问题来源

实际项目中，我们会遇到kafka消费不及时，系统发现最新的数据一致无法出现，这时候通过其他kafka工具发现原来指定的消费组lag太大，也就是我们系统要么出问题，要么需要启动更多的实例加快消费消息。每次通过kafka工具去查询lag基本都是手工而且耗时且慢，能否在自己的系统中集成查询指定消费组lag的功能，然后出现问题是可以管理界面中迅速查看lag呢？答案是当然可以。 kafka提供了这些API。

备注：这里的kafka工具是指kafka自带的命令行工具，或者其他第三方提供的kafka工具。

实现思路

首先我们需要引入的依赖如下。 其中kafka-clients是作为kafka客户端访问kakfa需要的依赖包，kafka_2.12是管理端需要的依赖包。
备注：我的kafka版本是kafka_2.12-1.0.0。

        <dependency>
            <groupId>org.apache.kafka</groupId>
            <artifactId>kafka_2.12</artifactId>
            <version>1.0.0</version>
        </dependency>

        <dependency>
            <groupId>org.apache.kafka</groupId>
            <artifactId>kafka-clients</artifactId>
            <version>1.0.0</version>
        </dependency>

kafka管理端的AdminClient可以查询消费组的详情。

AdminClient.ConsumerGroupSummary  consumerGroupSummary =  adminClient.describeConsumerGroup(GROUP_ID, 5000);

scala.collection.immutable.Map<TopicPartition, Object> maps =  adminClient.listGroupOffsets(GROUP_ID);

listGroupOffsets可以查询消费组的在指定topic的指定分区中的offset. 然后我们可以通过查询该TopicPartition最新的postion得到endOffset。
两者之差就是指定消费组在指定topic的某个分区上的lag信息。

 KafkaConsumer<String, String> consumer = getNewConsumer();
        consumer.assign(Arrays.asList(topicPartition));
        consumer.seekToEnd(Arrays.asList(topicPartition));
        long endOffset = consumer.position(topicPartition);

具体代码

完整的代码在这里，欢迎加星和fork。 谢谢！

package com.yq;

import kafka.admin.AdminClient;
import lombok.extern.slf4j.Slf4j;
import org.apache.kafka.clients.CommonClientConfigs;
import org.apache.kafka.clients.consumer.KafkaConsumer;
import org.apache.kafka.common.TopicPartition;
import scala.Option;
import scala.collection.immutable.List;

import java.util.Arrays;
import java.util.Properties;


/**
 * Simple to Introduction
 * className: SendMessageMain
 *
 * @author EricYang
 * @version 2019/01/10 11:30
 */
@Slf4j
public class AdminMain {

    private static final String SERVERS = "ubuntu:9092";
    private static final String GROUP_ID = "yq-consumer09";

    public static long getLogEndOffset(TopicPartition topicPartition){
        KafkaConsumer<String, String> consumer = getNewConsumer();
        consumer.assign(Arrays.asList(topicPartition));
        consumer.seekToEnd(Arrays.asList(topicPartition));
        long endOffset = consumer.position(topicPartition);
        consumer.close();
        return endOffset;
    }

    public static KafkaConsumer getNewConsumer(){
        Properties props = new Properties();
        props.put("bootstrap.servers", SERVERS);
        props.put("group.id", "test001");
        props.put("enable.auto.commit", "true");
        props.put("auto.offset.reset", "earliest");
        props.put("auto.commit.interval.ms", "1000");
        props.put("key.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");
        props.put("value.deserializer", "org.apache.kafka.common.serialization.StringDeserializer");
        KafkaConsumer<String, String> consumer = new KafkaConsumer<>(props);
        return consumer;
    }

    public static void main(String[] args) throws InstantiationException, IllegalAccessException {
        Properties props = new Properties();
        props.put(CommonClientConfigs.BOOTSTRAP_SERVERS_CONFIG, SERVERS);
        AdminClient adminClient = AdminClient.create(props);

        AdminClient.ConsumerGroupSummary  consumerGroupSummary =  adminClient.describeConsumerGroup(GROUP_ID, 5000);
        if(consumerGroupSummary.state().equals("Empty")){
            System.out.println("No grp summary");
        }
        System.out.println("consumerGrpSummary State " +  consumerGroupSummary.state());
        Option<List<AdminClient.ConsumerSummary>> consumerSummaryOption =  consumerGroupSummary.consumers();
        

        scala.collection.immutable.Map<TopicPartition, Object> maps =  adminClient.listGroupOffsets(GROUP_ID);
        scala.collection.Set<TopicPartition> topicPartitions = maps.keySet();
        scala.collection.immutable.List<TopicPartition> topicPartitionList = topicPartitions.reversed();
        for(int j =0; j< topicPartitionList.size(); j++){
            TopicPartition topicPartition = topicPartitionList.apply(j);

            String currentOffset = maps.get(topicPartition).get().toString();
            long groupLastEndOffset = getLogEndOffset(topicPartition);
            long lag =  groupLastEndOffset -Long.valueOf(currentOffset);
            System.out.println("topic："+topicPartition.topic()+",  partition:" + topicPartition.partition() + ", offset："
                    + currentOffset + ", groupLastEndOffset："+ groupLastEndOffset + ", lag:"+ lag);
        }
        
        adminClient.close();
    }
}

效果截图

kakfa-consumer-lag.png