I have installed CDH Hadoop hadoop-2.5.0-cdh5.3.2 and impala 2.1 from http://archive.cloudera.com/impala/ubuntu/precise/amd64/impala/pool/contrib/i/impala/ in a ubuntu 12.04 64 bit version .
I configured both hadoop and impala.
I want to use impala to query csv on hdfs directly .
My hadoop along with hdfs is perfectly up and running .
But whenever I am trying to make Impala up , I am getting the below error.
Failed on local exception: com.google.protobuf.InvalidProtocolBufferException: Message missing required fields: callId, status; Host Details : local host is: "localhost/127.0.0.1"; destination host is: "localhost":54310;
I understand that HADOOP 2 is using protobuf version 2.5 but impala that I have installed is using protobuf version 2.4 .
Please help me how sort out this problem .
How to install a Hadoop 2.x version working with Impala ???
Thanks !!!
I resolved the issue . The problem was due to protobuf version mismatch.
The Impala is using HDFS directly without taking the help of the mapreduce framework . So in order to talk with each other they need to use the common protobuf version , because the message format is different for different protobuf version .
Solution:-
Compile both impala and hadoop using same protobuf version .
OrUse a Hadoop/Hive/Impala for the same CDH version .