this version of libhadoop was built without snappy support. 解决方案

偏执的太偏执、 2022-09-06 14:20 101阅读 0赞

线上遇到该问题，hadoop checknative 显示是支持 Snappy 压缩的

[yarn@Centos hadoop]$ bin/hadoop checknative -a
    21/08/20 17:52:39 INFO conf.Configuration: Current hadoop.security.authentication = simple
    21/08/20 17:52:39 INFO bzip2.Bzip2Factory: Successfully loaded & initialized native-bzip2 library system-native
    21/08/20 17:52:39 INFO zlib.ZlibFactory: Successfully loaded & initialized native-zlib library
    Native library checking:
    hadoop:  true /home/yarn/software/hadoop-2.7.2/lib/native/libhadoop.so.1.0.0
    zlib:    true /lib64/libz.so.1
    snappy:  true /home/yarn/software/hadoop-2.7.2/lib/native/libsnappy.so.1
    lz4:     true revision:99
    bzip2:   true /lib64/libbz2.so.1
    openssl: false Cannot load libcrypto.so (libcrypto.so: cannot open shared object file: No such file or directory)!

对 Map 输出开启snappy 压缩的部分作业，出现以下错误， 但也有部分作业运行正常。由于近期进行了线上灰度升级过程，  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM3OTY0Mzc5_size_16_color_FFFFFF_t_70]  
再经观察凡是出现 this version of libhadoop was built without snappy support 的作业，AppMaster 都跑在了升级后的节点上，出问题的 Task 均在 旧版的节点上，而在新版的节点上的作业能跑成功。可以确定，MRAppMaster 在启动任务时添加了某个环境变量导致的。

在 launch\_container.sh 中找到  
![在这里插入图片描述][7ae875afdbd6444da77921152fcc655c.png]  
查看 MRAppMaster 添加该环境变量的代码，发现是由 mapreduce.admin.user.env 配置设置的该值。

解决方案  
方案一：提交作业的时候，添加 -Dmapreduce.admin.user.env=‘LD\_LIBRARY\_PATH=/home/yarn/software/hadoop/lib/native’

方案二：在集群中的 mapred-site.xml 中添加配置

<property>
          <name>mapreduce.admin.user.env</name>
          <value>LD_LIBRARY_PATH=/home/yarn/software/hadoop/lib/native</value>
    </property>

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzM3OTY0Mzc5_size_16_color_FFFFFF_t_70]: /images/20220829/e8403575dbf54f588beef94595e6596d.png
[7ae875afdbd6444da77921152fcc655c.png]: /images/20220829/70dfd92a423e4012a3a8bda00883aa77.png