Flink集群搭建
Flink集群搭建
集群规划
| 节点 | node01 | node02 | node03 |
|---|---|---|---|
| 角色 | JobManager TaskManager | TaskManager | TaskManager |
下载并解压安装包
wget https://repo.huaweicloud.com/apache/flink/flink-1.17.0/flink-1.17.0-bin-scala_2.12.tgz
tar -zxvf flink-1.17.0-bin-scala_2.12.tgz
mv flink-1.17.0 flink
修改集群配置
vim /usr/local/program/flink/conf/flink-conf.yaml
1.修改flink-conf.yaml文件
# jobmanager.rpc.address: localhost
# jobmanager.bind-host: localhost
jobmanager.rpc.address: node01
jobmanager.bind-host: 0.0.0.0
# rest.address: localhost
# rest.bind-address: localhost
rest.address: node01
rest.bind-address: 0.0.0.0
# taskmanager.host: localhost
# taskmanager.bind-host: localhost
taskmanager.host: node01
taskmanager.bind-host: 0.0.0.0
注意：需要在/etc/hosts文件中配置各个节点信息
172.29.234.1 node01 node01
172.29.234.2 node02 node02
172.29.234.3 node03 node03
2.修改workers文件
# localhost
node01
node02
node03
3.修改masters文件
# localhost:8081
node01:8081
分发安装目录
[root@node01 program]# pwd
/usr/local/program
[root@node01 program]# ls
flink jdk8
[root@node01 program]# scp -r flink node02:/usr/local/program/flink
[root@node01 program]# scp -r flink node03:/usr/local/program/flink
在node02、node03节点，修改flink-conf.yaml 配置
1.node02节点
# taskmanager.host: localhost
taskmanager.host: node02
2.node03节点
# taskmanager.host: localhost
taskmanager.host: node03
启动集群
# 启动集群
./bin/start-cluster.sh
# 停止集群
./bin/stop-cluster.sh
在node01节点服务器上执行start-cluster.sh脚本以启动Flink集群
[root@node01 bin]# cd /usr/local/program/flink/bin
[root@node01 bin]# ./start-cluster.sh
Starting cluster.
Starting standalonesession daemon on host node01.
Starting taskexecutor daemon on host node01.
Starting taskexecutor daemon on host node02.
Starting taskexecutor daemon on host node03.
查看进程情况
[root@node01 bin]# jps
6788 StandaloneSessionClusterEntrypoint
7256 Jps
7116 TaskManagerRunner
[root@node02 conf]# jps
16884 TaskManagerRunner
16959 Jps
[root@node03 conf]# jps
17139 TaskManagerRunner
17214 Jps
访问Web UI
当如上所示一样后，代表启动成功，此时可以访问http://node01:8081对flink集群和任务进行监控管理。
注意：关闭防火墙，否则可能无法访问，或者集群的TaskManager数量、Slot数量显示异常
systemctl stop firewalld
提交任务
[root@node01 bin]# flink run ../examples/streaming/WordCount.jar
查看运行结果
[root@node01 bin]# tail flink-*-taskexecutor-*.out
也可以通过Flink的 Web UI来监视集群的状态和正在运行的作业
Flink集群HA高可用
概述
集群规划
| 节点 | node01 | node02 | node03 |
|---|---|---|---|
| 角色 | JobManager TaskManager | JobManager TaskManager | TaskManager |
配置flink
基于Flink集群的node01节点配置的情况下，修改conf/flink-conf.yaml文件，增加如下配置：
# 配置使用zookeeper来开启高可用模式
high-availability.type: zookeeper
# 配置zookeeper的地址，采用zookeeper集群时，可以使用逗号来分隔多个节点地址
high-availability.zookeeper.quorum: node01:2181,node02:2181,node03:2181
# 在zookeeper上存储flink集群元信息的路径
high-availability.zookeeper.path.root: /flink
# 集群id 放置集群的所有必需协调数据
high-availability.cluster-id: /cluster_one
# 持久化存储JobManager元数据的地址，zookeeper上存储的只是指向该元数据的指针信息
high-availability.storageDir: hdfs://node01:9000/flink/recovery
配置master、workers
修改conf/masters文件，配置master节点
node01:8081
node02:8081
修改conf/workers文件，配置worker节点
node01
node02
node03
配置ZK
编辑vim zoo.cfg文件
server.1=node01:2888:3888
server.2=node02:2888:3888
server.3=node03:2888:3888
分发安装目录
[root@node01 program]# pwd
/usr/local/program
[root@node01 program]# ls
flink jdk8
[root@node01 program]# scp -r flink node02:/usr/local/program/flink
[root@node01 program]# scp -r flink node03:/usr/local/program/flink
在node02、node03节点，修改flink-conf.yaml 配置
1.node02节点
jobmanager.rpc.address: node02
taskmanager.host: node02
2.node03节点
taskmanager.host: node03
启动HA集群
分发Flink相关配置到其他节点，然后确保Hadoop和ZooKeeper已经启动后，使用以下命令来启动集群：
[root@node01 flink]# bin/start-cluster.sh
Starting HA cluster with 2 masters.
Starting standalonesession daemon on host node01.
Starting standalonesession daemon on host node02.
Starting taskexecutor daemon on host node01.
Starting taskexecutor daemon on host node02.
Starting taskexecutor daemon on host node03.
访问http://node01:8081
访问http://node02:8081
测试
查看ZK：JobManager节点信息
kill node01节点上的JobManager进程
[root@node01 flink]# jps
2564 DataNode
3508 NodeManager
18741 Jps
7784 QuorumPeerMain
16666 TaskManagerRunner
2363 NameNode
16300 StandaloneSessionClusterEntrypoint
3117 ResourceManager
[root@node01 flink]# kill -9 16300
查看Active JobManager是否变化
Flink参数配置
# jobmanager地址
jobmanager.rpc.address: node01
# JobManager 的 JVM 堆内存大小，默认为 1024m
jobmanager.heap.size: 1024m
# rpc通信端口
jobmanager.rpc.port: 6123
# 进程使用的全部内存大小,可以根据集群规模进行适当调整
jobmanager.memory.process.size：1600m
# Taskmanager 的 JVM 堆内存大小，默认为 1024m
taskmanager.heap.size: 1024m
# 进程使用的全部内存大小,可以根据集群规模进行适当调整
taskmanager.memory.process.size: 1728m
# 每个TaskManager能够分配的Slot数量进行配置，默认为1
# 通常设置为 CPU 核心的数量，或其一半
# Slot就是TaskManager中具体运行一个任务所分配的计算资源
taskmanager.numberOfTaskSlots: 1
# flink任务执行的并行度，默认为1
# 优先级低于代码中进行的并行度配置和任务提交时使用参数指定的并行度数量
parallelism.default: 1
# 重启策略
jobmanager.execution.failover-strategy: region
# 存储临时文件的路径，如果没有配置，则默认采用服务器的临时目录，如 LInux 的 /tmp 目录
io.tmp.dirs: /tmp
参考Flink的官方手册：更多配置
配置历史服务器
概述
配置
创建存储目录
[root@node01 flink]# hadoop fs -mkdir -p /logs/flink-job
在flink-config.yaml中添加如下配置
#==============================================================================
# HistoryServer
#==============================================================================
# The HistoryServer is started and stopped via bin/historyserver.sh (start|stop)
# Directory to upload completed jobs to. Add this directory to the list of
# monitored directories of the HistoryServer as well (see below).
#jobmanager.archive.fs.dir: hdfs:///completed-jobs/
jobmanager.archive.fs.dir: hdfs://node01:9000/logs/flink-job
# The address under which the web-based HistoryServer listens.
#historyserver.web.address: 0.0.0.0
historyserver.web.address: node01
# The port under which the web-based HistoryServer listens.
#historyserver.web.port: 8082
historyserver.web.port: 8082
# Comma separated list of directories to monitor for completed jobs.
#historyserver.archive.fs.dir: hdfs:///completed-jobs/
historyserver.archive.fs.dir: hdfs://node01:9000/logs/flink-job
# Interval in milliseconds for refreshing the monitored directories.
#historyserver.archive.fs.refresh-interval: 10000
historyserver.archive.fs.refresh-interval: 5000
启动、停止历史服务器
启动历史服务器
[root@node01 flink]# bin/historyserver.sh start
Starting historyserver daemon on host node01.
停止历史服务器
[root@node01 flink]# bin/historyserver.sh stop
Stopping historyserver daemon (pid: 30749) on host node01.
提交一个Job任务
[root@node01 flink]# bin/flink run -t yarn-per-job -c com.atguigu.wc.WordCountStreamUnboundedDemo /root/FlinkTutorial-1.17-1.0-SNAPSHOT.jar
2023-06-12 23:41:00,719 INFO org.apache.hadoop.hdfs.protocol.datatransfer.sasl.SaslDataTransferClient [] - SASL encryption trust check: localHostTrusted = false, remoteHostTrusted = false
2023-06-12 23:41:00,742 INFO org.apache.hadoop.hdfs.protocol.datatransfer.sasl.SaslDataTransferClient [] - SASL encryption trust check: localHostTrusted = false, remoteHostTrusted = false
2023-06-12 23:41:00,761 INFO org.apache.flink.yarn.YarnClusterDescriptor [] - Cannot use kerberos delegation token manager, no valid kerberos credentials provided.
2023-06-12 23:41:00,766 INFO org.apache.flink.yarn.YarnClusterDescriptor [] - Submitting application master application_1686577483648_0012
2023-06-12 23:41:00,792 INFO org.apache.hadoop.yarn.client.api.impl.YarnClientImpl [] - Submitted application application_1686577483648_0012
2023-06-12 23:41:00,792 INFO org.apache.flink.yarn.YarnClusterDescriptor [] - Waiting for the cluster to be allocated
2023-06-12 23:41:00,793 INFO org.apache.flink.yarn.YarnClusterDescriptor [] - Deploying cluster, current state ACCEPTED
2023-06-12 23:41:04,565 INFO org.apache.flink.yarn.YarnClusterDescriptor [] - YARN application has been deployed successfully.
2023-06-12 23:41:04,565 INFO org.apache.flink.yarn.YarnClusterDescriptor [] - Found Web Interface node02:38887 of application 'application_1686577483648_0012'.
Job has been submitted with JobID cd41d983c93d8eb906c9aa899dcdefd0
访问http://node01:8088/cluster查看Hadoop
访问Web UI查看提交任务信息
查看历史Job信息
在浏览器地址栏输入：http://node01:8082 查看已经停止的 job 的统计信息
停止提交任务
[root@node01 flink]# bin/flink cancel -t yarn-per-job -Dyarn.application.id=application_1686577483648_0012 cd41d983c93d8eb906c9aa899dcdefd0
访问http://node01:9870/explorer.html#/logs/flink-job查看HDFS中的归档文件
等一段时间，几分钟后查看历史服务器
查看Job具体信息
