Wednesday 1 October 2014

Solve HiveServer2 Health Bad in Configuration


If HiveServer2s Health becomes Bad in CDH5.1, change some default values through Cloudera Manger.

Some default values:
  • The ‘Automatically Restart Process’ property is disabled by default. When set, the role's process is automatically (and transparently) restarted in the event of an unexpected failure.
  • Cloudera Manager killed the HS2 process because the ‘Kill When Out of Memory’ property   for the role is enabled. When set, a SIGKILL signal is sent to the role process when java.lang.OutOfMemoryError is thrown.
  • the HS2 java heap size is set to the default value 256MB. 
  • Hive Metastore Server max heap size is set to the default 1GB.
  • Zookeeper's maxClientCnxns is set at the default of 60. 
1. Increase the size of the HS2 and Hive Metastore Server heap to 2GB:
CM > Hive > Configuration > HS2 > Resource Management > Java Heap Size of HiveServer2 
CM > Hive > Configuration > Hive Metastore Server > Resource Management > Java Heap Size of Hive Metastore Server in Bytes 

2. Enable auto-restart:
CM > Hive > Configuration > HS2 > Advanced > Automatically Restart Process
CM > Hive > Configuration > Hive Metastore Server > Advanced > Automatically Restart Process

3. Increase Zookeeper maxClientCnxns to 1000
CM > Zookeeper > Configuration > Server Default Group > Maximum Client Connections

No comments:

Post a Comment