简体繁体 English

我无法在 hadoop3.3 中启动 jobhistoryserver

[英]I cannot start jobhistoryserver in hadoop3.3

原文 2022-07-16 17:01:55 8 1 hadoop/ mapreduce/ hadoop-yarn

part of my shell script -- hdp.sh to start hadoop cluster :我的 shell 脚本的一部分 - hdp.sh 启动 hadoop 集群：

but jobhistoryserver not started:但 jobhistoryserver 未启动：

process 'JobHistoryServer' should be there进程“JobHistoryServer”应该在那里

and I found that jobhistoryserver could start alone util I excut : $ hdp.sh start我发现jobhistoryserver可以单独启动util我执行： $ hdp.sh start

process 'JobHistoryServer' just disappear进程“JobHistoryServer”就消失了

I don't know why.我不知道为什么。

mapreduce-site.xml mapreduce-site.xml

yarn-site.xml纱线站点.xml

1 个解决方案

The Job History Server is a Hadoop 1 process and doesn't need ran in Hadoop 2 or 3. Job History Server 是 Hadoop 1 进程，不需要在 Hadoop 2 或 3 中运行。

YARN's ResourceManager is the replacement. YARN 的 ResourceManager 是替代品。 If the resource manager or node manager don't start, then you need to look at the actual process logs, not jps.如果资源管理器或节点管理器没有启动，那么你需要查看实际的进程日志，而不是 jps。

Regarding your script, Hortonworks "HDP" also doesn't exist anymore, so it seems you're referring to old documentation and you should be following the latest installation processes , which do not mention the history server.关于您的脚本，Hortonworks“HDP”也不再存在，因此您似乎指的是旧文档，您应该遵循最新的安装过程，其中没有提及历史服务器。

I examined the script, in my opinion, it would be better if you asked by typing it, not as an image.我检查了脚本，在我看来，如果您通过键入而不是图像来询问会更好。

I suppose you must be add sleep 10 after the yarn.sh line.我想你必须在 yarn.sh 行之后添加 sleep 10 。