Hadoop Name Node HA Hive failures

I have two namenodes in HA environment. And Hive configured to point to HA namespace. But intermediatly my Hive fails pointing to passive namenode giving below error even tough my active namenode is still in service. Kindly help me dig where the issues is. Even Zkfc logs dosent show any failover happening when Hive fails.

Couldn't set up IO streams; Host Details : local host is: "my node/10.10.11.6"; destination host is: "passive node":8020;

Answers


To prevent hive server from opening too many connections with namenode, we need to set ipc.client.connection.maxidletime to the default value of 10 seconds. By default, PHD will set this parameter to 1 hour in the core-site.xml which can cause out of memory errors on HiveServer2.

<property> <name>ipc.client.connection.maxidletime</name> <value>10000</value> </property>

Refer Below…

https://issues.apache.org/jira/browse/HIVE-6866 https://discuss.pivotal.io/hc/en-us/articles/201646766-How-to-Configure-HiveServer2-and-use-a-Beeline-Client-on-a-Pivotal-HD-Cluster


Need Your Help

Passing a C++ instance into Python outside of BOOST_PYTHON_MODULE macro

python c++ boost boost-python

I want to pass an instance of TestObj from the C++ code into python. The code posted here produces the error in cout: "No to_python (by-value) converter found for C++ type: class TestObj". If I mov...

In SQL Management studio, what is the overhead of displaying the result of a SELECT statement?

tsql ssms sqlperformance

When you perform a SELECT over a large table in SQL Management Studio by default, the result is displayed in a Grid table. As we can imagine, when the result set is million lines long, the data wil...