When Impala builds, it depends on Hadoop. More specifically, it links (in a
C++ sense) against 'libhdfs.so' and 'libhadoop.so' to enable reading from
HDFS during query execution. (It also does this for Kudu.) On the frontend,
it uses Maven to grab dependencies (especially HDFS, Hive, Sentry, HBase,
and Kudu) to compile the relevant Java code.
If you're building Impala from scratch, you'll want to build Impala against
the same versions of the components as you're running in your cluster.
On Mon, Apr 9, 2018 at 9:53 AM, Nikitin Artem <artem.niki...@advlab.io>
> Hi all. I’m trying to add impala to existing hadoop cluster. I’m using the
> instructions described at https://cwiki.apache.org/
> confluence/display/IMPALA/Building+Impala but I don’t understand, why
> these scripts downloading Hadoop cluster components (I’m already have a
> cluster). Please help.