Kafka - 即使应用程序配置为恰好一次和最高持久性，也会丢失消息
Kafka - Loosing messages even if app is configured for exactly once and highest durability
在某些情况下（很少，但确实存在）我会收到重复项，即使所有内容都配置为高持久性并且我们只使用一次配置。
请在下面检查导致此问题的应用程序上下文和测试场景。
Kafka 集群设置
3 个 Kafka 代理（1 个在主机 1 上，2 个在主机 2 上，3 个在主机 3 上）
3 个 Zookeeper 实例（1 个在主机 1 上，2 个在主机 2 上，3 个在主机 3 上）
卡夫卡配置
broker.id=1,2,3
num.network.threads=2
num.io.threads=8
socket.send.buffer.bytes=102400
socket.receive.buffer.bytes=102400
socket.request.max.bytes=104857600
log.dirs=/home/kafka/logs/kafka
min.insync.replicas=3
transaction.state.log.min.isr=3
default.replication.factor=3
log.retention.minutes=600
log.segment.bytes=1073741824
log.retention.check.interval.ms=300000
zookeeper.connect=host1:2181,host2:2181,host3:2181
zookeeper.connection.timeout.ms=6000
group.initial.rebalance.delay.ms=1000
log.message.timestamp.type=LogAppendTime
delete.topic.enable=true
auto.create.topics.enable=false
unclean.leader.election.enable=false
ZooKeeper 配置
tickTime=2000
dataDir=/home/kafka/logs/zk
clientPort=2181
maxClientCnxns=0
initLimit=5
syncLimit=2
server.1=host1:2888:3888
server.2=host2:2888:3888
server.3=host3:2888:3888
autopurge.snapRetainCount=3
autopurge.purgeInterval=24
Kafka 内部主题描述
Topic:__transaction_state PartitionCount:50 ReplicationFactor:3 Configs:segment.bytes=104857600,unclean.leader.election.enable=false,compression.type=uncompressed,cleanup.policy=compact,min.insync.replicas=3
Topic: __transaction_state Partition: 0 Leader: 1 Replicas: 3,2,1 Isr: 1,2,3
Topic:__consumer_offsets PartitionCount:50 ReplicationFactor:3 Configs:segment.bytes=104857600,unclean.leader.election.enable=false,min.insync.replicas=3,cleanup.policy=compact,compression.type=producer
Topic: __consumer_offsets Partition: 0 Leader: 1 Replicas: 3,2,1 Isr: 1,2,3
申请主题
Topic input-event
Topic:input-event PartitionCount:3 ReplicationFactor:3 Configs:retention.ms=28800001,unclean.leader.election.enable=false,min.insync.replicas=3,message.timestamp.difference.max.ms=28800000
Topic: input-event Partition: 0 Leader: 1 Replicas: 1,2,3 Isr: 1,2,3
Topic: input-event Partition: 1 Leader: 2 Replicas: 2,3,1 Isr: 1,2,3
Topic: input-event Partition: 2 Leader: 3 Replicas: 3,1,2 Isr: 1,2,3
Topic output-event
Topic:output-event PartitionCount:3 ReplicationFactor:3 Configs:retention.ms=28800001,unclean.leader.election.enable=false,min.insync.replicas=3,message.timestamp.difference.max.ms=28800000
Topic: output-event Partition: 0 Leader: 2 Replicas: 2,3,1 Isr: 1,2,3
Topic: output-event Partition: 1 Leader: 3 Replicas: 3,1,2 Isr: 1,2,3
Topic: output-event Partition: 2 Leader: 1 Replicas: 1,2,3 Isr: 1,2,3
应用程序消费者属性
o.a.k.clients.consumer.ConsumerConfig : ConsumerConfig values:
auto.commit.interval.ms = 5000
auto.offset.reset = earliest
bootstrap.servers = [host1:9092, host2:9092, host3:9092]
check.crcs = true
client.id =
connections.max.idle.ms = 540000
default.api.timeout.ms = 60000
enable.auto.commit = false
exclude.internal.topics = true
fetch.max.bytes = 134217728
fetch.max.wait.ms = 500
fetch.min.bytes = 1
group.id = groupId
heartbeat.interval.ms = 3000
interceptor.classes = []
internal.leave.group.on.close = true
isolation.level = read_committed
key.deserializer = class org.apache.kafka.common.serialization.ByteArrayDeserializer
max.partition.fetch.bytes = 134217728
max.poll.interval.ms = 300000
max.poll.records = 1
metadata.max.age.ms = 300000
metric.reporters = []
metrics.num.samples = 2
metrics.recording.level = INFO
metrics.sample.window.ms = 30000
partition.assignment.strategy = [class org.apache.kafka.clients.consumer.RangeAssignor]
receive.buffer.bytes = 65536
reconnect.backoff.max.ms = 1000
reconnect.backoff.ms = 1000
request.timeout.ms = 30000
retry.backoff.ms = 1000
sasl.client.callback.handler.class = null
sasl.jaas.config = null
sasl.kerberos.kinit.cmd = /usr/bin/kinit
sasl.kerberos.min.time.before.relogin = 60000
sasl.kerberos.service.name = null
sasl.kerberos.ticket.renew.jitter = 0.05
sasl.kerberos.ticket.renew.window.factor = 0.8
sasl.login.callback.handler.class = null
sasl.login.class = null
sasl.login.refresh.buffer.seconds = 300
sasl.login.refresh.min.period.seconds = 60
sasl.login.refresh.window.factor = 0.8
sasl.login.refresh.window.jitter = 0.05
sasl.mechanism = GSSAPI
security.protocol = PLAINTEXT
send.buffer.bytes = 131072
session.timeout.ms = 10000
ssl.cipher.suites = null
ssl.enabled.protocols = [TLSv1.2, TLSv1.1, TLSv1]
ssl.endpoint.identification.algorithm = https
ssl.key.password = null
ssl.keymanager.algorithm = SunX509
ssl.keystore.location = null
ssl.keystore.password = null
ssl.keystore.type = JKS
ssl.protocol = TLS
ssl.provider = null
ssl.secure.random.implementation = null
ssl.trustmanager.algorithm = PKIX
ssl.truststore.location = null
ssl.truststore.password = null
ssl.truststore.type = JKS
value.deserializer = class org.apache.kafka.common.serialization.ByteArrayDeserializer
应用程序生产者属性
bootstrapServers = "host1, host2, host3"
transactionIdPrefix = "my-producer-"${instance}"
"enable.idempotence" = "true"
"acks" = "all"
"retries" = "2147483647"
"transaction.timeout.ms" = "10000"
"max.in.flight.requests.per.connection" = "1"
"reconnect.backoff.max.ms" = "1000"
"reconnect.backoff.ms" = "1000"
"retry.backoff.ms" = "1000"
应用程序处理提交
使用 KafkaTransactionManager，我们启动事务，使用 KafkaTemplate 将消息写入输出主题并发送消费者偏移量（spring-kafka 2.2.8.RELEASE）。
测试expected/actual
写 32,000 条消息到输入主题
启动3个应用实例
开始一条一条地处理消息(max.poll.records = 1)
处理过程中，向host1和host2 Kafka Brokers并行发送SIGKILL(kill -9)50次
等待 60 秒
处理过程中，向host1和host3 Kafka Brokers并行发送SIGKILL(kill -9)50次
等待 60 秒
处理过程中，向host2和host3 Kafka Brokers并行发送SIGKILL(kill -9)50次
期望输出主题有 32,000 条消息，但是，有时我们实际上会得到重复的（至少一个）。
有时我们最终收到 32,000 条消息，但一切正常。
这个问题与交易 ID 没有正确设置在 topic.partition 级别有关，我们有两个生产者为同一个分区写了两次相同的消息。
这是一本好书：
https://tgrez.github.io/posts/2019-04-13-kafka-transactions.html
在某些情况下（很少，但确实存在）我会收到重复项，即使所有内容都配置为高持久性并且我们只使用一次配置。
请在下面检查导致此问题的应用程序上下文和测试场景。
Kafka 集群设置
3 个 Kafka 代理（1 个在主机 1 上，2 个在主机 2 上，3 个在主机 3 上）
3 个 Zookeeper 实例（1 个在主机 1 上，2 个在主机 2 上，3 个在主机 3 上）
卡夫卡配置
broker.id=1,2,3
num.network.threads=2
num.io.threads=8
socket.send.buffer.bytes=102400
socket.receive.buffer.bytes=102400
socket.request.max.bytes=104857600
log.dirs=/home/kafka/logs/kafka
min.insync.replicas=3
transaction.state.log.min.isr=3
default.replication.factor=3
log.retention.minutes=600
log.segment.bytes=1073741824
log.retention.check.interval.ms=300000
zookeeper.connect=host1:2181,host2:2181,host3:2181
zookeeper.connection.timeout.ms=6000
group.initial.rebalance.delay.ms=1000
log.message.timestamp.type=LogAppendTime
delete.topic.enable=true
auto.create.topics.enable=false
unclean.leader.election.enable=false
ZooKeeper 配置
tickTime=2000
dataDir=/home/kafka/logs/zk
clientPort=2181
maxClientCnxns=0
initLimit=5
syncLimit=2
server.1=host1:2888:3888
server.2=host2:2888:3888
server.3=host3:2888:3888
autopurge.snapRetainCount=3
autopurge.purgeInterval=24
Kafka 内部主题描述
Topic:__transaction_state PartitionCount:50 ReplicationFactor:3 Configs:segment.bytes=104857600,unclean.leader.election.enable=false,compression.type=uncompressed,cleanup.policy=compact,min.insync.replicas=3
Topic: __transaction_state Partition: 0 Leader: 1 Replicas: 3,2,1 Isr: 1,2,3
Topic:__consumer_offsets PartitionCount:50 ReplicationFactor:3 Configs:segment.bytes=104857600,unclean.leader.election.enable=false,min.insync.replicas=3,cleanup.policy=compact,compression.type=producer
Topic: __consumer_offsets Partition: 0 Leader: 1 Replicas: 3,2,1 Isr: 1,2,3
申请主题
Topic input-event
Topic:input-event PartitionCount:3 ReplicationFactor:3 Configs:retention.ms=28800001,unclean.leader.election.enable=false,min.insync.replicas=3,message.timestamp.difference.max.ms=28800000
Topic: input-event Partition: 0 Leader: 1 Replicas: 1,2,3 Isr: 1,2,3
Topic: input-event Partition: 1 Leader: 2 Replicas: 2,3,1 Isr: 1,2,3
Topic: input-event Partition: 2 Leader: 3 Replicas: 3,1,2 Isr: 1,2,3
Topic output-event
Topic:output-event PartitionCount:3 ReplicationFactor:3 Configs:retention.ms=28800001,unclean.leader.election.enable=false,min.insync.replicas=3,message.timestamp.difference.max.ms=28800000
Topic: output-event Partition: 0 Leader: 2 Replicas: 2,3,1 Isr: 1,2,3
Topic: output-event Partition: 1 Leader: 3 Replicas: 3,1,2 Isr: 1,2,3
Topic: output-event Partition: 2 Leader: 1 Replicas: 1,2,3 Isr: 1,2,3
应用程序消费者属性
o.a.k.clients.consumer.ConsumerConfig : ConsumerConfig values:
auto.commit.interval.ms = 5000
auto.offset.reset = earliest
bootstrap.servers = [host1:9092, host2:9092, host3:9092]
check.crcs = true
client.id =
connections.max.idle.ms = 540000
default.api.timeout.ms = 60000
enable.auto.commit = false
exclude.internal.topics = true
fetch.max.bytes = 134217728
fetch.max.wait.ms = 500
fetch.min.bytes = 1
group.id = groupId
heartbeat.interval.ms = 3000
interceptor.classes = []
internal.leave.group.on.close = true
isolation.level = read_committed
key.deserializer = class org.apache.kafka.common.serialization.ByteArrayDeserializer
max.partition.fetch.bytes = 134217728
max.poll.interval.ms = 300000
max.poll.records = 1
metadata.max.age.ms = 300000
metric.reporters = []
metrics.num.samples = 2
metrics.recording.level = INFO
metrics.sample.window.ms = 30000
partition.assignment.strategy = [class org.apache.kafka.clients.consumer.RangeAssignor]
receive.buffer.bytes = 65536
reconnect.backoff.max.ms = 1000
reconnect.backoff.ms = 1000
request.timeout.ms = 30000
retry.backoff.ms = 1000
sasl.client.callback.handler.class = null
sasl.jaas.config = null
sasl.kerberos.kinit.cmd = /usr/bin/kinit
sasl.kerberos.min.time.before.relogin = 60000
sasl.kerberos.service.name = null
sasl.kerberos.ticket.renew.jitter = 0.05
sasl.kerberos.ticket.renew.window.factor = 0.8
sasl.login.callback.handler.class = null
sasl.login.class = null
sasl.login.refresh.buffer.seconds = 300
sasl.login.refresh.min.period.seconds = 60
sasl.login.refresh.window.factor = 0.8
sasl.login.refresh.window.jitter = 0.05
sasl.mechanism = GSSAPI
security.protocol = PLAINTEXT
send.buffer.bytes = 131072
session.timeout.ms = 10000
ssl.cipher.suites = null
ssl.enabled.protocols = [TLSv1.2, TLSv1.1, TLSv1]
ssl.endpoint.identification.algorithm = https
ssl.key.password = null
ssl.keymanager.algorithm = SunX509
ssl.keystore.location = null
ssl.keystore.password = null
ssl.keystore.type = JKS
ssl.protocol = TLS
ssl.provider = null
ssl.secure.random.implementation = null
ssl.trustmanager.algorithm = PKIX
ssl.truststore.location = null
ssl.truststore.password = null
ssl.truststore.type = JKS
value.deserializer = class org.apache.kafka.common.serialization.ByteArrayDeserializer
应用程序生产者属性
bootstrapServers = "host1, host2, host3"
transactionIdPrefix = "my-producer-"${instance}"
"enable.idempotence" = "true"
"acks" = "all"
"retries" = "2147483647"
"transaction.timeout.ms" = "10000"
"max.in.flight.requests.per.connection" = "1"
"reconnect.backoff.max.ms" = "1000"
"reconnect.backoff.ms" = "1000"
"retry.backoff.ms" = "1000"
应用程序处理提交
使用 KafkaTransactionManager，我们启动事务，使用 KafkaTemplate 将消息写入输出主题并发送消费者偏移量（spring-kafka 2.2.8.RELEASE）。
测试expected/actual
写 32,000 条消息到输入主题
启动3个应用实例
开始一条一条地处理消息(max.poll.records = 1)
处理过程中，向host1和host2 Kafka Brokers并行发送SIGKILL(kill -9)50次
等待 60 秒
处理过程中，向host1和host3 Kafka Brokers并行发送SIGKILL(kill -9)50次
等待 60 秒
处理过程中，向host2和host3 Kafka Brokers并行发送SIGKILL(kill -9)50次
期望输出主题有 32,000 条消息，但是，有时我们实际上会得到重复的（至少一个）。
有时我们最终收到 32,000 条消息，但一切正常。
这个问题与交易 ID 没有正确设置在 topic.partition 级别有关，我们有两个生产者为同一个分区写了两次相同的消息。
这是一本好书： https://tgrez.github.io/posts/2019-04-13-kafka-transactions.html