概要描述

本文描述Yarn的ResourceManager均处于standby状态的解决方案

详细描述

问题描述

Manager页面配置服务重启YARN，卡在了ResourceManager上，后台 kubectl get pods -owide | grep yarn 可以查看到pod状态均为1/1 running

可以去查询两个resourcemanager服务器上/var/log/yarn1下的resource manager日志可以看到都是 Already in standby state

解决方案

1 进入zookeeper

未开启安全情况下

source TDH-Client/init.sh
export HADOOP_USER_NAME=zookeeper
export CLIENT_JVMFLAGS="-Djava.security.auth.login.config=/etc/zookeeper1/conf/jaas.conf -Dtranswarp.maintenance.only.mode=true"
TDH-Client/zookeeper/bin/zookeeper-client

开启安全情况下

source TDH-Client/init.sh
#kinit 认证zookeeper账号
klist -kt /etc/zookeeper1/conf/zookeeper.keytab
# 根据上面列出的principle 进行认证
kinit -kt /etc/zookeeper1/conf/zookeeper.keytab zookeeper/ddmss00@TDH
export CLIENT_JVMFLAGS="-Djava.security.auth.login.config=/etc/zookeeper1/conf/jaas.conf -Dtranswarp.maintenance.only.mode=true"
TDH-Client/zookeeper/bin/zookeeper-client -server [zookeeper一个角色的ip地址]:2181
rmr /yarn-leader-election

2 配置服务重启YARN