Reputation: 18299
I have just started Oozie today, and I've noticed an inconsistent error while running Hive actions. When I run the same exact Oozie workflow, sometimes it succeeds, and other times it fails with this error for the Hive action:
FAILED: SemanticException Unable to fetch table <table_name>
1) What is the cause of this?
2) What is the work around?
Currently I just run the Oozie job again until it succeeds, which sometimes takes multiple tries.
I am using Cloudera CDH4, MRv1. I have manually installed as opposed to using Cloudera Manager. I have a three node cluster. The Master node contains the NameNode, SecondaryNameNode, JobTracker, and HMaster; in addition, the Master node contains the slave components: DataNode, TaskTracker, HRegionServer. The Two Slave nodes also contain those slave components.
On the MasterNode, I installed the hive-metastore package along with the hive package. I am using a MySQL database for my metastore db, not the Derby database. I was unable to get the hive-server2 package to work because I did not install the Zookeeper ensemble.
Please let me know if I may provide any more information that could lead to the diagnosis of this error.
Upvotes: 2
Views: 2919
Reputation: 549
Its quite a possibility that multiple instances of hive on different putty terminal are running.This answer is not in reference with Cloudera and may be just one reason for the error.
Upvotes: 0
Reputation: 139
Have you parameterized the oozie flow? If yes, then check the parameter value is not getting passed successfully.
Upvotes: 0