So while that query should definitely work on the latest version of Impala,
2.7.0 is a pretty old version and the syntax has changed somewhat since
then, for example I believe in 2.7 we used "DISTRIBUTE BY" instead of
"PARTITION BY" for Kudu tables (see
https://issues.apache.org/jira/browse/IMPALA-4561)

Unfortunately, the documentation on our website only goes back to 2.8 (
http://impala.apache.org/impala-docs.html) and I'm not sure how easy it
would be to get the docs for 2.7 anymore. Probably the best solution is to
use a more recent version of Impala (we've made a ton of improvements since
2.7 that you would probably want to take advantage of anyways)


On Wed, Dec 11, 2019 at 9:48 AM l vic <[email protected]> wrote:

>  What version of Impala do you use?
>
> Impala v2.7.0-cdh5-IMPALA_KUDU-cdh5 (48f1ad3)
>
> Can you create tables with other formats?
>
> just "create table" works but i do not know what format the created table
> is... Any way to check?
>
> On Wed, Dec 11, 2019 at 11:46 AM Zoltán Borók-Nagy <
> [email protected]> wrote:
>
>> Hi,
>>
>> That statement should work. What version of Impala do you use?
>> Can you create tables with other formats?
>>
>> BR,
>>     Zoltan
>>
>>
>> On Wed, Dec 11, 2019 at 4:03 PM l vic <[email protected]> wrote:
>>
>>> If I create table from impala, how can i make sure i create table in
>>> kudu?
>>> example from
>>> https://kudu.apache.org/docs/kudu_impala_integration.html#kudu_impala_create_table
>>> CREATE TABLE my_first_table ( id BIGINT, name STRING, PRIMARY KEY(id) )
>>> PARTITION BY HASH PARTITIONS 16 STORED AS KUDU;
>>> If i try this statement in impala shell, sql errors out on "stored as":
>>>
>>> "ERROR: AnalysisException: Syntax error in line 1:
>>>
>>> stored as kudu"
>>>
>>> Does it create table in hdfs and expects file format on "stored as",
>>> such as TEXTFILE? if so, how can i explicitly force it to create table in
>>> kudu?
>>>
>>> Thank you.
>>>
>>

Reply via email to