0

If I develop my map-reduce logic with a older hadoop lib version like 0.20.0 and then crated a jar file and run the jar file on a cluster on which the cluster version is higher like 1.0.3 then what will happen? Does the cluster uses the same version of hadoop that compiled my code with or does it ignore what I have compiled it with? Would you please explain how this process work on hadoop in general?

I know that it will run but I see some logs which looks to be the older version of hadoop.common was used. The jar file does not include the lib also.

4

1 回答 1

0

首先，您需要从胖 jar 中排除 hadoop-core 库，hadoop 提供了这一点。

我认为会发生什么，如果您使用了任何 API 已从版本更改为版本的库，您将收到一个异常指示。

理论上，如果 API 没有改变，它应该可以工作，尽管不能保证。最好在同一个版本上编译和运行。

于 2012-09-20T16:28:14.367 回答