Help Center/ MapReduce Service/ Troubleshooting/ Using Hive/ Health Status of a HiveServer or MetaStore Instance Is Concerning
Updated on 2023-11-30 GMT+08:00

Health Status of a HiveServer or MetaStore Instance Is Concerning

Symptom

The health status of the HiveServer or MetaStore instance is Concerning.

Cause Analysis

The HiveServer or MetaStore instance cannot be started properly. For example, when modifying the MetaStore/HiveServer GC parameter, you can view the startup log of the corresponding process, for example, the hiveserver.out(hadoop-omm-jar-192-168-1-18.out) file. The following exception occurs:

Error: Could not find or load main class Xmx2048M

The preceding information indicates that Xmx2048M is used as the startup parameter of the Java process instead of the JVM during the startup of the JVM. As shown in the following information, the hyphen (-) is deleted mistakenly.

METASTORE_GC_OPTS=Xms1024M Xmx2048M -DIgnoreReplayReqDetect 
-XX\:CMSFullGCsBeforeCompaction\=1 -XX\:+UseConcMarkSweepGC 
-XX\:+CMSParallelRemarkEnabled -XX\:+UseCMSCompactAtFullCollection 
-XX\:+ExplicitGCInvokesConcurrent -server -XX\:MetaspaceSize\=128M 
-XX\:MaxMetaspaceSize\=256M

Solution

Check the latest changes to detect incorrect settings.

METASTORE_GC_OPTS=Xms1024M -Xmx2048M -DIgnoreReplayReqDetect 
-XX\:CMSFullGCsBeforeCompaction\=1 -XX\:+UseConcMarkSweepGC 
-XX\:+CMSParallelRemarkEnabled -XX\:+UseCMSCompactAtFullCollection 
-XX\:+ExplicitGCInvokesConcurrent -server -XX\:MetaspaceSize\=128M 
-XX\:MaxMetaspaceSize\=256M