# 1. 安装ZooKeeper集群

# 解压安装
$ tar -xzf apache-zookeeper-3.8.0-bin.tar.gz -C /bigdata/app/
$ ln -s /bigdata/app/apache-zookeeper-3.8.0-bin /bigdata/app/zookeeper

# 配置zoo.cfg
$ cat > /bigdata/app/zookeeper/conf/zoo.cfg << 'EOF' tickTime=2000 initLimit=10 syncLimit=5 dataDir=/bigdata/fgdata/zookeeper clientPort=2181 server.1=fgedu-node1:2888:3888 server.2=fgedu-node2:2888:3888 server.3=fgedu-node3:2888:3888 autopurge.snapRetainCount=5 autopurge.purgeInterval=24 EOF # 创建myid文件 $ echo "1" > /bigdata/fgdata/zookeeper/myid # node1
$ echo “2” > /bigdata/fgdata/zookeeper/myid # node2
$ echo “3” > /bigdata/fgdata/zookeeper/myid # node3

# 启动ZooKeeper
$ /bigdata/app/zookeeper/bin/zkServer.sh start
$ /bigdata/app/zookeeper/bin/zkServer.sh status

ZooKeeper JMX enabled by default
Mode: leader

# 2. 配置HDFS HA

# 配置core-site.xml
$ cat > /bigdata/app/hadoop/etc/hadoop/core-site.xml << 'EOF'
fs.defaultFS
hdfs://fgeduns ha.zookeeper.quorum
fgedu-node1:2181,fgedu-node2:2181,fgedu-node3:2181
EOF

# 配置hdfs-site.xml
$ cat > /bigdata/app/hadoop/etc/hadoop/hdfs-site.xml << 'EOF'
dfs.nameservices
fgeduns dfs.ha.namenodes.fgeduns
nn1,nn2 dfs.namenode.rpc-address.fgeduns.nn1
fgedu-node1:8020 dfs.namenode.rpc-address.fgeduns.nn2
fgedu-node2:8020 dfs.namenode.http-address.fgeduns.nn1
fgedu-node1:9870 dfs.namenode.http-address.fgeduns.nn2
fgedu-node2:9870 dfs.namenode.shared.edits.dir
qjournal://fgedu-node3:8485;fgedu-node4:8485;fgedu-node5:8485/fgeduns dfs.client.failover.proxy.provider.fgeduns
org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider dfs.ha.fencing.methods
sshfence dfs.ha.fencing.ssh.private-key-files
/root/.ssh/id_rsa dfs.ha.automatic-failover.enabled
true dfs.journalnode.edits.dir
/bigdata/fgdata/journalnode
EOF

# 3. 启动JournalNode
$ hdfs –daemon start journalnode

# 4. 格式化NameNode
# 在node1执行
$ hdfs namenode -format

# 5. 启动NameNode(node1)
$ hdfs –daemon start namenode

# 6. 同步NameNode(node2)
$ hdfs namenode -bootstrapStandby

# 7. 启动NameNode(node2)
$ hdfs –daemon start namenode

# 8. 格式化ZKFC
$ hdfs zkfc -formatZK

# 9. 启动ZKFC
$ hdfs –daemon start zkfc

# 10. 启动DataNode
$ hdfs –daemon start datanode

# 11. 验证HA
$ hdfs haadmin -getAllServiceState

fgedu-node1:8020 active
fgedu-node2:8020 standby

# 1. 配置YARN HA

# 配置yarn-site.xml
$ cat > /bigdata/app/hadoop/etc/hadoop/yarn-site.xml << 'EOF'
yarn.resourcemanager.ha.enabled
true yarn.resourcemanager.cluster-id
fgeducluster yarn.resourcemanager.ha.rm-ids
rm1,rm2 yarn.resourcemanager.hostname.rm1
fgedu-node1 yarn.resourcemanager.hostname.rm2
fgedu-node2 yarn.resourcemanager.webapp.address.rm1
fgedu-node1:8088 yarn.resourcemanager.webapp.address.rm2
fgedu-node2:8088 yarn.resourcemanager.zk-address
fgedu-node1:2181,fgedu-node2:2181,fgedu-node3:2181 yarn.resourcemanager.recovery.enabled
true yarn.resourcemanager.store.class
org.apache.hadoop.yarn.server.resourcemanager.recovery.ZKRMStateStore yarn.nodemanager.aux-services
mapreduce_shuffle
EOF

# 2. 启动ResourceManager
$ yarn –daemon start resourcemanager

# 3. 启动NodeManager
$ yarn –daemon start nodemanager

# 4. 验证YARN HA
$ yarn rmadmin -getServiceState rm1
active

$ yarn rmadmin -getServiceState rm2
standby

# 5. 测试故障转移
# 手动切换
$ yarn rmadmin -transitionToStandby rm1
$ yarn rmadmin -transitionToActive rm2

# 验证状态
$ yarn rmadmin -getServiceState rm1
standby

$ yarn rmadmin -getServiceState rm2
active

# 1. 状态检查脚本
$ cat > /bigdata/scripts/ha/check_ha_status.sh << 'EOF' #!/bin/bash # check_ha_status.sh # from:www.itpux.com.qq113257174.wx:itpux-com # web: http://www.fgedu.net.cn LOG_FILE="/bigdata/logs/automation/ha_status.log" log_info() { echo "[$(date '+%Y-%m-%d %H:%M:%S')] INFO: $1" | tee -a $LOG_FILE; } # 检查ZooKeeper check_zookeeper() { log_info "检查ZooKeeper状态..." for node in fgedu-node1 fgedu-node2 fgedu-node3; do status=$(ssh $node "/bigdata/app/zookeeper/bin/zkServer.sh status 2>&1 | grep Mode”)
log_info “$node: $status”
done
}

# 检查NameNode HA
check_namenode_ha() {
log_info “检查NameNode HA状态…”
hdfs haadmin -getAllServiceState | while read line; do
log_info “NameNode: $line”
done
}

# 检查YARN HA
check_yarn_ha() {
log_info “检查YARN HA状态…”
rm1_state=$(yarn rmadmin -getServiceState rm1)
rm2_state=$(yarn rmadmin -getServiceState rm2)
log_info “RM1: $rm1_state”
log_info “RM2: $rm2_state”
}

# 主函数
main() {
log_info “========== HA状态检查 ==========”
check_zookeeper
check_namenode_ha
check_yarn_ha
log_info “========== 检查完成 ==========”
}

main
EOF

$ chmod +x /bigdata/scripts/ha/check_ha_status.sh
$ /bigdata/scripts/ha/check_ha_status.sh

[2026-04-08 10:00:00] INFO: ========== HA状态检查 ==========
[2026-04-08 10:00:01] INFO: 检查ZooKeeper状态…
[2026-04-08 10:00:02] INFO: fgedu-node1: Mode: leader
[2026-04-08 10:00:03] INFO: fgedu-node2: Mode: follower
[2026-04-08 10:00:04] INFO: fgedu-node3: Mode: follower
[2026-04-08 10:00:05] INFO: 检查NameNode HA状态…
[2026-04-08 10:00:06] INFO: NameNode: fgedu-node1:8020 active
[2026-04-08 10:00:06] INFO: NameNode: fgedu-node2:8020 standby
[2026-04-08 10:00:07] INFO: 检查YARN HA状态…
[2026-04-08 10:00:08] INFO: RM1: active
[2026-04-08 10:00:08] INFO: RM2: standby
[2026-04-08 10:00:08] INFO: ========== 检查完成 ==========

# 2. 故障转移脚本
$ cat > /bigdata/scripts/ha/failover.sh << 'EOF' #!/bin/bash # failover.sh # from:www.itpux.com.qq113257174.wx:itpux-com # web: http://www.fgedu.net.cn if [ "$#" -ne 2 ]; then echo "用法: $0 ”
echo “component: namenode|resourcemanager”
echo “action: status|switch”
exit 1
fi

COMPONENT=$1
ACTION=$2

case $COMPONENT in
namenode)
case $ACTION in
status)
hdfs haadmin -getAllServiceState
;;
switch)
current_active=$(hdfs haadmin -getAllServiceState | grep active | cut -d: -f1)
if [ “$current_active” = “fgedu-node1” ]; then
hdfs haadmin -failover nn1 nn2
else
hdfs haadmin -failover nn2 nn1
fi
hdfs haadmin -getAllServiceState
;;
esac
;;
resourcemanager)
case $ACTION in
status)
yarn rmadmin -getServiceState rm1
yarn rmadmin -getServiceState rm2
;;
switch)
current_active=$(yarn rmadmin -getServiceState rm1)
if [ “$current_active” = “active” ]; then
yarn rmadmin -transitionToStandby rm1
yarn rmadmin -transitionToActive rm2
else
yarn rmadmin -transitionToStandby rm2
yarn rmadmin -transitionToActive rm1
fi
yarn rmadmin -getServiceState rm1
yarn rmadmin -getServiceState rm2
;;
esac
;;
esac
EOF

$ chmod +x /bigdata/scripts/ha/failover.sh

# 查看状态
$ /bigdata/scripts/ha/failover.sh namenode status

fgedu-node1:8020 active
fgedu-node2:8020 standby

# 手动切换
$ /bigdata/scripts/ha/failover.sh namenode switch

fgedu-node2:8020 active
fgedu-node1:8020 standby