The connection to HMS seems to be working fine. Here are some things to consider: 1. create a table in Hive and run invalidate metadata in Impala. Does this work? 2. Check the impalad.INFO and catalogd.INFO for any exceptions/errors 3. Keep in mind that invalidate metadata is asynchronous, i.e. it returns immediately and then in the background the catalog objects are sent to the impalad nodes. This could take some time, so the question is if the db shows up eventually (after a few secs) or never.   Dimitris ... View more

Does it work if you create the database through Impala?  ... View more

I meant the query profile of query you run in Impala 2.0 by setting MEM_LIMIT=15G. Just post the section with the estimated/peak memory usage per node. Is it the same profile as the one you posted earlier?   Dimitris ... View more

Hi,    The explain is not really usefull for identifying the memory offenders. Can you post the output of profile (http://www.cloudera.com/content/cloudera/en/documentation/cloudera-impala/latest/topics/impala_explain_plan.html#perf_profile_unique_1) after you've run the query? That will shed some light on the actual memory requirements of each execution node in your query.   Thanks   Dimitris ... View more

Hi,    Quick question. Do you have statistics in the tables you're querying? It doesn't look like it. Can you run compute stats and see if the problem persists. Also, the log messages you posted are just warnings. The problem is not with the scan nodes but with the computation of the hash join.    Dimitris  ... View more

Hi,  Is the query that is failing running in isolation or are other workloads running at the same time? Can you post the profile information for that query and any relevant sections from the impalad log?   Dimitris ... View more