Hadoop集群regionserver节点故障
全触点Hadoop集群出现regionserver频繁掉的现象,大部分掉的都是140的regionserver节点,查看日志后发现两处信息
![](https://img2022.cnblogs.com/blog/2599613/202202/2599613-20220222212740317-1473613885.png)
![](https://img2022.cnblogs.com/blog/2599613/202202/2599613-20220222212740325-760788297.png)
![](http://media.teamshub.com/10000/tm/2019/02/20/58c73c13-c584-4e9e-8422-4c3df6a0b63f/61b06a73-3491-4407-86ed-54ea74f3938d.png)
查看相关主机信息并无异常,日志信息显示由于与zk的session超时,查看MaxSessionTimeout、zookeeper.session.timeout这2个参数,发现参数设置过小
![](https://img2022.cnblogs.com/blog/2599613/202202/2599613-20220222212740116-756677811.png)
![](https://img2022.cnblogs.com/blog/2599613/202202/2599613-20220222212740264-1160247942.png)
建议调整参数
MaxSessionTimeout 调整为210000
MaxSessionTimeout 调整为210000
zookeeper.session.timeout 调整180000
20190219凌晨两点至三点,137及140两台主机的regionserver节点均故障,相关的nmon监控及regionserver日志信息请查看附件
20190219凌晨两点至三点,137及140两台主机的regionserver节点均故障,相关的nmon监控及regionserver日志信息请查看附件