kafka消费偏移常见问题

L13763338360

已于 2022-10-09 15:35:08 修改

阅读量5.6k

点赞数

分类专栏： kafka 文章标签： kafka java 分布式

于 2020-04-15 12:52:03 首次发布

本文链接：https://blog.csdn.net/L13763338360/article/details/105529388

版权

kafka 专栏收录该内容

20 篇文章 4 订阅

订阅专栏

kafka消费偏移量维护的方式有两种：

kafka自身维护
zookeeper维护

1.kafka维护消费偏移

1.1 查看kafka维护的消费组列表

./bin/kafka-consumer-groups.sh --new-consumer --bootstrap-server localhost:9092 --list

$ ./bin/kafka-consumer-groups.sh --new-consumer --bootstrap-server localhost:9092 --list
The [new-consumer] option is deprecated and will be removed in a future major release.The new consumer is used by default if the [bootstrap-server] option is provided.
Note: This will not show information about old Zookeeper-based consumers.
creditpayment_db_t_partner_pre_apply_group
fund_trade_center_db_t_oc_task_complete_apply_group
cardloan_user_count_group_test
console-consumer-55818
p2p_db_t_user_info_group_test
p2p_db_t_user_invest_group_test
apply_db_group

……

1.2 查看 kafka维护的某个消费组

./bin/kafka-consumer-groups.sh --new-consumer --bootstrap-server localhost:9092 --group apply_db_group --describe

$ ./bin/kafka-consumer-groups.sh --new-consumer --bootstrap-server localhost:9092 --group apply_db_group --describe
The [new-consumer] option is deprecated and will be removed in a future major release.The new consumer is used by default if the [bootstrap-server] option is provided.
Note: This will not show information about old Zookeeper-based consumers.

TOPIC PARTITION CURRENT-OFFSET LOG-END-OFFSET LAG CONSUMER-ID HOST CLIENT-ID
sdc2loan_cardloan-apply_s_binlog_pro 0 11605539 11605554 15 consumer-97-1acf945d-a2d3-4be8-8660-410c036e289d /10.2.0.45 consumer-97

1.3 kafka维护的消费偏移

如果是用kafka默认的api【org.apache.kafka.clients.consumer.KafkaConsumer】来消费，其消费者的offset会更新到一个kafka自带的topic下：__consumer_offsets

__consumer_offsets默认有50个分区

[zk: localhost:2181(CONNECTED) 4] ls /kafka/brokers/topics/__consumer_offsets/partitions
[44, 45, 46, 47, 48, 49, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43]

查看__consumer_offsets主题所有信息

./bin/kafka-console-consumer.sh --topic __consumer_offsets --bootstrap-server localhost:9092 --formatter "kafka.coordinator.group.GroupMetadataManager\$OffsetsMessageFormatter" --consumer.config config/consumer.properties --from-beginning

查看__consumer_offsets主题下某个分区所有信息

bin/kafka-simple-consumer-shell.sh --topic __consumer_offsets --partition 11 --broker-list localhost:9092,localhost:9093,localhost:9094 --formatter "kafka.coordinator.group.GroupMetadataManager\$OffsetsMessageFormatter"

计算指定group在__consumer_offsets的哪个分区
```
Math.abs(groupID.hashCode()) % numPartitions
```
所以在本例中，对应的分区=Math.abs("console-consumer-46965".hashCode()) % 50 = 11，即__consumer_offsets的分区11保存了这个consumer group的位移信息。

2.zookeeper维护消费偏移

2.1 创建生产者

bin/kafka-console-producer.sh --broker-list localhost:9092 --topic test

$ bin/kafka-console-producer.sh --broker-list localhost:9092 --topic test
>1
>2
>3
>1

2.2 创建消费组

bin/kafka-console-consumer.sh --zookeeper localhost:2181/kafka --topic test --consumer-property group.id=my_group

bin/kafka-console-consumer.sh --zookeeper localhost:2181/kafka --topic test --consumer-property group.id=my_group
Using the ConsoleConsumer with old consumer is deprecated and will be removed in a future major release. Consider using the new consumer by passing [bootstrap-server] instead of [zookeeper].
3
1
2
1

2.3 查看zookeeper维护的消费组列表

./bin/kafka-consumer-groups.sh --zookeeper localhost:2181/kafka --list

$ ./bin/kafka-consumer-groups.sh --zookeeper localhost:2181/kafka --list
Note: This will only show information about consumers that use ZooKeeper (not those using the Java consumer API).
console-consumer-7101
console-consumer-82464
my_group
console-consumer-2010

2.3 查看zookeeper维护的某个消费组

./bin/kafka-consumer-groups.sh --zookeeper localhost:2181 --group test-consumer-group1 --describe

$ ./bin/kafka-consumer-groups.sh --zookeeper localhost:2181/kafka --group my_group1 --describe
Note: This will only show information about consumers that use ZooKeeper (not those using the Java consumer API).

TOPIC PARTITION CURRENT-OFFSET LOG-END-OFFSET LAG CONSUMER-ID
test 0 15 17 2 -
test 1 16 18 2 -
test 2 14 16 2 -

2.4 zookeeper维护的消费偏移

如果是用javaapi【kafka.javaapi.consumer.ConsumerConnector】来消费，消费者的offset会更新到zookeeper目录下：/kafka/consumers/{group}/offsets/{topic}/{partition}

get /kafka/consumers/my_group1/offsets/test/0

[zk: localhost:2181(CONNECTED) 5] get /kafka/consumers/my_group1/offsets/test/0
15
cZxid = 0x328a8e539
ctime = Tue Apr 14 19:01:13 CST 2020
mZxid = 0x328b43eac
mtime = Tue Apr 14 23:51:14 CST 2020
pZxid = 0x328a8e539
cversion = 0
dataVersion = 4
aclVersion = 0
ephemeralOwner = 0x0
dataLength = 2
numChildren = 0

2.5 查看topic的offset的范围

查看offset的最小值

$ bin/kafka-run-class.sh kafka.tools.GetOffsetShell --broker-list localhost:9092 -topic test --time -2
test:2:0
test:1:0
test:0:0

查看offset的最大值

$ bin/kafka-run-class.sh kafka.tools.GetOffsetShell --broker-list localhost:9092 -topic test --time -1
test:2:16
test:1:18
test:0:17

3. 消费偏移的更新方式

无论是kafka默认api，还是java的api，offset的更新方式都有两种：自动提交和手动提交

3.1 自动提交（默认方式）

Kafka中偏移量的自动提交是由参数enable_auto_commit和auto_commit_interval_ms控制的，当enable_auto_commit=True时，Kafka在消费的过程中会以频率为auto_commit_interval_ms向Kafka自带的topic(__consumer_offsets)进行偏移量提交，具体提交到哪个Partation：Math.abs(groupID.hashCode()) % numPartitions。

这种方式也被称为at most once，fetch到消息后就可以更新offset，无论是否消费成功。

3.2 手动提交

鉴于Kafka自动提交offset的不灵活性和不精确性(只能是按指定频率的提交)，Kafka提供了手动提交offset策略。手动提交能对偏移量更加灵活精准地控制，以保证消息不被重复消费以及消息不被丢失。

对于手动提交offset主要有3种方式：1.同步提交 2.异步提交 3.异步+同步组合的方式提交

3.2.1 同步提交