kafka bin目录下的各种脚本,功能如下:
脚本名称 | 用途描述 |
---|---|
connect-distributed.sh | 连接kafka集群模式 |
connect-standalone.sh | 连接kafka单机模式 |
kafka-acls.sh | todo |
kafka-broker-api-versions.sh | todo |
kafka-configs.sh | 配置管理脚本 |
kafka-console-consumer.sh | kafka消费者控制台 |
kafka-console-producer.sh | kafka生产者控制台 |
kafka-consumer-groups.sh | kafka消费者组相关信息 |
kafka-consumer-perf-test.sh | kafka消费者性能测试脚本 |
kafka-delegation-tokens.sh | todo |
kafka-delete-records.sh | 删除低水位的日志文件 |
kafka-log-dirs.sh | kafka消息日志目录信息 |
kafka-mirror-maker.sh | 不同数据中心kafka集群复制工具 |
kafka-preferred-replica-election.sh | 触发preferred replica选举 |
kafka-producer-perf-test.sh | kafka生产者性能测试脚本 |
kafka-reassign-partitions.sh | 分区重分配脚本 |
kafka-replay-log-producer.sh | todo |
kafka-replica-verification.sh | 复制进度验证脚本 |
kafka-run-class.sh | todo |
kafka-server-start.sh | 启动kafka服务 |
kafka-server-stop.sh | 停止kafka服务 |
kafka-simple-consumer-shell.sh | deprecated,推荐使用kafka-console-consumer.sh |
kafka-streams-application-reset.sh | todo |
kafka-topics.sh | topic管理脚本 |
kafka-verifiable-consumer.sh | 可检验的kafka消费者 |
kafka-verifiable-producer.sh | 可检验的kafka生产者 |
trogdor.sh | todo |
zookeeper-security-migration.sh | todo |
zookeeper-server-start.sh | 启动zk服务 |
zookeeper-server-stop.sh | 停止zk服务 |
zookeeper-shell.sh | zk客户端 |
connect-distributed.sh&connect-standalone.sh
Kafka Connect是在0.9以后加入的功能,主要是用来将其他系统的数据导入到Kafka,然后再将Kafka中的数据导出到另外的系统。可以用来做实时数据同步的ETL,数据实时分析处理等。
主要有2种模式:Standalone(单机模式)和Distribute(分布式模式)。
单机主要用来开发,测试,分布式的用于生产环境。
用法比较复杂,建议参考:Kafka Connect教程详解 https://3gods.com/bigdata/Kafka-Connect-Details.html
kafka-broker-api-versions.sh
用法:bin/kafka-broker-api-versions.sh –bootstrap-server 10.0.55.229:9092,10.0.55.229:9093,10.0.55.229:9094
kafka-configs.sh
配置管理脚本,这个脚本主要分两类用法:describe和alter。
describe相关用法:
查看每个topic的配置:bin/kafka-configs.sh –zookeeper localhost:2181 –describe –entity-type topics
部分结果如下:
Configs for topic 'afei' are
Configs for topic 'TOPIC-TEST-AFEI' are retention.ms=600000
Configs for topic '__consumer_offsets' are segment.bytes=104857600,cleanup.policy=compact,compression.type=producer
查看broker的配置:bin/kafka-configs.sh –bootstrap-server localhost:9092 –describe –entity-type brokers –entity-name 0
说明:0是broker.id,因为entity-type为brokers,所以entity-name表示broker.id。
alter相关用法:
给指定topic增加配置:bin/kafka-configs.sh –zookeeper localhost:2181 –alter –entity-type topics –entity-name TOPIC-TEST-AFEI –add-config retention.ms=600000
给指定topic删除配置:bin/kafka-configs.sh –zookeeper localhost:2181 –alter –entity-type topics –entity-name TOPIC-TEST-AFEI –delete-config max.message.bytes
通过该脚本可以管理的属性,可以通过执行bin/kafka-configs.sh得到的输出中–add-config的desc可以得到。
kafka-broker-api-versions.sh
用法:bin/kafka-broker-api-versions.sh –bootstrap-server localhost:9092
kafka-console-consumer.sh
用法:bin/kafka-console-consumer.sh –bootstrap-server localhost:9092 –topic afei [–group groupName] [–partition 目标分区]
这个命令后面还可带很多参数:
--key-deserializer:指定key的反序列化方式,默认是 org.apache.kafka.common.serialization.StringDeserializer
--value-deserializer:指定value的反序列化方式,默认是 org.apache.kafka.common.serialization.StringDeserializer
--from-beginning:从最早的消息开始消费,默认是从最新消息开始消费。
--offset: 从指定的消息位置开始消费,如果设置了这个参数,还需要带上--partition。否则会提示:The partition is required when offset is specified.
--timeout-ms:当消费者在这个参数指定时间间隔内没有收到消息就会推出,并抛出异常:kafka.consumer.ConsumerTimeoutException。
--whitelist:接收的topic白名单集合,和--topic二者取其一。例如:--whitelist "afei.*"(以afei开头的topic),--whitelist "afei"(指定afei这个topic),"afei|fly"(指定afei或者fly这两个topic)。另外一个参数--blacklist用法类似。
这个命令后面还可带很多参数:
--key-serializer:指定key的序列化方式,默认是 org.apache.kafka.common.serialization.StringSerializer
--value-serializer:指定value的序列化方式,默认是 org.apache.kafka.common.serialization.StringSerializer
bin/kafka-consumer-groups.sh –bootstrap-server localhost:9092 –list
查看某个消费者组:
bin/kafka-consumer-groups.sh –bootstrap-server localhost:9092 –group AfeiGroup –describe
输出结果如下:
Note: This will not show information about old Zookeeper-based consumers.
TOPIC PARTITION CURRENT-OFFSET LOG-END-OFFSET LAG CONSUMER-ID HOST CLIENT-ID
afei 0 8 8 0 consumer-1-7a46c647-8221-4aca-b6bf-ed14571fb0f1 /172.18.36.203 consumer-1
afei 4 10 10 0 consumer-1-7a46c647-8221-4aca-b6bf-ed14571fb0f1 /172.18.36.203 consumer-1
afei 1 8 8 0 consumer-1-7a46c647-8221-4aca-b6bf-ed14571fb0f1 /172.18.36.203 consumer-1
afei 3 6 6 0 consumer-1-7a46c647-8221-4aca-b6bf-ed14571fb0f1 /172.18.36.203 consumer-1
afei 2 9 9 0 consumer-1-7a46c647-8221-4aca-b6bf-ed14571fb0f1 /172.18.36.203 consumer-1
bin/kafka-consumer-perf-test.sh –broker-list localhost:9092 –group testGroup –topic afei –messages 1024
输出结果如下:
start.time, end.time, data.consumed.in.MB, MB.sec, data.consumed.in.nMsg, nMsg.sec, rebalance.time.ms, fetch.time.ms, fetch.MB.sec, fetch.nMsg.sec
2018-07-02 22:49:10:068, 2018-07-02 22:49:12:077, 0.0001, 0.0001, 41, 20.4082, 19, 1990, 0.0001, 20.6030
bin/kafka-delete-records.sh –bootstrap-server 10.0.55.229:9092,10.0.55.229:9093,10.0.55.229:9094 –offset-json-file offset.json
offset.json文件内容:
{
"partitions": [{
"topic": "afei",
"partition": 3,
"offset": 10
}],
"version": 1
}
执行结果如下,表示删除afei这个topic下分区为3的offset少于10的消息日志(不会删除offset=10的消息日志):
Executing records delete operation
Records delete operation completed:
partition: afei-3 low_watermark: 10
bin/kafka-log-dirs.sh –bootstrap-server localhost:9092 –describe –topic-list afei[,topicName2,topicNameN]
如果没有指定–topic-list,那么会输出所有kafka消息日志目录以及目录下所有topic信息。加上–topic-list参数后,输出结果如下,由这段结果可知,消息日志所在目录为/data/kafka-logs,并且afei这个topic有3个分区:
{
"version": 1,
"brokers": [{
"broker": 0,
"logDirs": [{
"logDir": "/data/kafka-logs",
"error": null,
"partitions": [{
"partition": "afei-1",
"size": 567,
"offsetLag": 0,
"isFuture": false
}, {
"partition": "afei-2",
"size": 639,
"offsetLag": 0,
"isFuture": false
}, {
"partition": "afei-0",
"size": 561,
"offsetLag": 0,
"isFuture": false
}]
}]
}]
}
bin/kafka-preferred-replica-election.sh –zookeeper 10.0.55.208:2181/wallet,10.0.55.209:2181/wallet,10.0.55.210:2181/wallet –path-to-json-file afei-preferred.json
(如果不带–path-to-json-file就是对所有topic进行preferred replica election),json文件内容如下::
{
"partitions": [{
"topic": "afei",
"partition": 0
},
{
"topic": "afei",
"partition": 1
},
{
"topic": "afei",
"partition": 2
}]
}
场景: 在创建一个topic时,kafka尽量将partition均分在所有的brokers上,并且将replicas也均分在不同的broker上。每个partitiion的所有replicas叫做”assigned replicas”,”assigned replicas”中的第一个replicas叫”preferred replica”,刚创建的topic一般”preferred replica”是leader。leader replica负责所有的读写。其他replica只是冷备状态,不接受读写请求。但随着时间推移,broker可能会主动停机甚至客观宕机,会发生leader选举迁移,导致机群的负载不均衡。我们期望对topic的leader进行重新负载均衡,让partition选择”preferred replica”做为leader。
kafka提供了一个参数auto.leader.rebalance.enable自动做这件事情,且默认为true,原理是一个后台线程检查并触发leader balance。但是并不建议把这个参数设置为true。因为担心这个自动选举发生在业务高峰期,从而导致影响业务。
验证:
操作比较简单,常见一个3个分区3个副本的topic,然后kill掉一个broker。这时候topic信息如下,我们可以看到broker.id为0的broker上有两个leader:
Topic:afei PartitionCount:3 ReplicationFactor:3 Configs:
Topic: afei Partition: 0 Leader: 0 Replicas: 0,1,2 Isr: 0,1,2
Topic: afei Partition: 1 Leader: 1 Replicas: 1,2,0 Isr: 0,1,2
Topic: afei Partition: 2 Leader: 0 Replicas: 2,0,1 Isr: 0,1,2
执行kafka-preferred-replica-election.sh脚本后,topic信息如下,leader均匀分布在3个不同的broker上,
Topic:afei PartitionCount:3 ReplicationFactor:3 Configs:
Topic: afei Partition: 0 Leader: 0 Replicas: 0,1,2 Isr: 0,1,2
Topic: afei Partition: 1 Leader: 1 Replicas: 1,2,0 Isr: 0,1,2
Topic: afei Partition: 2 Leader: 2 Replicas: 2,0,1 Isr: 0,1,2
kafka-producer-perf-test.sh
perf是performance的缩写,所以这个脚本是kafka生产者性能测试脚本。
kafka-reassign-partitions.sh
场景:将一些topic上的分区从当前所在broker移到其他比如新增的broker上。假设有个名为ORDER-DETAIL的topic,在broker.id为2的broker上:
Topic:ORDER-DETAIL PartitionCount:1 ReplicationFactor:1 Configs:
Topic: ORDER-DETAIL Partition: 0 Leader: 2 Replicas: 2 Isr: 2
现在想要把它移动到broker.id为1的broker上,执行脚本:
bin/kafka-reassign-partitions.sh –zookeeper 10.0.55.208:2181/wallet,10.0.55.209:2181/wallet,10.0.55.210:2181/wallet –topics-to-move-json-file move.json –broker-list “1” –generate
–generate 参数表示生成一个分区再分配配置,并不会真正的执行,命令执行结果如下:
Current partition replica assignment
{"version":1,"partitions":[{"topic":"ORDER-DETAIL","partition":0,"replicas":[2],"log_dirs":["any"]}]}
Proposed partition reassignment configuration
{"version":1,"partitions":[{"topic":"ORDER-DETAIL","partition":0,"replicas":[1],"log_dirs":["any"]}]}
我们只需要把第二段json内容保存到一个新建的final.json文件中(如果知道如何编写这段json内容,那么也可以不执行第一条命令),然后执行命令:bin/kafka-reassign-partitions.sh –zookeeper 10.0.55.208:2181/wallet,10.0.55.209:2181/wallet,10.0.55.210:2181/wallet –reassignment-json-file move_final.json –execute,此次执行的命令带有–execute参数,说明是真正的执行分区重分配。
通过这个命令还可以给某个topic增加副本,例如有一个名为ORDER-DETAIL的topic,有3个分区,但是只有1个副本,为了高可用,需要将副本数增加到2,那么编写replica.json文本内容如下:
{
"version": 1,
"partitions": [{
"topic": "ORDER-DETAIL",
"partition": 0,
"replicas": [1, 2]
},
{
"topic": "ORDER-DETAIL",
"partition": 1,
"replicas": [0, 2]
},
{
"topic": "ORDER-DETAIL",
"partition": 2,
"replicas": [1, 0]
}]
}
然后执行命令即可:
bin/kafka-reassign-partitions.sh –zookeeper 10.0.55.208:2181/wallet,10.0.55.209:2181/wallet,10.0.55.210:2181/wallet –reassignment-json-file replica.json
bin/kafka-replica-verification.sh –broker-list 10.0.55.229:9092,10.0.55.229:9093,10.0.55.229:9094 [–topic-white-list afei]
参数–topic-white-list指定要检查的目标topic。输出结果如下,如果输出信息为max lag is 0 for …表示这个topic的复制没有任何延迟:
2018-07-03 15:04:46,889: verification process is started.
2018-07-03 15:05:16,811: max lag is 0 for partition multi-1 at offset 0 among 5 partitions
2018-07-03 15:05:46,812: max lag is 0 for partition multi-1 at offset 0 among 5 partitions
... ...
bin/kafka-server-start.sh -daemon config/server.properties,指定配置文件并以守护进程模式启动。
bin/kafka-server-stop.sh 。
说明:这个命令会kill掉当前服务器上所有kafka broker。但是这个脚本可能执行结果为:No kafka server to stop
分析原因:我们先看一下kafka-server-stop.sh脚本内容,这个脚本非常简单,就是得到所有包含kafka.Kafka的进程ID,但是由于kafka启动依赖比较多的jar,导致kafka进程的ps结果输出内容比较长,而ps输出结果受到PAGE_SIZE(其值通过命令getconf PAGE_SIZE可以得到)的限制,从而导致ps结果中看不到kafka.Kafka,所以不能kill掉kafka server:
SIGNAL=${SIGNAL:-TERM}
PIDS=$(ps ax | grep -i 'kafka\.Kafka' | grep java | grep -v grep | awk '{print $1}')
if [ -z "$PIDS" ]; then
echo "No kafka server to stop"
exit 1
else
kill -s $SIGNAL $PIDS
fi
为了kafka-server-stop.sh脚本可以正常执行,建议修改脚本如下,通过bin脚本所在目录的上级目录来查找进程ID,从而kill相关进程:
cd `dirname $0`
BIN_DIR=`pwd`
cd ..
DEPLOY_DIR=`pwd`
SIGNAL=${SIGNAL:-TERM}
PIDS=$(ps ax | grep -i "${DEPLOY_DIR}" | grep java | grep -v grep | awk '{print $1}')
if [ -z "$PIDS" ]; then
echo "No kafka server to stop"
exit 1
else
kill -s $SIGNAL $PIDS
fi
bin/kafka-simple-consumer-shell.sh –broker-list 10.0.55.229:9092,10.0.55.229:9093,10.0.55.229:9094 –topic afei
创建topic: bin/kafka-topics.sh –zookeeper localhost:2181 –create –topic afei –partitions 3 –replication-factor 1
删除topic: bin/kafka-topics.sh –zookeeper localhost:2181 –delete –topic test
修改topic: bin/kafka-topics.sh –zookeeper localhost:2181 –alter –topic afei –partitions 5,修改topic时只能增加分区数量。
查询topic: bin/kafka-topics.sh –zookeeper localhost:2181 –describe [ –topic afei ],查询时如果带上–topic topicName,那么表示只查询该topic的详细信息。这时候还可以带上–unavailable-partitions 和–under-replicated-partitions任意一个参数。
bin/kafka-verifiable-consumer.sh –broker-list 10.0.55.229:9092,10.0.55.229:9093,10.0.55.229:9094 –topic afei –group-id groupName
这个脚本的作用是接收指定topic的消息消费,并发出消费者事件,例如:offset提交等。
bin/kafka-verifiable-producer.sh –broker-list 10.0.55.229:9092,10.0.55.229:9093,10.0.55.229:9094 –topic afei [–max-messages 64]
建议使用该脚本时增加参数–max-messages,否则会不停的发送消息。
这个脚本的作用是持续发送消息到指定的topic中,参数–max-messages限制最大发送消息数。且每条发送的消息都会有响应信息,这就是和kafka-console-producer.sh最大的不同:
[mwopr@jtcrtvdra35 kafka_2.12-1.1.0]$ bin/kafka-verifiable-producer.sh --broker-list 10.0.55.229:9092,10.0.55.229:9093,10.0.55.229:9094 --topic afei --max-messages 9
{"timestamp":1530515959900,"name":"startup_complete"}
{"timestamp":1530515960310,"name":"producer_send_success","key":null,"value":"1","offset":5,"topic":"afei","partition":0}
{"timestamp":1530515960315,"name":"producer_send_success","key":null,"value":"4","offset":6,"topic":"afei","partition":0}
{"timestamp":1530515960315,"name":"producer_send_success","key":null,"value":"7","offset":7,"topic":"afei","partition":0}
{"timestamp":1530515960316,"name":"producer_send_success","key":null,"value":"0","offset":5,"topic":"afei","partition":1}
{"timestamp":1530515960316,"name":"producer_send_success","key":null,"value":"3","offset":6,"topic":"afei","partition":1}
{"timestamp":1530515960316,"name":"producer_send_success","key":null,"value":"6","offset":7,"topic":"afei","partition":1}
{"timestamp":1530515960316,"name":"producer_send_success","key":null,"value":"2","offset":6,"topic":"afei","partition":2}
{"timestamp":1530515960316,"name":"producer_send_success","key":null,"value":"5","offset":7,"topic":"afei","partition":2}
{"timestamp":1530515960316,"name":"producer_send_success","key":null,"value":"8","offset":8,"topic":"afei","partition":2}
{"timestamp":1530515960333,"name":"shutdown_complete"}
{"timestamp":1530515960334,"name":"tool_data","sent":9,"acked":9,"target_throughput":-1,"avg_throughput":20.689655172413794}
afei这个topic有3个分区,使用kafka-verifiable-producer.sh发送9条消息。根据输出结果可以看出,往每个分区发送了3条消息。另外,我们可以通过设置参数–max-messages一个比较大的值,可以压测一下搭建的kafka集群环境。
bin/zookeeper-shell.sh localhost:2181[/path],
如果kafka集群的zk配置了chroot路径,那么需要加上/path,例如bin/zookeeper-shell.sh localhost:2181/mykafka,登陆zk后,就可以查看kafka写在zk上的节点信息。例如查看有哪些broker,以及broker的详细信息:
ls /brokers/ids
[0]
get /brokers/ids/0
{"listener_security_protocol_map":{"PLAINTEXT":"PLAINTEXT"},"endpoints":["PLAINTEXT://izwz91rhzhed2c54e6yx87z:9092"],"jmx_port":-1,"host":"izwz91rhzhed2c54e6yx87z","timestamp":"1530435834272","port":9092,"version":4}
cZxid = 0x2d3
ctime = Sun Jul 01 17:03:54 CST 2018
mZxid = 0x2d3
mtime = Sun Jul 01 17:03:54 CST 2018
pZxid = 0x2d3
cversion = 0
dataVersion = 0
aclVersion = 0
ephemeralOwner = 0x1642cb09421006c
dataLength = 216
numChildren = 0