We are trying to configure dremio to work together with our kerberized cloudera stack (CDP 7.3.1) via HDFS. The connection works, we are able to browse the data lake created against it. We can click a parquet file open and see the data.
When we run a query on the test dataset, the workers make a whole bunch of connections with the namenode and then workers and finally fail with a
VALIDATION ERROR: Failure getting source
In the query profile (
attached (16.4 KB)
) we see that it is trying to create a __support. table with a store as “json”. Right below that it fails saying that it cannot find the “json” table.
So is this an error on dremio side? Or should we look for the issue on our cloudera cluster? If so, what technology is this (hive/impala/…) - we can’t seem to find any errors about this on cloudera side.
I hope that somebody can help me save the little hair I have remaining.