<configuration> <property> <name>yarn.app.mapreduce.am.env</name> <value>HADOOP_MAPRED_HOME=/opt/apps/hadoop-3.1.1</value> </property> <property> <name>mapreduce.map.env</name> <value>HADOOP_MAPRED_HOME=/opt/apps/hadoop-3.1.1</value> </property> <property> <name>mapreduce.reduce.env</name> <value>HADOOP_MAPRED_HOME=/opt/apps/hadoop-3.1.1</value> </property> </configuration>
Tag Archives: Big data technology
Hive execution task report cannot find main class error
Reason: because my cluster is centos7 system, according to the data on the Internet, the buffer/cache caching mechanism of centos7 will continue to increase the number of buffers, which will occupy the memory and reduce the available memory of the program. As a result, the Java garbage collection mechanism recycles the yarnchild process, resulting in the failure to find the main class
solution: store the cache data in memory to disk, and then release the memory manually
It is found on the Internet that:
execute the sync command several times, and then execute it in turn:
echo 1 & gt/ proc/sys/vm/drop_ caches
echo 2 >/proc/sys/vm/drop_ caches
echo 3 >/proc/sys/vm/drop_ caches
However, when running the following statement to clear the cache, an error of permission denied is reported: – bash/proc/sys/VM/drop_ caches: Permission denied
sudo echo 1 >/proc/sys/vm/drop_ caches
sudo echo 2 >/proc/sys/vm/drop_ caches
sudo echo 3 >/proc/sys/vm/drop_ caches
sync
Bash refused to do so because of the redirection symbol “& gt;” It’s also bash’s order. Sudo only allows echo command to have root authority,
but not “& gt;” The command also has root permission, so bash will think that the command has no permission to write information.
resolvent:
“SH – C” command, which allows bash to execute a string as a complete command
sudo sh -c “echo 1 >/proc/sys/vm/drop_ caches”
sudo sh -c “echo 2 >/proc/sys/vm/drop_ caches”
sudo sh -c “echo 3 >/proc/sys/vm/drop_ caches”
Or
echo 1 | sudo TEE/proc/sys/VM/drop_ caches