About alex.behm

alex.behm · ‎05-10-2018

To debug wrong results, it's very helpful for us to get an Impala query profile of the query that returns wrong results. Would you be able to provide that to help us debug?

alex.behm · ‎05-10-2018

The IS NOT NULL predicate is not placed correcly in the execution plan, that's why the result are wrong. We've fied many similar issues, chances are this one is known and fixed. I will need to confirm. What version of Impala are you running? You might be able to work around the issue like this. Instead of "where company_name is not null" use "where coalesce(company_name, aais_company_number) is not null". You might need to "cast(aais_company_number as string)" I understand it's not pretty but hopefully it does the trick.

alex.behm · ‎05-07-2018

Yes to both.

alex.behm · ‎04-25-2018

This is expected to work. What's the view definition? Can you post the EXPLAIN plan of your query?

alex.behm · ‎04-17-2018

1. You need to add the appropriate PARTITION clause to your CREATE EXTERNAL TABLE. Looks like it should be PARTITIONED BY (`date` STRING) 2. You need to run ALTER TABLE RECOVER PARTITIONS to discover all the partitions for the table After that you should be good to go!

alex.behm · ‎04-11-2018

Looks like https://issues.apache.org/jira/browse/IMPALA-6454 which was recently fixed. The problem is with the casing of primary key or partition column names. Should work if you convert all column names to lower case in your CTAS.

alex.behm · ‎03-02-2018

Changing the log level will not work. You can disable query profile profile logging with the following impalad startup option: --log_query_to_file=false If successful, you should see "Query profile archival is disabled" in the impalad ERROR log.

alex.behm · ‎03-01-2018

Looks like a client display issue. Both Impala and Hive return a DOUBLE for round(). A DOUBLE cannot precisely represent 80.71 so I'm guessing that technically the results are the same, it's just that the Hive client "pretty prints" the results in a different way than Impala's. If you require precise rouding results it's recommended you use decimal: select round(cast(269586 / 334026 * 100 as decimal(17, 15)), 2); 80.71

alex.behm · ‎02-27-2018

Kudu has the capability to evaluate simple filters natively, e.g. using the primary index of a table, so Impala will push such filters directly to Kudu. More complex filters (e.g. those involving UDFs) are evaluated by Impala after receiving rows from Kudu. Impala clearly distinguishes the filters evaluated by Kudu and those by Impala in the explain plan.

alex.behm · ‎02-09-2018

Thanks for the report! I suspect you are hitting this issue: https://issues.apache.org/jira/browse/IMPALA-5186 As a workaround you can set the following query option: SET MT_DOP=0; Please let us know if that worked or not.

Online	Offline
Last Visited	‎05-10-2018 06:52 PM

Member Since	‎10-16-2013 11:04 AM
Last Visited	‎05-10-2018 06:52 PM
Posts	307
Kudos received	77

Cloudera Community

Re: External Table from Parquet folder returns emp...

Re: Impala SQL for KUDU does not work

Re: Impalad logs diskspace full

Re: Impala round function does not return expected...

Re: Is Impala a proces engine when I use kudu?

Re: Select in impala has different value with hive

Re: Impala - Is Not Null Doesn't Work for Views

Re: Impala - Is Not Null Doesn't Work for Views

Re: Impala - Is Not Null Doesn't Work for Views

Re: External Table from Parquet folder returns emp...

Re: Impala SQL for KUDU does not work

Re: Impalad logs diskspace full

Re: Impala round function does not return expected...

Re: Is Impala a proces engine when I use kudu?

Re: Impalad exit when execute 'compute stats' on ...