Using data from S3 on local cluster

[Tim Bittersohl <ti...@innoplexia.com>]

Hi,

 

I do have an urgent problem with the Hive server in Java, reading data from
Amazon S3.

 

I try to read data from a table which content is stored on Amazon S3 space
and process it with a local cluster.

 

This data is accessed setting the following value of the configuration of
the server, when creating it in Java:

 

fs.defaultFS=s3://my-bucket

 

Commands using no Hadoop job to perform (like "SELECT * FROM testable") do
work that way. But when adding a where clause, so that a Hadoop job is
created on the local cluster, it fails immediately after initialization.

I suppose this is because the server tries to use the S3 space to perform
the job or create job information.

 

How can I solve this?

 

Thanks

Tim