topic Re: "Execute" in Cloudera in SAS Programming

"Execute" in Cloudera

VanDalucas — Fri, 20 Jul 2018 08:25:03 GMT

Hi Guys,

I am trying to connect to a Cloudera cluster and run a query there by using an execute statement:

libname sas2 hadoop
uri='.........'
schema="data_analyst"
SERVER="1***.**.**.***"
TRANSCODE_FAIL=SILENT
DBMAX_TEXT=50;

libname vantage hadoop
uri='.........'
schema="vantage"
SERVER="***.**.**.***"
TRANSCODE_FAIL=SILENT
DBMAX_TEXT=50;

proc sql;
connect to HADOOP
(
SERVER="***.**.**.***"
uri='......'
);
execute(
create table sas.T0 as
SELECT
A.var1, A.var2,
B.var3,B.var4,
FROM
sas2.data_set_1 as A left join vantage.data_set_2 as B on A.column1=B.column2) by hadoop;
disconnect from hadoop;
quit;

The problem is that (since I want to "execute" in Cloudera) I cannot use the name sas2 as a Libref for the first schema. I must give it the name "data_analyst" which is not a valid SAS Libname... . So I cannot access/use the dataset sas2.data_set_1 in the query above. Any solution other than renaming the Cloudera schema "data_analyst" to something less than 9 characters ?

Re: "Execute" in Cloudera

LinusH — Fri, 20 Jul 2018 12:48:05 GMT

Not sure I'm following. The length of a schema in an external database doesn't have to follow SAS naming conventions.
Anything that goes on inside an execute block is SAS unaware of.
If you wish to access that schema via a libref, you just give it a SAS libref name if max 8 chars, the referred schema can be much longer.

Re: "Execute" in Cloudera

VanDalucas — Fri, 20 Jul 2018 15:04:06 GMT

Hi LinusH and thank you for your help!

if I let "sas2" as the name of the "data_analyst" schema found in Cloudera, then upon using a dataset of that schema (sas2.dataset1) in an execute statement the following error is produced: "[Error 10072]: Database does not exist: sas2" or "["Error 10001]: Dataset does not exist: dataset1"

But these problems disappear when the name of the libref is the same as the name of the schema..

thanks again