I am trying to fetch a message with a particular correlation id like explained in rabbitmq docs. However I see that the irrelevant messages gets dequeued. I do not want it to happen. How can I tell rabbitmq to not dequeue after I get message and get to know that this is not the one I was looking for. Please help me.
`
.
.
replyQueueName = channel.queueDeclare().getQueue();
consumer = new QueueingConsumer(channel);
channel.basicConsume(replyQueueName, false, consumer);
while (true) {
QueueingConsumer.Delivery delivery = consumer.nextDelivery();
System.out.println(delivery.getProperties().getCorrelationId());
if (delivery.getProperties().getCorrelationId().equals(corrId)) {
response = new String(delivery.getBody());
break;
}
}
`
You can't do what you want, the way you want. The "selective consumer" is an anti-pattern in RabbitMQ.
Instead, you should design your RabbitMQ setup so that your messages are routed to a queue that only contains messages for the intended consumer.
I wrote more about this, here: http://derickbailey.com/2015/07/22/airport-baggage-claims-selective-consumers-and-rabbitmq-anti-patterns/
If you can afford to lose the order of messages you can use the re-queueing mechanism.
Try turning off auto ack.
If not, you have to redesign your application to inject headers or routing keys to route to a particular queue.
Related
I'm trying to retrieve messages from a queue. I understand that RecieveMessageRequest has a threshold of 10 messages but when I tried I was able to receive only 2 out 3 messages in the queue. I read many threads which said adding setMaxNumberOfMessages(10) and increasing WaitTimeSeconds will fix it(Before adding this I received only one message out of 3) but it wasn't helpful.
FYI: I'm using a standard queue and all the messages were definitely there in the queue at the time of receive message request so it shouldn't have been a polling issue.
My implementation:
List<Message> messages;
ReceiveMessageRequest receiveMessageRequest = new ReceiveMessageRequest().withQueueUrl(queueUrl)
.withWaitTimeSeconds(10)
.withMaxNumberOfMessages(10);
messages = sqsConfig.getSQSClient().receiveMessage(receiveMessageRequest).getMessages();
I know it make not make a ton of sense but in multiple programming languages (certainly Java and Python) I've had to loop to get everything from a queue. In Java it's something like (using the V2 api's):
// note that "done" needs to be set somewhere else to stop the loop
while (!done) {
ReceiveMessageRequest receiveMessageRequest = ReceiveMessageRequest.builder()
.queueUrl(queueUrl)
.waitTimeSeconds(20)
.build();
List<Message> messages = sqsClient.receiveMessage(receiveMessageRequest).messages();
for (Message nextMessage : messages) {
// do something with the message
DeleteMessageRequest deleteMessageRequest = DeleteMessageRequest
.builder()
.queueUrl(queueUrl)
.receiptHandle(nextMessage.receiptHandle())
.build();
sqsClient.deleteMessage(deleteMessageRequest);
}
}
From what I've read it is because there are ultimately many servers behind SQS and a single call doesn't hit all of them - it takes multiple calls. This is not a busy loop as the waitTimeSeconds does block if there is nothing to do.
Try something like this to read SQS. It's not quite as elegant but it does work.
Is there a way we can push messages to RabbitMQ and have an expiry time for it and once it expires, it should provide a notification.
Or
Is there a way we can deliver the messages in RabbitMQ after a certain amount of time. For example, I want to push a message in the queue and wants it to get delivered after 10 seconds..and simultaneously next messages.
Regarding the first part of your question, the routing of messages that have expired due to a per-message TTL is a feature of the RabbitMQ dead letter exchange (DLX).
Regarding a delay, this is not something supported by RabbitMQ out of the box, nor in my opinion should it be a feature of a message broker. I can't imagine a legitimate use case where you would deliberately want to introduce a delay into a message queue. In fact, it is a design goal of any message broker to minimize delay with enqueued messages. If you find a delay to be appropriate, then it is also likely that a message queue is not the appropriate means of conveyance.
The RabbitMQ Delayed Message Plugin adds a new exchange type to RabbitMQ where messages routed by that exchange can be delayed if the users chooses to do so.
You can use it in a way like described below.
// ... elided code ...
Map<String, Object> args = new HashMap<String, Object>();
args.put("x-delayed-type", "direct");
channel.exchangeDeclare("my-exchange", "x-delayed-message", true, false, args);
// ... more code ...
String queueA = "rabbitmq://host:5672/queue-a.exchange?queue=queue-a.exchange..etc
from(queueA)
.routeId("idForQueueA")
.onException(Exception.class)
.maximumRedeliveries(0)
// .processRef("sendEmailAlert") * not sure this belongs here*
.to(deadLetterQueueA)
.useOriginalMessage()
.end()
.processRef("dataProcessing")
.processRef("dataExporting")
.end();
Explaining the code above:
Messages are taken from queueA. Upon various processes being successful the message is consumed. If it fails its added to the dead letter queue "deadLetterQueueA". This all works ok.
My question is
When messages arrive in the deadletter queue I want to add alerts so we know to do something about it... How could I to add an email alert when a message arrives in the dead letter queue. I dont want to lose the original message if the alert fails - nor do I want the alert to consume the message.
My thoughts are.. I would need to split the message on an exception so its sent to two different queues? One for the alert which then sends out an email alert and then consumes itself. Then one for the dead letter queue that just sites there? However I'm not sure how to do this?
You can split a message to go to multiple endpoints using a multicast (details here):
.useOriginalMessage().multicast().to(deadLetterQueueA, "smtp://username#host:port?options")
This uses the camel mail component endpoints described here. Alternatively, you can continue processing the message after the to. So something like:
.useOriginalMessage()
.to(deadLetterQueueA)
.transform().simple("Hi <name>, there has been an error on the object ${body.toString}")
.to("smtp://username#host:port?options")
If you had multiple recipients, you could use a recipients list
public class EmailListBean {
#RecipientList
public String[] emails() {
return new String[] {"smtp://joe#host:port?options",
"smtp://fred#host:port?options"};
}
}
.useOriginalMessage()
.to(deadLetterQueueA)
.transform().simple("...")
.bean(EmailListBean.class)
Be careful of using JMS queues to store messages while waiting for a human to action them. I don't know what sort of message traffic you're getting. I'm assuming if you want to send an email for every failure, it's not a lot. But I would normally be wary of this sort of thing, and chose to use logging or database persistence to store the results of errors, and only use a JMS error queue to notify other processes or consumers of the error or to schedule a re-try.
There are two ways you can do this , but based on your message volume you might not want to send email on every failed message.
You can use the solution provided by AndyN , or you can use the Advisory Topics ActiveMQ.Advisory.MessageDLQd.Queue.* , whenever a message gets in to the DLQ the enqueue count of the topic will increase by 1 . By monitoring the Queue Depth you might now be able to send a mail to based on the number of the errors that ocurred.
If you want to do it at the producer end. You can use any one of the solutions provided by AndyN
RabbitMQ's Channel#basicConsume method gives us the following arguments:
channel.basicConsume(queueName, autoAck, consumerTag, noLocal,
exclusive, arguments, callback);
Giving us the ability to tell RabbitMQ exactly which queue we want to consume from.
But Channel#basicPublish has no such equivalency:
channel.basicPublish(exchangeName, routingKey, mandatory, immediateFlag,
basicProperties, messageAsBytes);
Why can't I specify the queue to publish to here?!? How do I get a Channel publishing to, say, a queue named logging? Thanks in advance!
To expand on #Tien Nguyen's answer, there is a "cheat" in RabbitMQ that effectively lets you publish directly to a queue. Each queue is automatically bound to the AMQP default exchange, with the queue's name as the routing key. The default exchange is also known as the "nameless exchange" - ie its name is the empty string. So if you publish to the exchange named "" with routing key equal to your queue's name, the message will go to just that queue. It is going through an exchange as #John said, it's just not one that you need to declare or bind yourself.
I don't have the Java client handy to try this code, but it should work.
channel.basicPublish("", myQueueName, false, false, null, myMessageAsBytes);
That said, this is mostly contrary to the spirit of how RabbitMQ works. For normal application flow you should declare and bind exchanges. But for exceptional cases the "cheat" can be useful. For example, I believe this is how the Rabbit Admin Console allows you to manually publish messages to a queue without all the ceremony of creating and binding exchanges.
Basically queues can be binded to an exchange based on routingKeys.
Assume that you have 3 different publishers.
Publisher1 sending message to exchange with routingKey "events"
Publisher2 sending message to exchange with routingKey "tasks"
Publisher3 sending message to exchange with routingKey "jobs"
You can have a consumer that consumes only messages with specific routhingKey.
For example in order to have a consumer for "events" messages you declare like this
channel.queueBind(queueName, exchangeName, "events");
If you want to consume all the messages coming to the exchange you give the routing as '#'
So in short what i can say is,
1. Messages will be published to an exchange.
2. Queues will be bound to exchange based on routingKeys.
3. RabbitMQ will forward messages with matching routing keys to the corresponding queues.
Please see the tutorial - http://www.rabbitmq.com/tutorials/tutorial-three-java.html
The core idea in the messaging model in RabbitMQ is that the producer never sends any messages directly to a queue. Actually, quite often the producer doesn't even know if a message will be delivered to any queue at all. Instead, the producer can only send messages to an exchange
please try this:
channel.basicPublish("", yourQueueName, null,
message.getBytes((Charset.forName("UTF-8"))));
It worked for my project.
Is it possible to send message to particular receiver using JMS Queue(HornetQ)?
Among so many receivers, I want certain message to be received by receiver which
are running on Linux OS.
Every suggestion is appriciated.
Thanks.
You can set a message property using Message.setObjectProperty(String, Object) and then have your consumers select the messages they are interested in using Session.createConsumer(Destination, String)
Sender example:
Message message = session.createMessage();
message.setObjectProperty("OS", "LINUX");
producer.send(message);
Receiver example:
MessageConsumer consumer = session.createConsumer(destination, "OS = 'LINUX'");
//Use consumer to receive messages.
The receiver in the example will ignore (they will go to some other receiver) all messages that do not match the selector. In this case all message where the 'OS' property is not 'LINUX' will be ignored by this consumer.
You can set properties of JMS message: http://download.oracle.com/javaee/1.4/api/javax/jms/TextMessage.html and filter messages at client side.
For example,
message.setStringProperty("TARGET_OS", "LINUX") - at sender
http://www.mkyong.com/java/how-to-detect-os-in-java-systemgetpropertyosname/ - detect OS at receivers and filter messages with correct TARGET_OS property
You can use JMS selectors on the consumer side to look for messages that fit specific criteria.
Not sure if I am missing something, you could keep things simple by having multiple queues - specific to each platform, then the linux based consumers can listen to the linux specific queue alone. Now your challenge probably will be to route the messages to the appropriate queue from the producer side, that should be fairly easy if the routing is based on some attribute of the message?