Apache Camel - Aggregator Hello World

Introduction

Camel Aggregator提供開發人員能夠將大量訊息合成一個的功能。假設你的系統會接受外來通知，並將內容寫進資料庫中；一次連線寫一筆資料會比較快，還是透過一次連線寫入五筆資料快呢? 正常來說，一次連線寫入五筆同屬性資料會是比較快的。它與Throttler的最大區別在於，Throttler限制了client的請求，而Aggregator則是收集起來一次“後送”。因此，如果使用Aggregator，你的程式必需要有對應的處理方式。

我將透過HTTP GET請求/events/{id}做為範例，說明如何使用Aggregator去將單位時間內的請求，以{id}去分組並Aggregate成各別分組訊息。本篇文章中使用到兩個RouteBuilder，分別為RestRouteBuilder與AggregatorGroupRouteBuilder。RestRouteBuilder負責REST核心相關設定，可參考先前文章。接下來直接說明AggregatorGroupRouteBuilder。

(程式碼可參考link)

AggregatorGroupRouteBuilder

package org.tonylin.practice.camel.aggregator;
 
import static com.google.common.base.Preconditions.checkState;
 
import org.apache.camel.builder.RouteBuilder;
import org.apache.camel.processor.aggregate.GroupedExchangeAggregationStrategy;
 
public class AggregatorGroupRouteBuilder extends RouteBuilder {
 
	private final static String GET_EVENTS = "GET_EVENTS";
 
	private Object eventHandler;
	private int period = 500;
 
	public AggregatorGroupRouteBuilder(Object eventHandler) {
		this.eventHandler = eventHandler;
 
	}
 
	public void setPeriod(int period) {
		this.period = period;
	}
 
	@Override
	public void configure() throws Exception {
		checkState(eventHandler!=null, "Can't find eventHandler");
 
		rest("/events/{id}").get().route().id(GET_EVENTS)
		.aggregate(new GroupedExchangeAggregationStrategy())
		.header("id")
		.completionInterval(period)
		.bean(eventHandler).endRest();
	}
}

以下是在configure中的幾個重點:

aggregate(new GroupedExchangeAggregationStrategy()): 使用GroupedExchangeAggregationStrategy去做aggregate，aggregate的內容為Exchange。如果要加入條件限制，是以Exchange去操作。
header(“id”): 與aggregate一起使用，代表以header id去分組。在這範例中，指得就是event id。
completionInterval(period): 以每period的單位時間去做aggregate。
bean(eventHandler): 請求的處理者，必須要有能力處理aggregate後的訊息。

接下來讓我們透過單元測試展示效果。

Unit Test

測試有兩個目標，testGroupedExchange用以確認aggregate group後的結果，另外一個testCompletionInterval則是確認單位時間設定的作用。以下為程式碼主要結構，主要測試程式碼稍後做說明:

package org.tonylin.practice.camel.aggregator;
 
import static com.google.common.base.Preconditions.checkState;
 
import java.util.Arrays;
import java.util.List;
import java.util.Map;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.TimeUnit;
import java.util.stream.Collectors;
 
import org.apache.camel.Exchange;
import org.apache.camel.Handler;
import org.apache.camel.RoutesBuilder;
import org.apache.camel.test.junit4.CamelTestSupport;
import org.apache.http.HttpResponse;
import org.apache.http.client.HttpClient;
import org.apache.http.client.methods.HttpGet;
import org.apache.http.impl.client.HttpClientBuilder;
import org.junit.Test;
import org.tonylin.practice.camel.rest.RestRouteBuilder;
 
public class AggregatorGroupRouteBuilderTest extends CamelTestSupport  {
 
	private RestHandler hander = new RestHandler();
 
	private HttpClient client = HttpClientBuilder.create().build();
 
	@Override
	protected RoutesBuilder[] createRouteBuilders() throws Exception {
		AggregatorGroupRouteBuilder aggregatorGroupRouteBuilder = new AggregatorGroupRouteBuilder(hander);
		aggregatorGroupRouteBuilder.setPeriod(500);
 
		return new RoutesBuilder[] {
				new RestRouteBuilder(),
				aggregatorGroupRouteBuilder
		};
	}
 
	@Test
	public void testGroupedExchange() throws Exception {
		// skip
	}
	@Test
	public void testCompletionInterval() throws Exception {
		// skip
	}
}

此測試中的RestHandler程式碼如下，它會透過GROUPED_EXCHANGE取得aggregator處理後的內容，並根據header id儲存到map中，供測試程式碼做assertion。除此之外，在這裡使用CountDownLatch了去確認有收到預期訊息數量:

public static class RestHandler {
	private Map<String, List<Exchange>> requestData = new ConcurrentHashMap<String, List<Exchange>>();
	private CountDownLatch countDownLatch;
 
	@SuppressWarnings("unchecked")
	@Handler
	public void handle(Exchange exchange) {
		List<Exchange> groupExchanges = exchange.getProperty(Exchange.GROUPED_EXCHANGE, List.class);
		String id = groupExchanges.get(0).getIn().getHeader("id", String.class);
		requestData.put(id, groupExchanges);
 
		if( countDownLatch != null )
			countDownLatch.countDown();
	}
 
	public Map<String, List<Exchange>> getRequestData(){
		return requestData;
	}
 
	public void expectNum(int num) {
		countDownLatch = new CountDownLatch(num);
	}
 
	public void waitCompletion(int timeout) throws InterruptedException {
		checkState(countDownLatch!=null);
		countDownLatch.await(timeout, TimeUnit.MILLISECONDS);
	}
 
	public void clear() {
		requestData.clear();
	}
}

首先讓我說明testGroupedExchange。測試程式碼中，送了四個訊息，其中包含了三組不同id；這樣的請求，我們預期RestHandler中，總共會收到三組訊息，其中id 123那組，應包含兩個訊息:

private HttpResponse requestWithEventId(String id) {
	try {
		HttpGet httpGet = new HttpGet("http://localhost:8080/events/" + id);
		return client.execute(httpGet);
	} catch( Exception e ) {
		throw new RuntimeException(e);
	}
}
 
@Test
public void testGroupedExchange() throws Exception {
	// given request event id
	List<String> eventIds = Arrays.asList("123", "123", "456", "789");
	hander.expectNum(3);
 
	// when
	List<HttpResponse> responses = eventIds.parallelStream().map(this::requestWithEventId).collect(Collectors.toList());
 
	// then
	responses.forEach(response->{
		assertEquals(200, response.getStatusLine().getStatusCode());
	});
 
	hander.waitCompletion(1000);
 
	Map<String, List<Exchange>> requestData = hander.getRequestData();
	assertEquals(3, requestData.size());
	assertEquals(2, requestData.get("123").size());
	assertEquals(1, requestData.get("456").size());
	assertEquals(1, requestData.get("789").size());
}

從上述程式碼中，可以發現hander.waitCompletion(1000)是放在確認response之後；所以client只要將訊息發送到aggregator後，就會拿到response code，並不會等到全部處理結束才回去。接著是testCompletionInterval。這裡我直接透過請求相同id兩次，並分兩輪送出。這樣做可以確認aggregator有做到根據completionInterval的分批效果:

private void requestTwiceWithId(String id) throws Exception {
	// request event {id} twice
	List<String> eventIds = Arrays.asList(id, id);
	hander.expectNum(1);
	List<HttpResponse> responses = eventIds.parallelStream().map(this::requestWithEventId).collect(Collectors.toList());
	responses.forEach(response->{
		assertEquals(200, response.getStatusLine().getStatusCode());
	});
 
	hander.waitCompletion(1000);
 
	// confirm request result
	Map<String, List<Exchange>> requestData = hander.getRequestData();
	assertEquals(1, requestData.size());
	assertEquals(2, requestData.get("123").size());
 
	// clear first request data
	hander.clear();
	assertTrue(hander.getRequestData().isEmpty());
}
 
@Test
public void testCompletionInterval() throws Exception {
	requestTwiceWithId("123");
	requestTwiceWithId("123");
}

Camel Aggregator其實提供了很多細部操作，你也可以根據自身需求做AggregationStrategy；但時間有限，請容我以後有機會再分享。

Library Info (Gradle Config)

以下是我在寫這篇文章時，所使用的libraries版本:

ext {
	camelVersion='2.23.1'
	nettyAllVersion='4.1.34.Final'
	guavaVersion='27.1-jre'
	log4jVersion='1.2.17'
	slf4jVersion='1.7.26'
	httpClientVersion='4.5.7'
}

dependencies {
    compile group: 'org.apache.camel', name: 'camel-core', version: "$camelVersion"
    compile group: 'org.apache.camel', name: 'camel-netty4-http', version: "$camelVersion"
    compile group: 'org.apache.camel', name: 'camel-http-common', version: "$camelVersion"
    compile group: 'org.apache.camel', name: 'camel-netty4', version: "$camelVersion"
    compile group: 'io.netty', name: 'netty-all', version: "$nettyAllVersion"
    compile group: 'com.google.guava', name: 'guava', version: "$guavaVersion"
    compile group: 'log4j', name: 'log4j', version: "$log4jVersion"
    compile group: 'org.slf4j', name: 'slf4j-api', version: "$slf4jVersion"
    runtime group: 'org.slf4j', name: 'slf4j-log4j12', version: "$slf4jVersion"
    testCompile group: 'org.apache.camel', name: 'camel-test', version: "$camelVersion"
    testCompile group: 'org.apache.httpcomponents', name: 'httpclient', version: "$httpClientVersion"
    testCompile 'junit:junit:4.12'
}

Reference

Camel - Aggregator

Show NIC selection when setting the network command with the device option

Problem 在answer file中設定網卡名稱後，安裝時會停在以下畫面: 所使用的command參數如下: network --onboot = yes --bootproto =dhcp --ipv6 =auto --device =eth1 Diagnostic Result 這樣的參數，以前試驗過是可以安裝完成的。因此在發生這個問題後，我檢查了它的debug console: 從console得知，eth1可能是沒有連接網路線或者是網路太慢而導致的問題。後來和Ivy再三確認，有問題的是有接網路線的網卡，且問題是發生在activate階段: Solution 我想既然有retry應該就有次數或者timeout限制，因此發現在Anaconda的說明文件中( link )，有提到dhcptimeout這個boot參數。看了一些人的使用範例，應該是可以直接串在isolinux.cfg中，如下: default linux ksdevice = link ip =dhcp ks =cdrom: / ks.cfg dhcptimeout = 90 然而我在RHEL/CentOS 6.7與6.8試驗後都無效。因此我就拿了顯示的錯誤字串，問問Google大師，想找一下Anaconda source code來看一下。最後找到別人根據Anaconda code修改的版本: link ，關鍵在於setupIfaceStruct函式中的setupIfaceStruct與readNetConfig: setupIfaceStruct: 會在dhcp時設定dhcptimeout。 readNetConfig: 在writeEnabledNetInfo將timeout寫入dhclient config中；在wait_for_iface_activation內會根據timeout做retry。再來從log與code可以得知，它讀取的檔案是answer file而不是boot command line。因此我接下來的測試，就是在answer file的network command上加入dhcptimeout: network --onboot = yes --bootproto =dhcp --ipv6 =auto --device =eth1 --dhcptimeo

閱讀完整內容

阿兩的筆記本 Ryoutsu's Notebook

搜尋此網誌