitsdk.models.RabbitClient' );
// In your application
wirebox.getInstance( 'ConsumerClient' )
.startConsumer(
queue='myQueue',
consumer=(message, channel, log)=>{
log.info( 'Message received: #message.getBody()#' );
} );
wirebox.getInstance( 'PublishClient' )
.publish( 'My message', 'myQueue' )
.publish( 'Another message', 'myQueue' )
.publish( 'Still more messages', 'myQueue' );
Interactions with Rabbit actually happen over a channel, but the rabbitClient has a convenience method for all operations that automatically create and close the channel for you.
All methods that don't return some explicit value return this
so you can do method chaining like so:
rabbitClient
.queueDeclare( 'myQueue' )
.queuePurge( 'myQueue' )
.publish( 'My message', 'myQueue' );
If you have a large number of operations and you want to perform them on the same channel then the best approach is to use the batch()
method in the client CFC. It's less network traffic to re-use the same channel
so sending 1,000 messages is something you'd want to do over a single channel for the best performance.
rabbitClient.batch( (channel)=> {
channel
.queueDeclare( 'myQueue' )
.queuePurge( 'myQueue' )
// This could easily be a dynamic loop over a query, etc
.publish( 'Message 1', 'myQueue' )
.publish( 'Message 2', 'myQueue' )
.publish( 'Message 3', 'myQueue' )
.publish( 'Message 4', 'myQueue' );
});
The batch()
method accepts a closure and
You can minimally publish a message with the routing key and message text.
rabbitClient.publish( 'My message', 'myQueue' );
You can publish complex data like structs or arrays and the message will be automatically JSON serialized for you and deserialized when you read the message back.
var data = {
name:'brad',
age:40,
hair:'red',
likes:[
'music',
'computers',
'procrastinating'
]
};
rabbitClient.publish( data, 'myQueue' );
Rabbit also allows you to send additional properties for a message. Most properties are a string value, but headers are a struct of name/value pairs. You can see a full list
of supported properties below in the Message
section.
rabbitClient.publish(
body='My Message',
routingKey='myQueue',
props={
'contentEncoding':'UTF-8',
'contentType':'text/plain',
'deliveryMode':1,
'headers':{
'header 1' :'value 1',
'header 2' :'value 2',
'header 3' :'value 3'
},
'priority':5
}
);
You can consume a single message at a time like so. See the Message
section below.
var message = rabbitClient.getMessage( 'myQueue' );
The recommended way to process messages is to start a consumer. Each channel can have a single consumer started. A consumer only needs to be started once.
The consumer will run until it is stopped, the channel is closed, or the client is shutdown.
A consumer runs in a separate thread so starting the consumer is a non-blocking operation. You pass a UDF/closure or a CFC instance to the consumer which will be called once for every message that comes in.
Arguments:
Here is an example of using ad-hoc closures to handle messages and errors:
var channel = rabbitClient
.startConsumer(
queue='myQueue',
autoAcknowledge=false,
consumer=(message, channel, log)=>{
log.info( 'Consumer 1 Message received: #message.getBody()#' );
message.acknowledge();
},
error=(message, channel, log, exception)=>{
log.error( 'Error processing message #message.getBody()#. Error message: #exception.message#' );
} );
And here is an example of specying a component. Note, the consumer
and error
parameters default to onMessage
and onError
. They are provided here just for clarity, but could be omitted if not changed from the default. Also, component
can be either the instance or a CFC path or a WireBox mapping. wirebox.getInstance()
is used to create it.
var channel = getRabbitClient()
.startConsumer(
queue='myQueue',
autoAcknowledge=false,
component='MyConsumer,
consumer='onMessage',
error='onError'
);
// MyConsumer.cfc
component {
function onMessage( message, channel, log ) {
log.info( 'My Consumer received message #message.getBody()#' );
message.acknowledge();
}
function onError( message, channel, log, exception ) {
log.error( 'Error processing message #message.getBody()#. Error message: #exception.message#' );
}
}
The consumer
callback method (whether a closure or in a CFC) receives three arguments:
Message
CFC outlined belowThe error
callback method (whether a closure or in a CFC) receives four arguments:
Message
CFC outlined belowconsumer
method.The startConsume()
method returns a Channel
instance which you can use to stop the consumer later. When you start more than one consumer, a new channel is created for each one.
Save the channel refernce so you can stop the consumer later. Closing the channel will also stop the consumer.
var channel1 = rabbitClient.startConsumer( 'myQueue', ()=>{} );
var channel2 = rabbitClient.startConsumer( 'myQueue', ()=>{} );
var channel3 = rabbitClient.startConsumer( 'myQueue', ()=>{} );
// Later on, after you're done processing messages:
channel1.close()
channel2.close()
channel3.close()
When you have more than one consumer, they will compete for messages and work them in parallel. By default, a consumer will only pull one message at a time to work. This is ideal for several consumers completing tasks that take a while to process and you want to evenly distribute load among the consumers. If you have a very large number of messages and processing time is very fast, you can improve performance by increasing your prefetch amount so a consumer grabs more than one message to work at a time reduces network traffic between messages.
rabbitClient.startConsumer( name='myQueue', consumer=()=>{}, prefetch=10 );
The consumer above will fetch up to 10 messages at a time to process (if that many are available in the queue)
Transient that represents a message received from the server for you to process. Each message instance has the following getters you can call:
By default, messages will be automatically acknowledged, which removes them from the queue. For transient, or unimportant messages, this is fine. However, if you want to protect against an important
message getting lost, set autoAcknowledge
to false when getting your message or starting the consumer and explicitly acknowledge it like so:
message.acknowledge();
Do not call this method if autoAcknowledge
is enabled, or you will receive an error.
When you start a consumer thread, you can simply return true
from your callback UDF to acknowledge the message as processed.
If you are unable to process a message (such as a DB connection issue or other unrecoverable error) you can reject the message which will tell the Rabbit server that it was not processed.
message.reject();
Do not call this method if autoAcknowledge
is enabled, or you will receive an error.
When you start a consumer thread, you can simply return false
from your callback UDF to reject the message.
The default behavior when rejecting a message is NOT to requeue the message, meaning it is lost. If you want the message to be placed back in the queue for redelivery, then call reject like so:
message.reject( requeue=true );
Be careful, as this can create a huge amount of traffic if the same failing message gets redelivered over and over again. You probably want to log when messages are rejected and redelivered so you can keep an eye on why it is happening.
You can create, remove, and purge queues from the channel API.
Creating a queue is idempotent, meaning you can call this as many times as you like, and if the queue already exists, nothing happens.
rabbitClient.queueDeclare( 'myQueue' );
Creating a queue has the following additional arguments
A queue binding controls what exchange and routing keys will deliver to the queue.
rabbitClient
.queueDeclare( 'myQueue' )
.queueBind( queue='myQueue', exchange='amq.direct', routingKey='routing.key' );
rabbitClient.queueDelete( 'myQueue' );
Deleting a queue has the following additional arguments
Purging a queue removes all messages. This is probably only something you'd need for testing.
rabbitClient.queuePurge( 'myQueue' );
Returns true if queue exists, false if it doesn't. Be careful calling this method under load as it catches a thrown exception if the queue doesn't exist so it probably doesn't perform great if the queue you are checking doesn't exist most of the time.
var exists = rabbitClient.queueExists( 'myQueue' );
Returns count of the messages of the given queue. Doesn't count messages which waiting acknowledges or published to queue during transaction but not committed yet.
var count = rabbitClient.getQueueMessageCount( 'myQueue' );
All Rabbit interactions actually happen over a channel. All of the methods on the rabbitClient CFC automatically create and close a channel on your behalf for every operation. You can also create your own channel to use directly so you can re-use it. Do not share the same channel between threads. The channel's are technically thread safe, but only one thread can use the channel at a time so it would create an unwanted bottleneck.
var channel = rabbitClient.createChannel();
Make sure to call close()
on a channel when you're done. This is important or the channel will be left open forever and there is a limited number of channels that can be open for a connection.
In case you're code errors, it is recommended to use a finally
block to ensure the close always happens.
try {
var channel = rabbitClient.createChannel();
channel.queueDeclare( 'myQueue' );
} finally {
channel.close();
}
The way AMQP works is that if any communication error happens, the channel will be automatically closed which means no more actions can be take with it. Please account for this in your code and ask the client for a fresh channel if the previous one encounters an error.
RabbitMQ can be used for synchronous calls where the client blocks until a reply is received via the RPC Client. There are two pieces that must be in place, an RPC Server listening for calls, and an RPC client making them. The RPC Client "sends" requests by placing them into a queue which is being listened to by the RPC Server. The server will send the reply back using a direct-reply feature of Rabbit which behaves as a pseudo-queue to re-deliver the reply over the same channel as the original message.
You can start up as many RPC servers you like anywhere. Just specify the name of the queue they'll be listening to. An RPC Server is just a customized version of a consumer thread, but with special assumptions about the message semantics and auto-replying. One RPC Server can only process a single message at a time, so start more than one if you need high throughput.
An RPC server can use a UDF or a CFC instance to process the incoming calls. When using a UDF, the function will receive the name of the RPC method and args which may be an array or a struct. Just like starting a consumer, a channel reference will be passed back, which will stop the server when closed.
channel = rabbitClient
.startRPCServer(
queue='RPC_requests',
consumer=(method,args)=>{
// Whatever code you want here:
if( method == 'processA' ) {
return myService.processA( argumentCollection=args );
} else if( method == 'processB' ) {
return myService.processB( argumentCollection=args );
} else {
throw( "RPC method [#method#] not found." );
}
}
);
When passing a CFC instance, the RPC server will automatically call the method whose name matches the incoming method
parameter, passing all of the args as the argument collection for you.
channel = getRabbitClient()
.startRPCServer(
queue=queueName,
consumer=new models.myService()
);
Once the RPC server is running somewhere, you are ready to create and use an RPC Client which is pointing to the same queue. A different RPC Client will be created and persisted inside the rabbitClient for each unique exchange/routingKey/timeout combination. You can re-use the client object across threads, and each call will block until the server has sent back the response from your remote procedure call.
RPCClient = rabbitClient.RPCClient( 'RPC_requests' );
result RPCClient.processA( 'foo' );
The above code will create an RPC call with the method name processA
and a positional parameter of foo
. The RPCClient uses onMissingMethod()
to allow any method call to be proxied to the server cleanly.
If you are not going to use an RPC Client again during the life of your application, you can close it like so.
RPCClient.$close();
The close method is prefixed with a dollar sign on this class so it doesn't interfere with the onMissingMethod()
operation.
It is important that your blocking RPC calls time out if the server is not running or not responding in a timely fashion. The default timeout of the client is 15 seconds. You can override the default timeout for all RPC calls when creating the client lie so:
RPCClient = rabbitClient.RPCClient( 'RPC_requests', 30 );
And you can override the timeout on a per-call basis by using the $call()
method like so:
resullt = RPCClient.$call( 'processA', { my : 'args' }, 30 )
Both of those examples override the timeout to 30 seconds. If no reply has been received, an exception will be thrown. It is not possible to pass a call-specific timeout override when using the onMissingMethod()
proxy.
If an exception is thrown from the RPC Server's UDF or component method during invocation, an exception with the same message, detail, and type will be thrown from the client.
Due to limitations of both Lucee and Adobe in regards to how they rethrow serialized exception objects, the original stack trace will be in the detail
and the tag context will not be the
original one from the server-side exception.
Initial development for this module sponsored by Avoya Travel.
Copyright Since 2005 ColdBox Framework by Luis Majano and Ortus Solutions, Corp www.ortussolutions.com
Because of His grace, this project exists. If you don't like this, then don't read it, its not for you.
"Therefore being justified by faith, we have peace with God through our Lord Jesus Christ: By whom also we have access by faith into this grace wherein we stand, and rejoice in hope of the glory of God. And not only so, but we glory in tribulations also: knowing that tribulation worketh patience; And patience, experience; and experience, hope: And hope maketh not ashamed; because the love of God is shed abroad in our hearts by the Holy Ghost which is given unto us. ." Romans 5:5
"I am the way, and the truth, and the life; no one comes to the Father, but by me (JESUS)" Jn 14:1-12
$
box install rabbitsdk