Aggregate EIP

The Aggregator from the EIP patterns allows you to combine a number of messages together into a single message.

image

A correlation Expression is used to determine the messages which should be aggregated together. If you want to aggregate all messages into a single message, just use a constant expression. An AggregationStrategy is used to combine all the message exchanges for a single correlation key into a single message exchange.

Aggregator options

The Aggregate EIP supports 27 options which are listed below:

Name Description Default Type

correlationExpression

Required The expression used to calculate the correlation key to use for aggregation. The Exchange which has the same correlation key is aggregated together. If the correlation key could not be evaluated an Exception is thrown. You can disable this by using the ignoreBadCorrelationKeys option.

ExpressionSubElementDefinition

completionPredicate

A Predicate to indicate when an aggregated exchange is complete. If this is not specified and the AggregationStrategy object implements Predicate, the aggregationStrategy object will be used as the completionPredicate.

ExpressionSubElementDefinition

completionTimeoutExpression

Time in millis that an aggregated exchange should be inactive before its complete (timeout). This option can be set as either a fixed value or using an Expression which allows you to evaluate a timeout dynamically - will use Long as result. If both are set Camel will fallback to use the fixed value if the Expression result was null or 0. You cannot use this option together with completionInterval, only one of the two can be used. By default the timeout checker runs every second, you can use the completionTimeoutCheckerInterval option to configure how frequently to run the checker. The timeout is an approximation and there is no guarantee that the a timeout is triggered exactly after the timeout value. It is not recommended to use very low timeout values or checker intervals.

ExpressionSubElementDefinition

completionSizeExpression

Number of messages aggregated before the aggregation is complete. This option can be set as either a fixed value or using an Expression which allows you to evaluate a size dynamically - will use Integer as result. If both are set Camel will fallback to use the fixed value if the Expression result was null or 0.

ExpressionSubElementDefinition

optimisticLockRetryPolicy

Allows to configure retry settings when using optimistic locking.

OptimisticLockRetryPolicyDefinition

parallelProcessing

When aggregated are completed they are being send out of the aggregator. This option indicates whether or not Camel should use a thread pool with multiple threads for concurrency. If no custom thread pool has been specified then Camel creates a default pool with 10 concurrent threads.

false

Boolean

optimisticLocking

Turns on using optimistic locking, which requires the aggregationRepository being used, is supporting this by implementing org.apache.camel.spi.OptimisticLockingAggregationRepository .

false

Boolean

executorServiceRef

If using parallelProcessing you can specify a custom thread pool to be used. In fact also if you are not using parallelProcessing this custom thread pool is used to send out aggregated exchanges as well.

String

timeoutCheckerExecutorService Ref

If using either of the completionTimeout, completionTimeoutExpression, or completionInterval options a background thread is created to check for the completion for every aggregator. Set this option to provide a custom thread pool to be used rather than creating a new thread for every aggregator.

String

aggregationRepositoryRef

Sets the custom aggregate repository to use. Will by default use org.apache.camel.processor.aggregate.MemoryAggregationRepository

String

strategyRef

A reference to lookup the AggregationStrategy in the Registry. Configuring an AggregationStrategy is required, and is used to merge the incoming Exchange with the existing already merged exchanges. At first call the oldExchange parameter is null. On subsequent invocations the oldExchange contains the merged exchanges and newExchange is of course the new incoming Exchange.

String

strategyMethodName

This option can be used to explicit declare the method name to use, when using POJOs as the AggregationStrategy.

String

strategyMethodAllowNull

If this option is false then the aggregate method is not used for the very first aggregation. If this option is true then null values is used as the oldExchange (at the very first aggregation), when using POJOs as the AggregationStrategy.

false

Boolean

completionSize

Number of messages aggregated before the aggregation is complete. This option can be set as either a fixed value or using an Expression which allows you to evaluate a size dynamically - will use Integer as result. If both are set Camel will fallback to use the fixed value if the Expression result was null or 0.

Integer

completionInterval

A repeating period in millis by which the aggregator will complete all current aggregated exchanges. Camel has a background task which is triggered every period. You cannot use this option together with completionTimeout, only one of them can be used.

Long

completionTimeout

Time in millis that an aggregated exchange should be inactive before its complete (timeout). This option can be set as either a fixed value or using an Expression which allows you to evaluate a timeout dynamically - will use Long as result. If both are set Camel will fallback to use the fixed value if the Expression result was null or 0. You cannot use this option together with completionInterval, only one of the two can be used. By default the timeout checker runs every second, you can use the completionTimeoutCheckerInterval option to configure how frequently to run the checker. The timeout is an approximation and there is no guarantee that the a timeout is triggered exactly after the timeout value. It is not recommended to use very low timeout values or checker intervals.

Long

completionTimeoutChecker Interval

Interval in millis that is used by the background task that checks for timeouts ( org.apache.camel.TimeoutMap ). By default the timeout checker runs every second. The timeout is an approximation and there is no guarantee that the a timeout is triggered exactly after the timeout value. It is not recommended to use very low timeout values or checker intervals.

1000

Long

completionFromBatchConsumer

Enables the batch completion mode where we aggregate from a org.apache.camel.BatchConsumer and aggregate the total number of exchanges the org.apache.camel.BatchConsumer has reported as total by checking the exchange property org.apache.camel.Exchange#BATCH_COMPLETE when its complete. This option cannot be used together with discardOnAggregationFailure.

false

Boolean

completionOnNewCorrelation Group

Enables completion on all previous groups when a new incoming correlation group. This can for example be used to complete groups with same correlation keys when they are in consecutive order. Notice when this is enabled then only 1 correlation group can be in progress as when a new correlation group starts, then the previous groups is forced completed.

false

Boolean

eagerCheckCompletion

Use eager completion checking which means that the completionPredicate will use the incoming Exchange. As opposed to without eager completion checking the completionPredicate will use the aggregated Exchange.

false

Boolean

ignoreInvalidCorrelationKeys

If a correlation key cannot be successfully evaluated it will be ignored by logging a DEBUG and then just ignore the incoming Exchange.

false

Boolean

closeCorrelationKeyOn Completion

Closes a correlation key when its complete. Any late received exchanges which has a correlation key that has been closed, it will be defined and a ClosedCorrelationKeyException is thrown.

Integer

discardOnCompletionTimeout

Discards the aggregated message on completion timeout. This means on timeout the aggregated message is dropped and not sent out of the aggregator.

false

Boolean

discardOnAggregationFailure

Discards the aggregated message when aggregation failed (an exception was thrown from AggregationStrategy . This means the partly aggregated message is dropped and not sent out of the aggregator. This option cannot be used together with completionFromBatchConsumer.

false

Boolean

forceCompletionOnStop

Indicates to complete all current aggregated exchanges when the context is stopped

false

Boolean

completeAllOnStop

Indicates to wait to complete all current and partial (pending) aggregated exchanges when the context is stopped. This also means that we will wait for all pending exchanges which are stored in the aggregation repository to complete so the repository is empty before we can stop. You may want to enable this when using the memory based aggregation repository that is memory based only, and do not store data on disk. When this option is enabled, then the aggregator is waiting to complete all those exchanges before its stopped, when stopping CamelContext or the route using it.

false

Boolean

aggregateControllerRef

To use a org.apache.camel.processor.aggregate.AggregateController to allow external sources to control this aggregator.

String

About AggregationStrategy

The AggregationStrategy is used for aggregating the old (lookup by its correlation id) and the new exchanges together into a single exchange. Possible implementations include performing some kind of combining or delta processing, such as adding line items together into an invoice or just using the newest exchange and removing old exchanges such as for state tracking or market data prices; where old values are of little use.

Notice the aggregation strategy is a mandatory option and must be provided to the aggregator.

In the aggregate method, do not create a new exchange instance to return, instead return either the old or new exchange from the input parameters; favor returning the old exchange whenever possible.

Here are a few example AggregationStrategy implementations that should help you create your own custom strategy.

//simply combines Exchange String body values using '+' as a delimiter
class StringAggregationStrategy implements AggregationStrategy {

    public Exchange aggregate(Exchange oldExchange, Exchange newExchange) {
        if (oldExchange == null) {
            return newExchange;
        }

        String oldBody = oldExchange.getIn().getBody(String.class);
        String newBody = newExchange.getIn().getBody(String.class);
        oldExchange.getIn().setBody(oldBody + "+" + newBody);
        return oldExchange;
    }
}

//simply combines Exchange body values into an ArrayList<Object>
class ArrayListAggregationStrategy implements AggregationStrategy {

    public Exchange aggregate(Exchange oldExchange, Exchange newExchange) {
        Object newBody = newExchange.getIn().getBody();
        ArrayList<Object> list = null;
        if (oldExchange == null) {
            list = new ArrayList<Object>();
            list.add(newBody);
            newExchange.getIn().setBody(list);
            return newExchange;
        } else {
            list = oldExchange.getIn().getBody(ArrayList.class);
            list.add(newBody);
            return oldExchange;
        }
    }
}

About completion

When aggregation Exchanges at some point you need to indicate that the aggregated exchanges is complete, so they can be send out of the aggregator. Camel allows you to indicate completion in various ways as follows:

  • completionTimeout - Is an inactivity timeout in which is triggered if no new exchanges have been aggregated for that particular correlation key within the period.

  • completionInterval - Once every X period all the current aggregated exchanges are completed.

  • completionSize - Is a number indicating that after X aggregated exchanges it’s complete.

  • completionPredicate - Runs a Predicate when a new exchange is aggregated to determine if we are complete or not. The configured aggregationStrategy can implement the Predicate interface and will be used as the completionPredicate if no completionPredicate is configured. The configured aggregationStrategy can implement PreCompletionAwareAggregationStrategy and will be used as the completionPredicate in pre-complete check mode. See further below for more details.

  • completionFromBatchConsumer - Special option for Batch Consumer which allows you to complete when all the messages from the batch has been aggregated.

  • forceCompletionOnStop - Indicates to complete all current aggregated exchanges when the context is stopped

  • Using a AggregateController - which allows to use an external source to complete groups or all groups. This can be done using Java or JMX API.

Notice that all the completion ways are per correlation key. And you can combine them in any way you like. It’s basically the first which triggers that wins. So you can use a completion size together with a completion timeout. Only completionTimeout and completionInterval cannot be used at the same time.

Notice the completion is a mandatory option and must be provided to the aggregator. If not provided Camel will thrown an Exception on startup.

Pre-completion mode

There can be use-cases where you want the incoming Exchange to determine if the correlation group should pre-complete, and then the incoming Exchange is starting a new group from scratch. To determine this the AggregationStrategy can implement PreCompletionAwareAggregationStrategy which has a preComplete method:

    /**
     * Determines if the aggregation should complete the current group, and start a new group, or the aggregation
     * should continue using the current group.
     *
     * @param oldExchange the oldest exchange (is <tt>null</tt> on first aggregation as we only have the new exchange)
     * @param newExchange the newest exchange (can be <tt>null</tt> if there was no data possible to acquire)
     * @return <tt>true</tt> to complete current group and start a new group, or <tt>false</tt> to keep using current
     */
    boolean preComplete(Exchange oldExchange, Exchange newExchange);

If the preComplete method returns true, then the existing groups is completed (without aggregating the incoming exchange (newExchange). And then the newExchange is used to start the correlation group from scratch so the group would contain only that new incoming exchange. This is known as pre-completion mode. And when the aggregation is in pre-completion mode, then only the following completions are in use

  • aggregationStrategy must implement PreCompletionAwareAggregationStrategy xxx

  • completionTimeout or completionInterval can also be used as fallback completions

  • any other completion are not used (such as by size, from batch consumer etc)

  • eagerCheckCompletion is implied as true, but the option has no effect

Persistent AggregationRepository

The aggregator provides a pluggable repository which you can implement your own org.apache.camel.spi.AggregationRepository.
If you need persistent repository then you can use either Camel LevelDB, or SQL Component components.

Using TimeoutAwareAggregationStrategy

If your aggregation strategy implements TimeoutAwareAggregationStrategy, then Camel will invoke the timeout method when the timeout occurs. Notice that the values for index and total parameters will be -1, and the timeout parameter will be provided only if configured as a fixed value. You must not throw any exceptions from the timeout method.

Using CompletionAwareAggregationStrategy

If your aggregation strategy implements CompletionAwareAggregationStrategy, then Camel will invoke the onComplete method when the aggregated Exchange is completed. This allows you to do any last minute custom logic such as to cleanup some resources, or additional work on the exchange as it’s now completed.
You must not throw any exceptions from the onCompletion method.

Completing current group decided from the AggregationStrategy

The AggregationStrategy can now included a property on the returned Exchange that contains a boolean to indicate if the current group should be completed. This allows to overrule any existing completion predicates / sizes / timeouts etc, and complete the group.

For example the following logic (from an unit test) will complete the group if the message body size is larger than 5. This is done by setting the exchange property Exchange.AGGREGATION_COMPLETE_CURRENT_GROUP to true.

    public final class MyCompletionStrategy implements AggregationStrategy {
        @Override
        public Exchange aggregate(Exchange oldExchange, Exchange newExchange) {
            if (oldExchange == null) {
                return newExchange;
            }
            String body = oldExchange.getIn().getBody(String.class) + "+"
                + newExchange.getIn().getBody(String.class);
            oldExchange.getIn().setBody(body);
            if (body.length() >= 5) {
                oldExchange.setProperty(Exchange.AGGREGATION_COMPLETE_CURRENT_GROUP, true);
            }
            return oldExchange;
        }
    }

Completing all previous group decided from the AggregationStrategy

The AggregationStrategy can now included a property on the returned Exchange that contains a boolean to indicate if all previous groups should be completed. This allows to overrule any existing completion predicates / sizes / timeouts etc, and complete all the existing previous group.

For example the following logic (from an unit test) will complete all the previous group when a new aggregation group is started. This is done by setting the property Exchange.AGGREGATION_COMPLETE_ALL_GROUPS to true.

    public final class MyCompletionStrategy implements AggregationStrategy {
        @Override
        public Exchange aggregate(Exchange oldExchange, Exchange newExchange) {
            if (oldExchange == null) {
                // we start a new correlation group, so complete all previous groups
                newExchange.setProperty(Exchange.AGGREGATION_COMPLETE_ALL_GROUPS, true);
                return newExchange;
            }

            String body1 = oldExchange.getIn().getBody(String.class);
            String body2 = newExchange.getIn().getBody(String.class);

            oldExchange.getIn().setBody(body1 + body2);
            return oldExchange;
        }
    }

Manually Force the Completion of All Aggregated Exchanges Immediately

You can manually trigger completion of all current aggregated exchanges by sending an exchange containing the exchange property Exchange.AGGREGATION_COMPLETE_ALL_GROUPS set to true. The message is considered a signal message only, the message headers/contents will not be processed otherwise.

You can alternatively set the exchange property Exchange.AGGREGATION_COMPLETE_ALL_GROUPS_INCLUSIVE to true to trigger completion of all groups after processing the current message.

Using a List<V> in AggregationStrategy

If you want to aggregate some value from the messages <V> into a List<V> then we have added a org.apache.camel.processor.aggregate.AbstractListAggregationStrategy abstract class that makes this easier. The completed Exchange that is sent out of the aggregator will contain the List<V> in the message body.

For example to aggregate a List<Integer> you can extend this class as shown below, and implement the getValue method:

Using AggregateController

The org.apache.camel.processor.aggregate.AggregateController allows you to control the aggregate at runtime using Java or JMX API. This can be used to force completing groups of exchanges, or query its current runtime statistics.

The aggregator provides a default implementation if no custom have been configured, which can be accessed using getAggregateController() method. Though it may be easier to configure a controller in the route using aggregateController as shown below:

private AggregateController controller = new DefaultAggregateController();

from("direct:start")
   .aggregate(header("id"), new MyAggregationStrategy())
      .completionSize(10).id("myAggregator")
      .aggregateController(controller)
      .to("mock:aggregated");

Then there is API on AggregateController to force completion. For example to complete a group with key foo

int groups = controller.forceCompletionOfGroup("foo");

The number return would be the number of groups completed. In this case it would be 1 if the foo group existed and was completed. If foo does not exists then 0 is returned.

There is also an api to complete all groups

int groups = controller.forceCompletionOfAllGroups();

To configure this from XML DSL

<bean id="myController" class="org.apache.camel.processor.aggregate.DefaultAggregateController"/>

  <camelContext xmlns="http://camel.apache.org/schema/spring">
        <route>
            <from uri="direct:start"/>
            <aggregate strategyRef="myAppender" completionSize="10"
                       aggregateControllerRef="myController">
                <correlationExpression>
                    <header>id</header>
                </correlationExpression>
                <to uri="mock:result"/>
            </aggregate>
        </route>
    </camelContext>

There is also JMX API on the aggregator which is available under the processors node in the Camel JMX tree.

Using GroupedExchanges

In the route below we group all the exchanges together using groupExchanges():

from("direct:start")
    // aggregate all using same expression
    .aggregate(constant(true))
    // wait for 0.5 seconds to aggregate
    .completionTimeout(500L)
    // group the exchanges so we get one single exchange containing all the others
    .groupExchanges()
    .to("mock:result");

As a result we have one outgoing Exchange being routed the "mock:result" endpoint. The exchange is a holder containing all the incoming Exchanges.

The output of the aggregator will then contain the exchanges grouped together in a list as shown below:

List<Exchange> grouped = exchange.getIn().getBody(List.class);

Using POJOs as AggregationStrategy

To use the AggregationStrategy you had to implement the org.apache.camel.AggregationStrategy interface, which means your logic would be tied to the Camel API. You can use a POJO for the logic and let Camel adapt to your POJO. To use a POJO a convention must be followed:

  • there must be a public method to use

  • the method must not be void

  • the method can be static or non-static

  • the method must have 2 or more parameters

  • the parameters is paired so the first 50% is applied to the oldExchange and the reminder 50% is for the newExchange

  • .. meaning that there must be an equal number of parameters, eg 2, 4, 6 etc.

The paired methods is expected to be ordered as follows:

  • the first parameter is the message body

  • the 2nd parameter is a Map of the headers

  • the 3rd parameter is a Map of the Exchange properties

This convention is best explained with some examples.

In the method below, we have only 2 parameters, so the 1st parameter is the body of the oldExchange, and the 2nd is paired to the body of the newExchange:

public String append(String existing, String next) {
  return existing + next;
}

In the method below, we have only 4 parameters, so the 1st parameter is the body of the oldExchange, and the 2nd is the Map of the oldExchange headers, and the 3rd is paired to the body of the newExchange, and the 4th parameter is the Map of the newExchange headers:

public String append(String existing, Map existingHeaders, String next, Map nextHeaders) {
  return existing + next;
}

And finally if we have 6 parameters the we also have the properties of the Exchanges:

public String append(String existing, Map existingHeaders, Map existingProperties,
                     String next, Map nextHeaders, Map nextProperties) {
  return existing + next;
}

To use this with the Aggregate EIP we can use a POJO with the aggregate logic as follows:

public class MyBodyAppender {

    public String append(String existing, String next) {
        return next + existing;
    }

}

And then in the Camel route we create an instance of our bean, and then refer to the bean in the route using bean method from org.apache.camel.builder.AggregationStrategies as shown:

private MyBodyAppender appender = new MyBodyAppender();

public void configure() throws Exception {
    from("direct:start")
        .aggregate(constant(true), AggregationStrategies.bean(appender, "append"))
            .completionSize(3)
            .to("mock:result");
}

We can also provide the bean type directly:

public void configure() throws Exception {
    from("direct:start")
        .aggregate(constant(true), AggregationStrategies.bean(MyBodyAppender.class, "append"))
            .completionSize(3)
            .to("mock:result");
}

And if the bean has only one method we do not need to specify the name of the method:

public void configure() throws Exception {
    from("direct:start")
        .aggregate(constant(true), AggregationStrategies.bean(MyBodyAppender.class))
            .completionSize(3)
            .to("mock:result");
}

And the append method could be static:

public class MyBodyAppender {

    public static String append(String existing, String next) {
        return next + existing;
    }

}

If you are using XML DSL then we need to declare a <bean> with the POJO:

<bean id="myAppender" class="com.foo.MyBodyAppender"/>

And in the Camel route we use strategyRef to refer to the bean by its id, and the strategyMethodName can be used to define the method name to call:

<camelContext xmlns="http://camel.apache.org/schema/spring">
    <route>
        <from uri="direct:start"/>
        <aggregate strategyRef="myAppender" strategyMethodName="append" completionSize="3">
            <correlationExpression>
                <constant>true</constant>
            </correlationExpression>
            <to uri="mock:result"/>
        </aggregate>
    </route>
</camelContext>

When using XML DSL you must define the POJO as a <bean>.

Aggregating when no data

By default when using POJOs as AggregationStrategy, then the method is only invoked when there is data to be aggregated (by default). You can use the option strategyMethodAllowNull to configure this. Where as without using POJOs then you may have null as oldExchange or newExchange parameters. For example the Aggregate EIP will invoke the AggregationStrategy with oldExchange as null, for the first Exchange incoming to the aggregator. And then for subsequent Exchanges then oldExchange and newExchange parameters are both not null.

Example with Content Enricher EIP and no data

Though with POJOs as AggregationStrategy we made this simpler and only call the method when oldExchange and newExchange is not null, as that would be the most common use-case. If you need to allow oldExchange or newExchange to be null, then you can configure this with the POJO using the AggregationStrategyBeanAdapter as shown below. On the bean adapter we call setAllowNullNewExchange to allow the new exchange to be null.

public void configure() throws Exception {
    AggregationStrategyBeanAdapter myStrategy = new AggregationStrategyBeanAdapter(appender, "append");
    myStrategy.setAllowNullOldExchange(true);
    myStrategy.setAllowNullNewExchange(true);

    from("direct:start")
        .pollEnrich("seda:foo", 1000, myStrategy)
            .to("mock:result");
}

This can be configured a bit easier using the beanAllowNull method from AggregationStrategies as shown:

public void configure() throws Exception {
    from("direct:start")
        .pollEnrich("seda:foo", 1000, AggregationStrategies.beanAllowNull(appender, "append"))
            .to("mock:result");
}

Then the append method in the POJO would need to deal with the situation that newExchange can be null:

public class MyBodyAppender {

    public String append(String existing, String next) {
        if (next == null) {
            return "NewWasNull" + existing;
        } else {
            return existing + next;
        }
    }

}

In the example above we use the Content Enricher EIP using pollEnrich. The newExchange will be null in the situation we could not get any data from the "seda:foo" endpoint, and therefore the timeout was hit after 1 second. So if we need to do some special merge logic we would need to set setAllowNullNewExchange=true, so the append method will be invoked. If we do not do that then when the timeout was hit, then the append method would normally not be invoked, meaning the Content Enricher did not merge/change the message.

In XML DSL you would configure the strategyMethodAllowNull option and set it to true as shown below:

<camelContext xmlns="http://camel.apache.org/schema/spring">
    <route>
        <from uri="direct:start"/>
        <aggregate strategyRef="myAppender"
                   strategyMethodName="append"
                   strategyMethodAllowNull="true"
                   completionSize="3">
            <correlationExpression>
                <constant>true</constant>
            </correlationExpression>
            <to uri="mock:result"/>
        </aggregate>
    </route>
</camelContext>

Different body types

When for example using strategyMethodAllowNull as true, then the parameter types of the message bodies does not have to be the same. For example suppose we want to aggregate from a com.foo.User type to a List<String> that contains the user name. We could code a POJO doing this as follows:

public static final class MyUserAppender {

    public List addUsers(List names, User user) {
        if (names == null) {
            names = new ArrayList();
        }
        names.add(user.getName());
        return names;
    }
}

Notice that the return type is a List which we want to contain the user names. The 1st parameter is the list of names, and then notice the 2nd parameter is the incoming com.foo.User type.