08-28-2017 03:29 AM
Hi, Dear all
The Hive I have is failing once in two or three days with numerous messages like the following one
2017-08-28 05:06:00,301 WARN org.apache.hadoop.hive.common.JvmPauseMonitor: [org.apache.hadoop.hive.common.JvmPauseMonitor$Monitor@3ce53f6a]: Detected pause in JVM or host machine (eg GC): pause of approximately 16290ms
.. from the /vaк/log/hive/....HIVESERVER...log
The cloudera manager initially shows this message as Hive service health is bad. Usually Hive keeps kicking for one more hour or so then it becomes 100% unresponsive until I reboot the entire Hive service.
I can't find any meaningfull error/warning messages in the log of the hive server. Everything look like usual, but the service is failing.
In addition it looks like the HIVEMETASTORE service is running OK on the same host at the same time.
Could anyone please point me where can I look up what's wrong?
08-28-2017 05:22 AM - edited 08-28-2017 05:24 AM
whats your hive / hiveserver2 version ?
what you have in those parameter
-XX:NewRatio= ? -XX:MaxHeapFreeRatio= ? -XX:MinHeapFreeRatio= ?
please refer this Jira
08-28-2017 07:16 AM
The Jira was looked through and looks like that is not the case.
We don't have a lot of partitions in the queries. The highest "Total number of paths:" i see in the logs is 667. I guess that's the affected partition count
Hive version is 1.1.0-cdh5.5.1
How do I look up the parameters using the ClouderaManager?