如何有效调整HiveServer2进程的JVM堆内存大小？

爱被打了一巴掌 2022-05-10 08:54 981阅读 0赞

我的运行环境：JDK1.8， HIVE1.2 ，HADOOP2.7  
HiveServer2在以下简称HS2

[hadoop@tony_hdp_01 bin]$ pwd
    ${HIVE_HOME}/bin
    [hadoop@tony_hdp_01 bin]$ ll
    total 40
    -rwxr-xr-x 1 hadoop hadoop  881 Nov 10  2017 beeline
    drwxrwxr-x 3 hadoop hadoop 4096 Mar  6  2018 ext
    -rwxr-xrwx 1 hadoop hadoop 9838 Nov 10  2017 hive
    -rwxr-xr-x 1 hadoop hadoop 1900 Nov 10  2017 hive-config.sh
    -rwxr-xr-x 1 hadoop hadoop  885 Nov 10  2017 hiveserver2
    -rwxr-xr-x 1 hadoop hadoop  880 Nov 10  2017 hplsql
    -rwxr-xr-x 1 hadoop hadoop  832 Nov 10  2017 metatool
    -rwxr-xr-x 1 hadoop hadoop  884 Nov 10  2017 schematool

// hive这个脚本运行时的参数实际上调用的是 [hive-config.sh][]

[hadoop@tony_hdp_01 bin]$ grep --color -C 3 -iE "hive-config.sh" hive
    bin=`dirname "$0"`
    bin=`cd "$bin"; pwd`
    
    . "$bin"/hive-config.sh
    
    TMP_USER_DIR="/tmp/${USER}"
    STDERR="${TMP_USER_DIR}/stderr"

// 修改HADOOP\_HEAPSIZE为12288M即12G  
\[hadoop@tony\_hdp\_01 bin\]$ cp [hive-config.sh][] hive-config.sh\_20180927  
\[hadoop@tony\_hdp\_01 bin\]$ vim [hive-config.sh][]  
export HADOOP\_HEAPSIZE=$\{HADOOP\_HEAPSIZE:-12288\}

// 改完以后kill掉HS2：后台我的自定义守护hiveServer2的脚本auto\_restart\_hive2.sh会自动启动HS2  
// 但是第二个 -Xmx仍然是 -Xmx2g

[hadoop@tony_hdp_01 hadoop]$ ps aux | grep -v grep | grep --color -iE "auto_clear|auto_restart"
    hadoop     912  0.0  0.0 113252  1628 ?        S    Aug31   0:05 /bin/bash ./auto_clear_hung_jobs.sh
    hadoop    3421  0.1  0.0 113636  1924 ?        S    Sep26   2:57 /bin/sh /home/hadoop/wps_cloud/admin_op/auto/auto_restart_hive2.sh
    [hadoop@tony_hdp_01 hadoop]$ ps aux|grep -v grep | grep --color "HiveServer2"  | grep --color Xmx
    hadoop    7217 19.3  0.7 4184168 938496 ?      Sl   14:26   1:11 /usr/lib/jvm/java-1.8.0-openjdk-1.8.0.171-7.b10.el7.x86_64/bin/java -Xmx12288m -Djava.library.path=${HADOOP_HOME}/lib:${HADOOP_HOME}/lib/native -Djava.library.path=${HADOOP_HOME}/lib:${HADOOP_HOME}/lib/native -Dhadoop.log.dir=${HADOOP_HOME}/logs -Dhadoop.log.file=hadoop.log -Dhadoop.home.dir=${HADOOP_HOME} -Dhadoop.id.str= -Dhadoop.root.logger=INFO,console -Djava.library.path=${HADOOP_HOME}/lib/native -Dhadoop.policy.file=hadoop-policy.xml -Djava.net.preferIPv4Stack=true -Xmx2g -Dproc_hiveserver2 -Dlog4j.configurationFile=hive-log4j2.properties -Djava.util.logging.config.file=${HIVE_HOME}/conf/parquet-logging.properties -Djline.terminal=jline.UnsupportedTerminal -Dhadoop.security.logger=INFO,NullAppender org.apache.hadoop.util.RunJar ${HIVE_HOME}/lib/hive-service-2.3.2.jar org.apache.hive.service.server.HiveServer2

// 联想到hive命令是通过hadoop jar命令加上hive有关的jar包执行的，那么就去从hadoop有关的配置文件中找到 -Xmx=2g出现的配置文件，再修改那个配置文件即可：

[hadoop@tony_hdp_01 hadoop]$ pwd
    ${HADOOP_HOME}/etc/hadoop
    [hadoop@tony_hdp_01 hadoop]$ grep -r --color '\-Xmx2g' ${HADOOP_HOME}/etc/hadoop/
    ${HADOOP_HOME}/etc/hadoop/hadoop-env.sh:export HADOOP_CLIENT_OPTS="-Xmx2g $HADOOP_CLIENT_OPTS"

// 修改 $\{HADOOP\_HOME\}/etc/hadoop/hadoop-env.sh 中的HADOOP\_CLIENT\_OPTS值为12g

[hadoop@tony_hdp_01 hadoop]$ cp hadoop-env.sh hadoop-env.sh_20180927
    [hadoop@tony_hdp_01 hadoop]$ vim hadoop-env.sh  # 将HADOOP_CLIENT_OPTS=-Xmx12g 即可
    [hadoop@tony_hdp_01 hadoop]$ ps aux|grep -v grep | grep --color "HiveServer2"  | grep --color Xmx
    hadoop    7217 19.3  0.7 4184168 938496 ?      Sl   14:26   1:11 /usr/lib/jvm/java-1.8.0-openjdk-1.8.0.171-7.b10.el7.x86_64/bin/java -Xmx12288m -Djava.library.path=${HADOOP_HOME}/lib:${HADOOP_HOME}/lib/native -Djava.library.path=${HADOOP_HOME}/lib:${HADOOP_HOME}/lib/native -Dhadoop.log.dir=${HADOOP_HOME}/logs -Dhadoop.log.file=hadoop.log -Dhadoop.home.dir=${HADOOP_HOME} -Dhadoop.id.str= -Dhadoop.root.logger=INFO,console -Djava.library.path=${HADOOP_HOME}/lib/native -Dhadoop.policy.file=hadoop-policy.xml -Djava.net.preferIPv4Stack=true -Xmx2g -Dproc_hiveserver2 -Dlog4j.configurationFile=hive-log4j2.properties -Djava.util.logging.config.file=${HIVE_HOME}/conf/parquet-logging.properties -Djline.terminal=jline.UnsupportedTerminal -Dhadoop.security.logger=INFO,NullAppender org.apache.hadoop.util.RunJar ${HIVE_HOME}/lib/hive-service-2.3.2.jar org.apache.hive.service.server.HiveServer2

// kill掉HS2：后台我的自定义守护hiveServer2的脚本auto\_restart\_hive2.sh会自动启动HiveServer2

[hadoop@tony_hdp_01 hadoop]$ kill -9 7217
    [hadoop@tony_hdp_01 hadoop]$ ps aux|grep -v grep | grep --color "HiveServer2"  | grep --color Xmx
    hadoop   13601  109  0.1 14863400 175580 ?     Sl   14:33   0:04 /usr/lib/jvm/java-1.8.0-openjdk-1.8.0.171-7.b10.el7.x86_64/bin/java -Xmx12288m -Djava.library.path=${HADOOP_HOME}/lib:${HADOOP_HOME}/lib/native -Djava.library.path=${HADOOP_HOME}/lib:${HADOOP_HOME}/lib/native -Dhadoop.log.dir=${HADOOP_HOME}/logs -Dhadoop.log.file=hadoop.log -Dhadoop.home.dir=${HADOOP_HOME} -Dhadoop.id.str= -Dhadoop.root.logger=INFO,console -Djava.library.path=${HADOOP_HOME}/lib/native -Dhadoop.policy.file=hadoop-policy.xml -Djava.net.preferIPv4Stack=true -Xmx12g -Dproc_hiveserver2 -Dlog4j.configurationFile=hive-log4j2.properties -Djava.util.logging.config.file=${HIVE_HOME}/conf/parquet-logging.properties -Djline.terminal=jline.UnsupportedTerminal -Dhadoop.security.logger=INFO,NullAppender org.apache.hadoop.util.RunJar ${HIVE_HOME}/lib/hive-service-2.3.2.jar org.apache.hive.service.server.HiveServer2

// 结果：JVM -Xmx12g 生效！那么也需要在提供HiveServer2服务的另外一台主机上做同样的修改、重启HS2

[hadoop@tony_hdp_02 bin]$ cd ${HADOOP_HOME}/etc/hadoop
    [hadoop@tony_hdp_02 hadoop]$ cp hadoop-env.sh hadoop-env.sh_20180927
    [hadoop@tony_hdp_02 hadoop]$ ll hadoop-env.sh*
    -rw-rw-r-- 1 hadoop hadoop 16255 May 18 09:38 hadoop-env.sh
    -rw-rw-r-- 1 hadoop hadoop 16255 Sep 27 14:40 hadoop-env.sh_20180927
    [hadoop@tony_hdp_02 hadoop]$ vim hadoop-env.sh
    [hadoop@tony_hdp_02 hadoop]$ ps aux|grep -v grep |grep --color -iE "auto_clear|auto_restart"
    hadoop    3769  0.1  0.0 113380  1780 ?        S    Sep26   2:28 /bin/sh /home/hadoop/wps_cloud/admin_op/auto/auto_restart_hive2.sh
    hadoop   18840  0.0  0.0 113252  1648 ?        S    Aug28   0:06 /bin/bash auto_clear_hung_jobs.sh
    [hadoop@tony_hdp_02 hadoop]$ ps aux|grep -v grep|grep --color "HiveServer2" | grep --color Xmx12
    hadoop   32461 35.9  2.0 6558060 2661848 ?     Sl   14:21   7:08 /usr/lib/jvm/java-1.8.0-openjdk-1.8.0.171-7.b10.el7.x86_64/bin/java -Xmx12288m -Djava.library.path=${HADOOP_HOME}/lib:${HADOOP_HOME}/lib/native -Dhadoop.log.dir=${HADOOP_HOME}/logs -Dhadoop.log.file=hadoop.log -Dhadoop.home.dir=${HADOOP_HOME} -Dhadoop.id.str= -Dhadoop.root.logger=INFO,console -Djava.library.path=${HADOOP_HOME}/lib/native -Dhadoop.policy.file=hadoop-policy.xml -Djava.net.preferIPv4Stack=true -Xmx12g -Dproc_hiveserver2 -Dlog4j.configurationFile=hive-log4j2.properties -Djava.util.logging.config.file=${HIVE_HOME}/conf/parquet-logging.properties -Djline.terminal=jline.UnsupportedTerminal -Dhadoop.security.logger=INFO,NullAppender org.apache.hadoop.util.RunJar ${HIVE_HOME}/lib/hive-service-2.3.2.jar org.apache.hive.service.server.HiveServer2

// 重启hs2，可以看到hiveServer2使用的JVM堆大小(12G)已生效！

[hadoop@tony_hdp_02 hadoop]$ kill -9 32461
    [hadoop@tony_hdp_02 hadoop]$ ps aux|grep -v grep|grep --color "HiveServer2" | grep --color Xmx12
    hadoop   23756  170  0.7 14996400 1023800 ?    Sl   14:41   0:44 /usr/lib/jvm/java-1.8.0-openjdk-1.8.0.171-7.b10.el7.x86_64/bin/java -Xmx12288m -Djava.library.path=${HADOOP_HOME}/lib:${HADOOP_HOME}/lib/native -Dhadoop.log.dir=${HADOOP_HOME}/logs -Dhadoop.log.file=hadoop.log -Dhadoop.home.dir=${HADOOP_HOME} -Dhadoop.id.str= -Dhadoop.root.logger=INFO,console -Djava.library.path=${HADOOP_HOME}/lib/native -Dhadoop.policy.file=hadoop-policy.xml -Djava.net.preferIPv4Stack=true -Xmx12g -Dproc_hiveserver2 -Dlog4j.configurationFile=hive-log4j2.properties -Djava.util.logging.config.file=${HIVE_HOME}/conf/parquet-logging.properties -Djline.terminal=jline.UnsupportedTerminal -Dhadoop.security.logger=INFO,NullAppender org.apache.hadoop.util.RunJar ${HIVE_HOME}/lib/hive-service-2.3.2.jar org.apache.hive.service.server.HiveServer2

[hive-config.sh]: http://hive-config.sh