The JMS component allows messages to be sent to (or consumed from) a JMS Queue or Topic. The implementation of the JMS Component uses Spring's JMS support for declarative transactions, using Spring's
JmsTemplate for sending and a
MessageListenerContainer for consuming.
Maven users will need to add the following dependency to their
pom.xml for this component:
destinationName is a JMS queue or topic name. By default, the
destinationName is interpreted as a queue name. For example, to connect to the queue,
You can include the optional
queue: prefix, if you prefer:
To connect to a topic, you must include the
topic: prefix. For example, to
connect to the topic,
You append query options to the URI using the following format,
The JMS component reuses Spring 2's
JmsTemplate for sending messages. This is not ideal for use in a non-J2EE container and typically requires some caching in the JMS provider to avoid poor performance.
If you intend to use Apache ActiveMQ as your Message Broker - which is a good choice as ActiveMQ rocks , then we recommend that you either:
- Use the ActiveMQ component, which is already optimized to use ActiveMQ efficiently
- Use the
PoolingConnectionFactory in ActiveMQ.
Transactions and Cache Levels
If you are consuming messages and using transactions (
transacted=true) then the default settings for cache level can impact performance.
If you are using XA transactions then you cannot cache as it can cause the XA transaction to not work properly.
If you are not using XA, then you should consider caching as it speeds up performance, such as setting
Through Camel 2.7.x, the default setting for
CACHE_CONSUMER. You will need to explicitly set
In Camel 2.8 onwards, the default setting for
CACHE_AUTO. This default auto detects the mode and sets the cache level accordingly to:
- CACHE_CONSUMER = if transacted=false
- CACHE_NONE = if transacted=true
So you can say the default setting is conservative. Consider using
cacheLevelName=CACHE_CONSUMER if you are using non-XA transactions.
If you wish to use durable topic subscriptions, you need to specify both clientId and durableSubscriptionName. The value of the
clientId must be unique and can only be used by a single JMS connection instance in your entire network. You may prefer to use Virtual Topics instead to avoid this limitation. More background on durable messaging here.
When using message headers, the JMS specification states that header names must be valid Java identifiers. So try to name your headers to be valid Java identifiers. One benefit of doing this is that you can then use your headers inside a JMS Selector (whose SQL92 syntax mandates Java identifier syntax for headers).
A simple strategy for mapping header names is used by default. The strategy is to replace any dots and hyphens in the header name as shown below and to reverse the replacement when the header name is restored from a JMS message sent over the wire. What does this mean? No more losing method names to invoke on a bean component, no more losing the filename header for the File Component, and so on.
The current header name strategy for accepting header names in Camel is as follows:
- Dots are replaced by
_DOT_ and the replacement is reversed when Camel consume the message
- Hyphen is replaced by
_HYPHEN_ and the replacement is reversed when Camel consumes the message
You can configure many different properties on the JMS endpoint which map to properties on the JMSConfiguration POJO.
The options are divided into two tables, the first one with the most common options used. The latter contains the rest.
Most commonly used options
Sets the JMS client ID to use. Note that this value, if specified, must be unique and can only be used by a single JMS connection instance. It is typically only required for durable topic subscriptions. You may prefer to use Virtual Topics instead.
Specifies the default number of concurrent consumers. From Camel 2.10.3 onwards this option can also be used when doing request/reply over JMS. See also the
maxMessagesPerTask option to control dynamic scaling up/down of threads.
true, a producer will behave like a InOnly exchange with the exception that
JMSReplyTo header is sent out and not be suppressed like in the case of
InOnly the producer will not wait for a reply. A consumer with this flag will behave like
InOnly. This feature can be used to bridge
InOut requests to another queue so that a route on the other queue will send it´s response directly back to the original
The durable subscriber name for specifying durable topic subscriptions. The
clientId option must be configured as well.
Specifies the maximum number of concurrent consumers. From Camel 2.10.3 onwards this option can also be used when doing request/reply over JMS. See also the
maxMessagesPerTask option to control dynamic scaling up/down of threads.
The number of messages per task. -1 is unlimited. If you use a range for concurrent consumers (eg min < max), then this option can be used to set a value to eg
100 to control how fast the consumers will shrink when less work is required.
true, if you want to send message using the QoS settings specified on the message, instead of the QoS settings on the JMS endpoint. The following three headers are considered
JMSExpiration. You can provide all or only some of them. If not provided, Camel will fall back to use the values from the endpoint instead. So, when using this option, the headers override the values from the endpoint. The
explicitQosEnabled option, by contrast, will only use options set on the endpoint, and not values from the message header.
Provides an explicit ReplyTo destination, which overrides any incoming value of
Message.getJMSReplyTo(). If you do Request Reply over JMS then make sure to read the section Request-reply over JMS further below for more details, and the
replyToType option as well.
Camel 2.9: Allows for explicitly specifying which kind of strategy to use for replyTo queues when doing request/reply over JMS. Possible values are:
Exclusive. By default Camel will use temporary queues. However if
replyTo has been configured, then
Shared is used by default. This option allows you to use exclusive queues instead of shared ones. See further below for more details, and especially the notes about the implications if running in a clustered environment, and the fact that
Shared reply queues has lower performance than its alternatives
Producer only: The timeout for waiting for a reply when using the InOut Exchange Pattern (in milliseconds). The default is 20 seconds. From Camel 2.13/2.12.3 onwards you can include the header
"CamelJmsRequestTimeout" to override this endpoint configured timeout value, and thus have per message individual timeout values. See below in section About time to live for more details. See also the requestTimeoutCheckerInterval option.
Sets the JMS Selector, which is an SQL 92 predicate that is used to filter messages within the broker. You may have to encode special characters such as = as %3D Before Camel 2.3.0, we don't support this option in CamelConsumerTemplate
When sending messages, specifies the time-to-live of the message (in milliseconds). See below in section About time to live for more details.
Specifies whether to use transacted mode for sending/receiving messages using the InOnly Exchange Pattern.
Camel 2.1: Specifies whether to test the connection on startup. This ensures that when Camel starts that all the JMS consumers have a valid connection to the JMS broker. If a connection cannot be granted then Camel throws an exception on startup. This ensures that Camel is not started with failed connections. From Camel 2.8 onwards also the JMS producers is tested as well.
All the other options
Specifies whether the consumer accept messages while it is stopping. You may consider enabling this option, if you start and stop JMS routes at runtime, while there are still messages enqued on the queue. If this option is
false, and you stop the JMS route, then messages may be rejected, and the JMS broker would have to attempt redeliveries, which yet again may be rejected, and eventually the message may be moved at a dead letter queue on the JMS broker. To avoid this its recommended to enable this option.
The JMS acknowledgement name, which is one of:
The JMS acknowledgement mode defined as an Integer. Allows you to set vendor-specific extensions to the acknowledgment mode. For the regular modes, it is preferable to use the
Camel 2.9.3/2.10.1: Whether to allow sending messages with no body. If this option is
false and the message body is null, then an
JMSException is thrown.
true, Camel will always make a JMS message copy of the message when it is passed to the producer for sending. Copying the message is needed in some situations, such as when a
replyToDestinationSelectorName is set (incidentally, Camel will set the
alwaysCopyMessage option to
true, if a
replyToDestinationSelectorName is set)
Camel 2.9: Whether the
JmsConsumer processes the Exchange asynchronously. If enabled then the
JmsConsumer may pickup the next message from the JMS queue, while the previous message is being processed asynchronously (by the Asynchronous Routing Engine). This means that messages may be processed not 100% strictly in order. If disabled (as default) then the Exchange is fully processed before the
JmsConsumer will pickup the next message from the JMS queue. Note if
transacted has been enabled, then
asyncConsumer=true does not run asynchronously, as transactions must be executed synchronously (Camel 3.0 may support async transactions).
Camel 2.10: Whether to startup the
JmsConsumer message listener asynchronously, when starting a route. For example if a
JmsConsumer cannot get a connection to a remote JMS broker, then it may block while retrying and/or failover. This will cause Camel to block while starting routes. By setting this option to
true, you will let routes startup, while the
JmsConsumer connects to the JMS broker using a dedicated thread in asynchronous mode. If this option is used, then beware that if the connection could not be established, then an exception is logged at
WARN level, and the consumer will not be able to receive messages; You can then restart the route to retry.
Camel 2.10: Whether to stop the
JmsConsumer message listener asynchronously, when stopping a route.
Specifies whether the consumer container should auto-startup.
CACHE_AUTO (Camel >= 2.8.0)
CACHE_CONSUMER (Camel <= 2.7.1)
Sets the cache level by name for the underlying JMS resources. Possible values are:
CACHE_SESSION. The default setting for Camel 2.8 and newer is
CACHE_AUTO. For Camel 2.7.1 and older the default is
CACHE_CONSUMER. See the Spring documentation and Transactions Cache Levels for more information.
Sets the cache level by ID for the underlying JMS resources. See
cacheLevelName option for more details.
The consumer type to use, which can be one of:
Custom. The consumer type determines which Spring JMS listener to use.
Default will use
Simple will use
Custom is specified, the
MessageListenerContainerFactory defined by the
messageListenerContainerFactoryRef option will determine what
org.springframework.jms.listener.AbstractMessageListenerContainer to use (new option in Camel 2.10.2 onwards). This option was temporary removed in Camel 2.7 and 2.8. But has been added back from Camel 2.9 onwards.
The default JMS connection factory to use for the
templateConnectionFactory, if neither is specified.
Camel 2.10.4: Specifies what default TaskExecutor type to use in the DefaultMessageListenerContainer, for both consumer endpoints and the ReplyTo consumer of producer endpoints. Possible values:
SimpleAsync (uses Spring's SimpleAsyncTaskExecutor) or
ThreadPool (uses Spring's ThreadPoolTaskExecutor with optimal values - cached threadpool-like). If not set, it defaults to the previous behaviour, which uses a cached thread pool for consumer endpoints and SimpleAsync for reply consumers. The use of
ThreadPool is recommended to reduce "thread trash" in elastic configurations with dynamically increasing and decreasing concurrent consumers.
Camel 2.12.2/2.13: Specifies the delivery mode to be used. Possibles values are those defined by
Specifies whether persistent delivery is used by default.
Specifies the JMS Destination object to use on this endpoint.
Specifies the JMS destination name to use on this endpoint.
org.springframework.jms.support.destination.DestinationResolver that allows you to use your own resolver (for example, to lookup the real destination in a JNDI registry).
Camel 2.8: Use this option to force disabling time to live. For example when you do request/reply over JMS, then Camel will by default use the
requestTimeout value as time to live on the message being sent. The problem is that the sender and receiver systems have to have their clocks synchronized, so they are in sync. This is not always so easy to archive. So you can use
disableTimeToLive=true to not set a time to live value on the sent message. Then the message will not expire on the receiver system. See below in section About time to live for more details.
Enables eager loading of JMS properties as soon as a message is received, which is generally inefficient, because the JMS properties might not be required. But this feature can sometimes catch early any issues with the underlying JMS provider and the use of JMS properties. This feature can also be used for testing purposes, to ensure JMS properties can be understood and handled correctly.
Specifies the JMS Exception Listener that is to be notified of any underlying JMS exceptions.
Camel 2.8.2, 2.9: Specifies a
org.springframework.util.ErrorHandler to be invoked in case of any uncaught exceptions thrown while processing a
Message. By default these exceptions will be logged at the WARN level, if no
errorHandler has been configured. From Camel 2.9.1: onwards you can configure logging level and whether stack traces should be logged using the below two options. This makes it much easier to configure, than having to code a custom
Camel 2.9.1: Allows to configure the default
errorHandler logging level for logging uncaught exceptions.
Camel 2.9.1: Allows to control whether stacktraces should be logged or not, by the default
Set if the
timeToLive qualities of service should be used when sending messages. This option is based on Spring's
timeToLive options are applied to the current endpoint. This contrasts with the
preserveMessageQos option, which operates at message granularity, reading QoS properties exclusively from the Camel In message headers.
Specifies whether the listener session should be exposed when consuming messages.
Camel 2.7: When using
mapJmsMessage=false Camel will create a new JMS message to send to a new JMS destination if you touch the headers (get or set) during the route. Set this option to
true to force Camel to send the original JMS message that was received.
Specifies the limit for idle executions of a receive task, not having received any message within its execution. If this limit is reached, the task will shut down and leave receiving to other executing tasks (in the case of dynamic scheduling; see the
maxConcurrentConsumers setting). There is additional doc available from Spring.
Camel 2.8.2, 2.9: Specify the limit for the number of consumers that are allowed to be idle at any given time.
Camel 2.10.3: Only applicable when sending to JMS destination using InOnly (eg fire and forget). Enabling this option will enrich the Camel Exchange with the actual JMSMessageID that was used by the JMS client when the message was sent to the JMS destination.
Camel 2.11.2/2.12: Whether to include all JMSXxxx properties when mapping from JMS to Camel Message. Setting this to
true will include properties such as
JMSXUserID etc. Note: If you are using a custom
headerFilterStrategy then this option does not apply.
Allows you to force the use of a specific
javax.jms.Message implementation for sending JMS messages. Possible values are:
Text. By default, Camel would determine which JMS message type to use from the In body type. This option allows you to specify it.
Pluggable strategy for encoding and decoding JMS keys so they can be compliant with the JMS specification. Camel provides two implementations out of the box:
default strategy will safely marshal dots and hyphens (
passthrough strategy leaves the key as is. Can be used for JMS brokers which do not care whether JMS header keys contain illegal characters. You can provide your own implementation of the
org.apache.camel.component.jms.JmsKeyFormatStrategy and refer to it using the
Allows you to use your own implementation of the
org.springframework.jms.core.JmsOperations interface. Camel uses
JmsTemplate as default. Can be used for testing purpose, but not used much as stated in the spring API docs.
true, Camel will create a
JmsTransactionManager, if there is no
transactionManager injected when option
The JMS connection factory used for consuming messages.
Specifies whether Camel should auto map the received JMS message to an appropiate payload type, such as
javax.jms.TextMessage to a
String etc. See section about how mapping works below for more details.
Limits the number of messages fetched at most, when browsing endpoints using Browse or JMX API.
To use a custom Spring
org.springframework.jms.support.converter.MessageConverter so you can be 100% in control how to map to/from a
When sending, specifies whether message IDs should be added.
Camel 2.10.2: Registry ID of the
MessageListenerContainerFactory used to determine what
org.springframework.jms.listener.AbstractMessageListenerContainer to use to consume messages. Setting this will automatically set
Specifies whether timestamps should be enabled by default on sending messages.
The password for the connector factory.
Values greater than 1 specify the message priority when sending (where 0 is the lowest priority and 9 is the highest). The
explicitQosEnabled option must also be enabled in order for this option to have any effect.
Specifies whether to inhibit the delivery of messages published by its own connection.
The timeout for receiving messages (in milliseconds).
Specifies the interval between recovery attempts, i.e. when a connection is being refreshed, in milliseconds. The default is 5000 ms, that is, 5 seconds.
Camel 2.9.1: Sets the cache level by name for the reply consumer when doing request/reply over JMS. This option only applies when using fixed reply queues (not temporary). Camel will by default use:
CACHE_CONSUMER for exclusive or shared w/
CACHE_SESSION for shared without
replyToSelectorName. Some JMS brokers such as IBM WebSphere may require to set the
replyToCacheLevelName=CACHE_NONE to work. Note: If using temporary queues then
CACHE_NONE is not allowed, and you must use a higher value such as
Sets the JMS Selector using the fixed name to be used so you can filter out your own replies from the others when using a shared queue (that is, if you are not using a temporary reply queue).
Specifies whether to use persistent delivery by default for replies.
Camel 2.9.2: Configures how often Camel should check for timed out Exchanges when doing request/reply over JMS.By default Camel checks once per second. But if you must react faster when a timeout occurs, then you can lower this interval, to check more frequently. The timeout is determined by the option requestTimeout.
@deprecated: Enabled by default, if you specify a
durableSubscriberName and a
Allows you to specify a custom task executor for consuming messages.
Camel 2.6: To use when using Spring 2.x with Camel. Allows you to specify a custom task executor for consuming messages.
The JMS connection factory used for sending messages.
@deprecated: Specifies whether to use transacted mode for sending messages using the InOut Exchange Pattern. Applies only to producer endpoints. See section Enabling Transacted Consumption for more details.
The Spring transaction manager to use.
The name of the transaction to use.
The timeout value of the transaction (in seconds), if using transacted mode.
If enabled and you are using Request Reply messaging (InOut) and an Exchange failed on the consumer side, then the caused
Exception will be send back in response as a
javax.jms.ObjectMessage. If the client is Camel, the returned
Exception is rethrown. This allows you to use Camel JMS as a bridge in your routing - for example, using persistent queues to enable robust routing. Notice that if you also have transferExchange enabled, this option takes precedence. The caught exception is required to be serializable. The original
Exception on the consumer side can be wrapped in an outer exception such as
org.apache.camel.RuntimeCamelException when returned to the producer.
You can transfer the exchange over the wire instead of just the body and headers. The following fields are transferred: In body, Out body, Fault body, In headers, Out headers, Fault headers, exchange properties, exchange exception. This requires that the objects are serializable. Camel will exclude any non-serializable objects and log it at
WARN level. You must enable this option on both the producer and consumer side, so Camel knows the payloads is an Exchange and not a regular payload.
The username for the connector factory.
JMSMessageID should always be used as
JMSCorrelationID for InOut messages.
@deprecated (removed from Camel 2.5 onwards): Specifies whether the old JMS API should be used.
Message Mapping between JMS and Camel
Camel automatically maps messages between
When sending a JMS message, Camel converts the message body to the following JMS message types:
The DOM will be converted to
When receiving a JMS message, Camel converts the JMS message to the following body type:
Disabling auto-mapping of JMS messages
You can use the
mapJmsMessage option to disable the auto-mapping above. If disabled, Camel will not try to map the received JMS message, but instead uses it directly as the payload. This allows you to avoid the overhead of mapping and let Camel just pass through the JMS message. For instance, it even allows you to route
javax.jms.ObjectMessage JMS messages with classes you do not have on the classpath.
Using a custom MessageConverter
You can use the
messageConverter option to do the mapping yourself in a Spring
For example, in the route below we use a custom message converter when sending a message to the JMS order queue:
You can also use a custom message converter when consuming from a JMS destination.
Controlling the mapping strategy selected
You can use the jmsMessageType option on the endpoint URL to force a specific message type for all messages.
In the route below, we poll files from a folder and send them as
javax.jms.TextMessage as we have forced the JMS producer endpoint to use text messages:
You can also specify the message type to use for each messabe by setting the header with the key
CamelJmsMessageType. For example:
The possible values are defined in the
Message format when sending
The exchange that is sent over the JMS wire must conform to the JMS Message spec.
exchange.in.header the following rules apply for the header keys:
- Keys starting with
JMSX are reserved.
exchange.in.headers keys must be literals and all be valid Java identifiers (do not use dots in the key name).
- Camel replaces dots & hyphens and the reverse when when consuming JMS messages:
. is replaced by
_DOT_ and the reverse replacement when Camel consumes the message.
- is replaced by
_HYPHEN_ and the reverse replacement when Camel consumes the message.
- See also the option
jmsKeyFormatStrategy, which allows use of your own custom strategy for formatting keys.
exchange.in.header, the following rules apply for the header values:
- The values must be primitives or their counter objects (such as
Character). The types,
BigInteger are all converted to their
toString() representation. All other types are dropped.
Camel will log with category
org.apache.camel.component.jms.JmsBinding at DEBUG level if it drops a given header value. For example:
Message format when receiving
Camel adds the following properties to the
Exchange when it receives a message:
The reply destination.
Camel adds the following JMS properties to the In message headers when it receives a JMS message:
The JMS correlation ID.
The JMS delivery mode.
The JMS destination.
The JMS expiration.
The JMS unique message ID.
The JMS priority (with 0 as the lowest priority and 9 as the highest).
Is the JMS message redelivered.
The JMS reply-to destination.
The JMS timestamp.
The JMS type.
The JMS group ID.
As all the above information is standard JMS you can check the JMS documentation for further details.
About using Camel to send and receive messages and JMSReplyTo
The JMS component is complex and you have to pay close attention to how it works in some cases. So this is a short summary of some of the areas/pitfalls to look for.
When Camel sends a message using its
JMSProducer, it checks the following conditions:
- The message exchange pattern,
- Whether a
JMSReplyTo was set in the endpoint or in the message headers,
- Whether any of the following options have been set on the JMS endpoint:
All this can be a tad complex to understand and configure to support your use case.
JmsProducer behaves as follows, depending on configuration:
Camel will expect a reply, set a temporary
JMSReplyTo, and after sending the message, it will start to listen for the reply message on the temporary queue.
JMSReplyTo is set
Camel will expect a reply and, after sending the message, it will start to listen for the reply message on the specified
Camel will send the message and not expect a reply.
JMSReplyTo is set
By default, Camel discards the
JMSReplyTo destination and clears the
JMSReplyTo header before sending the message. Camel then sends the message and does not expect a reply. Camel logs this in the log at
WARN level (changed to
DEBUG level from Camel 2.6 onwards. You can use
preserveMessageQuo=true to instruct Camel to keep the
JMSReplyTo. In all situations the
JmsProducer does not expect any reply and thus continue after sending the message.
JmsConsumer behaves as follows, depending on configuration:
Camel will send the reply back to the
Camel will not send a reply back, as the pattern is InOnly.
This option suppresses replies.
So pay attention to the message exchange pattern set on your exchanges.
If you send a message to a JMS destination in the middle of your route you can specify the exchange pattern to use, see more at Request Reply.
This is useful if you want to send an
InOnly message to a JMS topic:
Reuse endpoint and send to different destinations computed at runtime
If you need to send messages to a lot of different JMS destinations, it makes sense to reuse a JMS endpoint and specify the real destination in a message header. This allows Camel to reuse the same endpoint, but send to different destinations. This greatly reduces the number of endpoints created and economizes on memory and thread resources.
You can specify the destination in the following headers:
A destination object.
The destination name.
For example, the following route shows how you can compute a destination at run time and use it to override the destination appearing in the JMS URL:
The queue name,
dummy, is just a placeholder. It must be provided as part of the JMS endpoint URL, but it will be ignored in this example.
computeDestination bean, specify the real destination by setting the
CamelJmsDestinationName header as follows:
Then Camel will read this header and use it as the destination instead of the one configured on the endpoint. So, in this example Camel sends the message to
activemq:queue:order:2, assuming the
id value was 2.
If both the
CamelJmsDestination and the
CamelJmsDestinationName headers are set,
CamelJmsDestination takes priority.
Configuring different JMS providers
You can configure your JMS provider in Spring XML as follows:
Basically, you can configure as many JMS component instances as you wish and give them a unique name using the
id attribute. The preceding example configures an
activemq component. You could do the same to configure MQSeries, TibCo, BEA, Sonic and so on.
Once you have a named JMS component, you can then refer to endpoints within that component using URIs. For example for the component name,
activemq, you can then refer to destinations using the URI format,
activemq:[queue:|topic:]destinationName. You can use the same approach for all other JMS providers.
This works by the SpringCamelContext lazily fetching components from the spring context for the scheme name you use for Endpoint URIs and having the Component resolve the endpoint URIs.
Using JNDI to find the ConnectionFactory
If you are using a J2EE container, you might need to look up JNDI to find the JMS
ConnectionFactory rather than use the usual
<bean> mechanism in Spring. You can do this using Spring's factory bean or the new Spring XML namespace. For example:
See The jee schema in the Spring reference documentation for more details about JNDI lookup.
A common requirement with JMS is to consume messages concurrently in multiple threads in order to make an application more responsive. You can set the
concurrentConsumers option to specify the number of threads servicing the JMS endpoint, as follows:
You can configure this option in one of the following ways:
- On the
- On the endpoint URI or,
- By invoking
setConcurrentConsumers() directly on the
Concurrent Consuming with async consumer
Notice that each concurrent consumer will only pickup the next available message from the JMS broker, when the current message has been fully processed. You can set the option
asyncConsumer=true to let the consumer pickup the next message from the JMS queue, while the previous message is being processed asynchronously (by the Asynchronous Routing Engine). See more details in the table on top of the page about the
Request-reply over JMS
Camel supports Request Reply over JMS. In essence the MEP of the Exchange should be
InOut when you send a message to a JMS queue.
Camel offers a number of options to configure request/reply over JMS that influence performance and clustered environments. The table below summaries the options.
A temporary queue is used as reply queue, and automatic created by Camel. To use this do not specify a replyTo queue name. And you can optionally configure
replyToType=Temporary to make it stand out that temporary queues are in use.
A shared persistent queue is used as reply queue. The queue must be created beforehand, although some brokers can create them on the fly such as Apache ActiveMQ. To use this you must specify the replyTo queue name. And you can optionally configure
replyToType=Shared to make it stand out that shared queues are in use. A shared queue can be used in a clustered environment with multiple nodes running this Camel application at the same time. All using the same shared reply queue. This is possible because JMS Message selectors are used to correlate expected reply messages; this impacts performance though. JMS Message selectors is slower, and therefore not as fast as
Exclusive queues. See further below how to tweak this for better performance.
An exclusive persistent queue is used as reply queue. The queue must be created beforehand, although some brokers can create them on the fly such as Apache ActiveMQ. To use this you must specify the replyTo queue name. And you must configure
replyToType=Exclusive to instruct Camel to use exclusive queues, as
Shared is used by default, if a
replyTo queue name was configured. When using exclusive reply queues, then JMS Message selectors are not in use, and therefore other applications must not use this queue as well. An exclusive queue cannot be used in a clustered environment with multiple nodes running this Camel application at the same time; as we do not have control if the reply queue comes back to the same node that sent the request message; that is why shared queues use JMS Message selectors to make sure of this. Though if you configure each Exclusive reply queue with an unique name per node, then you can run this in a clustered environment. As then the reply message will be sent back to that queue for the given node, that awaits the reply message.
Camel 2.10.3: Allows to process reply messages concurrently using concurrent message listeners in use. You can specify a range using the
maxConcurrentConsumers options. Notice: That using
Shared reply queues may not work as well with concurrent listeners, so use this option with care.
Camel 2.10.3: Allows to process reply messages concurrently using concurrent message listeners in use. You can specify a range using the
maxConcurrentConsumers options. Notice: That using
Shared reply queues may not work as well with concurrent listeners, so use this option with care.
JmsProducer detects the
InOut and provides a
JMSReplyTo header with the reply destination to be used. By default Camel uses a temporary queue, but you can use the
replyTo option on the endpoint to specify a fixed reply queue (see more below about fixed reply queue).
Camel will automatic setup a consumer which listen on the reply queue, so you should not do anything.
This consumer is a Spring
DefaultMessageListenerContainer which listen for replies. However it's fixed to 1 concurrent consumer.
That means replies will be processed in sequence as there are only 1 thread to process the replies. If you want to process replies faster, then we need to use concurrency. But not using the
concurrentConsumer option. We should use the
threads from the Camel DSL instead, as shown in the route below:
In this route we instruct Camel to route replies asynchronously using a thread pool with 5 threads.
From Camel 2.10.3 onwards you can now configure the listener to use concurrent threads using the
maxConcurrentConsumers options. This allows you to easier configure this in Camel as shown below:
Request-reply over JMS and using a shared fixed reply queue
If you use a fixed reply queue when doing Request Reply over JMS as shown in the example below, then pay attention.
In this example the fixed reply queue named "bar" is used. By default Camel assumes the queue is shared when using fixed reply queues, and therefore it uses a
JMSSelector to only pickup the expected reply messages (eg based on the
JMSCorrelationID). See next section for exclusive fixed reply queues. That means its not as fast as temporary queues. You can speedup how often Camel will pull for reply messages using the
receiveTimeout option. By default its 1000 millis. So to make it faster you can set it to 250 millis to pull 4 times per second as shown:
Notice this will cause the Camel to send pull requests to the message broker more frequent, and thus require more network traffic.
It is generally recommended to use temporary queues if possible.
Request-reply over JMS and using an exclusive fixed reply queue
Available as of Camel 2.9
In the previous example, Camel would anticipate the fixed reply queue named "bar" was shared, and thus it uses a
JMSSelector to only consume reply messages which it expects. However there is a drawback doing this as JMS selectos is slower. Also the consumer on the reply queue is slower to update with new JMS selector ids. In fact it only updates when the
receiveTimeout option times out, which by default is 1 second. So in theory the reply messages could take up till about 1 sec to be detected. On the other hand if the fixed reply queue is exclusive to the Camel reply consumer, then we can avoid using the JMS selectors, and thus be more performant. In fact as fast as using temporary queues. So in Camel 2.9 onwards we introduced the
ReplyToType option which you can configure to
to tell Camel that the reply queue is exclusive as shown in the example below:
Mind that the queue must be exclusive to each and every endpoint. So if you have two routes, then they each need an unique reply queue as shown in the next example:
The same applies if you run in a clustered environment. Then each node in the cluster must use an unique reply queue name. As otherwise each node in the cluster may pickup messages which was intended as a reply on another node. For clustered environments its recommended to use shared reply queues instead.
Synchronizing clocks between senders and receivers
When doing messaging between systems, its desirable that the systems have synchronized clocks. For example when sending a JMS message, then you can set a time to live value on the message. Then the receiver can inspect this value, and determine if the message is already expired, and thus drop the message instead of consume and process it. However this requires that both sender and receiver have synchronized clocks. If you are using ActiveMQ then you can use the timestamp plugin to synchronize clocks.
About time to live
Read first above about synchronized clocks.
When you do request/reply (InOut) over JMS with Camel then Camel uses a timeout on the sender side, which is default 20 seconds from the
requestTimeout option. You can control this by setting a higher/lower value. However the time to live value is still set on the JMS message being send. So that requires the clocks to be synchronized between the systems. If they are not, then you may want to disable the time to live value being set. This is now possible using the
disableTimeToLive option from Camel 2.8 onwards. So if you set this option to
disableTimeToLive=true, then Camel does not set any time to live value when sending JMS messages. But the request timeout is still active. So for example if you do request/reply over JMS and have disabled time to live, then Camel will still use a timeout by 20 seconds (the
requestTimeout option). That option can of course also be configured. So the two options
disableTimeToLive gives you fine grained control when doing request/reply.
From Camel 2.13/2.12.3 onwards you can provide a header in the message to override and use as the request timeout value instead of the endpoint configured value. For example:
In the route above we have a endpoint configured
requestTimeout of 30 seconds. So Camel will wait up till 30 seconds for that reply message to come back on the bar queue. If no reply message is received then a
org.apache.camel.ExchangeTimedOutException is set on the Exchange and Camel continues routing the message, which would then fail due the exception, and Camel's error handler reacts.
If you want to use a per message timeout value, you can set the header with key
org.apache.camel.component.jms.JmsConstants#JMS_REQUEST_TIMEOUT which has constant value
"CamelJmsRequestTimeout" with a timeout value as long type.
For example we can use a bean to compute the timeout value per individual message, such as calling the
"whatIsTheTimeout" method on the service bean as shown below:
When you do fire and forget (InOut) over JMS with Camel then Camel by default does not set any time to live value on the message. You can configure a value by using the
timeToLive option. For example to indicate a 5 sec., you set
timeToLive=5000. The option
disableTimeToLive can be used to force disabling the time to live, also for InOnly messaging. The
requestTimeout option is not being used for InOnly messaging.
Enabling Transacted Consumption
A common requirement is to consume from a queue in a transaction and then process the message using the Camel route. To do this, just ensure that you set the following properties on the component/endpoint:
transacted = true
transactionManager = a Transsaction Manager - typically the
See the Transactional Client EIP pattern for further details.
Available as of Camel 2.10
You can leverage the DMLC transacted session API using the following properties on component/endpoint:
transacted = true
lazyCreateTransactionManager = false
The benefit of doing so is that the cacheLevel setting will be honored when using local transactions without a configured TransactionManager. When a TransactionManager is configured, no caching happens at DMLC level and its necessary to rely on a pooled connection factory. For more details about this kind of setup see here and here.
Using JMSReplyTo for late replies
When using Camel as a JMS listener, it sets an Exchange property with the value of the ReplyTo
javax.jms.Destination object, having the key
ReplyTo. You can obtain this
Destination as follows:
And then later use it to send a reply using regular JMS or Camel.
A different solution to sending a reply is to provide the
replyDestination object in the same Exchange property when sending. Camel will then pick up this property and use it for the real destination. The endpoint URI must include a dummy destination, however. For example:
Using a request timeout
In the sample below we send a Request Reply style message Exchange (we use the
requestBody method =
InOut) to the slow queue for further processing in Camel and we wait for a return reply:
JMS is used in many examples for other components as well. But we provide a few samples below to get started.
Receiving from JMS
In the following sample we configure a route that receives JMS messages and routes the message to a POJO:
You can of course use any of the EIP patterns so the route can be context based. For example, here's how to filter an order topic for the big spenders:
Sending to a JMS
In the sample below we poll a file folder and send the file content to a JMS topic. As we want the content of the file as a
TextMessage instead of a
BytesMessage, we need to convert the body to a
Camel also has annotations so you can use POJO Consuming and POJO Producing.
Spring DSL sample
The preceding examples use the Java DSL. Camel also supports Spring XML DSL. Here is the big spender sample using Spring DSL:
JMS appears in many of the examples for other components and EIP patterns, as well in this Camel documentation. So feel free to browse the documentation. If you have time, check out the this tutorial that uses JMS but focuses on how well Spring Remoting and Camel works together Tutorial-JmsRemoting.
Using JMS as a Dead Letter Queue storing Exchange
Normally, when using JMS as the transport, it only transfers the body and headers as the payload. If you want to use JMS with a Dead Letter Channel, using a JMS queue as the Dead Letter Queue, then normally the caused Exception is not stored in the JMS message. You can, however, use the transferExchange option on the JMS dead letter queue to instruct Camel to store the entire Exchange in the queue as a
javax.jms.ObjectMessage that holds a
org.apache.camel.impl.DefaultExchangeHolder. This allows you to consume from the Dead Letter Queue and retrieve the caused exception from the Exchange property with the key
Exchange.EXCEPTION_CAUGHT. The demo below illustrates this:
Then you can consume from the JMS queue and analyze the problem:
Using JMS as a Dead Letter Channel storing error only
You can use JMS to store the cause error message or to store a custom body, which you can initialize yourself. The following example uses the Message Translator EIP to do a transformation on the failed exchange before it is moved to the JMS dead letter queue:
Here we only store the original cause error message in the transform. You can, however, use any Expression to send whatever you like. For example, you can invoke a method on a Bean or use a custom processor.
Sending an InOnly message and keeping the JMSReplyTo header
When sending to a JMS destination using camel-jms the producer will use the MEP to detect if its InOnly or InOut messaging. However there can be times where you want to send an InOnly message but keeping the JMSReplyTo header. To do so you have to instruct Camel to keep it, otherwise the JMSReplyTo header will be dropped.
For example to send an InOnly message to the foo queue, but with a JMSReplyTo with bar queue you can do as follows:
Notice we use
preserveMessageQos=true to instruct Camel to keep the JMSReplyTo header.
Setting JMS provider options on the destination
Some JMS providers, like IBM's WebSphere MQ need options to be set on the JMS destination. For example, you may need to specify the targetClient option. Since targetClient is a WebSphere MQ option and not a Camel URI option, you need to set that on the JMS destination name like so:
Some versions of WMQ won't accept this option on the destination name and you will get an exception like:
com.ibm.msg.client.jms.DetailedJMSException: JMSCC0005: The specified value 'MY_QUEUE?targetClient=1' is not allowed for 'XMSC_DESTINATION_NAME'
A workaround is to use a custom DestinationResolver: