Flink典型的任务处理过程如下所示:
Flink安装包下载地址:http://flink.apache.org/downloads.html ,选择对应Hadoop的Flink版本下载
Standalone 模式
快速入门教程地址:https://ci.apache.org/projects/flink/flink-docs-release-1.6/quickstart/setup_quickstart.html
1. 软件要求
- Java 1.8.x或更高版本,
- ssh(必须运行sshd才能使用管理远程组件的Flink脚本)
集群部署规划
节点 | master | worker | zookeeper |
master | master | zookeeper | |
slave1 | worker | zookeeper | |
slave2 | worker | zookeeper |
2. 解压
tar zxvf flink-1.7.1-bin-hadoop27-scala_2.11.tgz
3. 修改配置文件
[[email protected] conf]$ ls
flink-conf.yaml log4j-console.properties log4j-yarn-session.properties logback.xml masters sql-client-defaults.yaml
log4j-cli.properties log4j.properties logback-console.xml logback-yarn.xml slaves zoo.cfg
修改flink-conf.yaml
taskmanager.numberOfTaskSlots:2
jobmanager.rpc.address:master
可选配置:
- 每个JobManager(
jobmanager.heap.mb
)的可用内存量, - 每个TaskManager(
taskmanager.heap.mb
)的可用内存量, - 每台机器的可用CPU数量(
taskmanager.numberOfTaskSlots
), - 集群中的CPU总数(
parallelism.default
)和 - 临时目录(
taskmanager.tmp.dirs
)
3.1. HA配置文件
#jobmanager.rpc.address:master
high-availability:zookeeper #指定高可用模式(必须)
high-availability.zookeeper.quorum:master:2181,slave1:2181,slave2:2181 #ZooKeeper仲裁是ZooKeeper服务器的复制组,它提供分布式协调服务(必须)
high-availability.storageDir:hdfs:///flink/ha/ #JobManager元数据保存在文件系统storageDir中,只有指向此状态的指针存储在ZooKeeper中(必须)
high-availability.zookeeper.path.root:/flink #根ZooKeeper节点,在该节点下放置所有集群节点(推荐)
high-availability.cluster-id:/flinkCluster #自定义集群(推荐)
state.backend: filesystem
state.checkpoints.dir: hdfs:///flink/checkpoints
state.savepoints.dir: hdfs:///flink/checkpoints
修改conf/zoo.cfg
server.1=master:2888:3888
server.2=slave1:2888:3888
server.3=slave2:2888:3888
修改conf/masters
master:8081
修改slaves
slave1
slave2
4. 启动Hadoop
[[email protected] /]# start-dfs.sh
19/01/11 06:35:51 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
Starting namenodes on [master]
master: starting namenode, logging to /usr/local/hadoop/hadoop-2.8.3/logs/hadoop-root-namenode-master.out
slave1: starting datanode, logging to /usr/local/hadoop/hadoop-2.8.3/logs/hadoop-root-datanode-slave1.out
slave2: starting datanode, logging to /usr/local/hadoop/hadoop-2.8.3/logs/hadoop-root-datanode-slave2.out
master: starting datanode, logging to /usr/local/hadoop/hadoop-2.8.3/logs/hadoop-root-datanode-master.out
Starting secondary namenodes [0.0.0.0]
0.0.0.0: starting secondarynamenode, logging to /usr/local/hadoop/hadoop-2.8.3/logs/hadoop-root-secondarynamenode-master.out
19/01/11 06:36:14 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
5. 启动Zookeeper
[[email protected] /]# start-zookeeper-quorum.sh
Starting zookeeper daemon on host master.
Starting zookeeper daemon on host slave1.
Starting zookeeper daemon on host slave2.
6. 启动flink
[[email protected] /]# start-cluster.sh
Starting HA cluster with 1 masters.
Starting standalonesession daemon on host master.
Starting taskexecutor daemon on host slave1.
Starting taskexecutor daemon on host slave2.