Created on 07-12-2017 07:15 AM - edited 09-16-2022 04:55 AM
We are facing the CLI Error intermittently, when parsing thru Impala(Both Parquet/Standard) and Hive tables using SAS ODBC Connection.
Impala Error:
ERROR: CLI cursor fetch error: [Cloudera][ImpalaODBC] (120) Error while retrieving data from in
Impala: [HY000] : RPC Error: SSL_read: Resource temporarily unavailable
Hive Error:
ERROR: CLI cursor fetch error: [Cloudera][Hardy] (35) Error from server: error code: '0' error
message: 'Invalid OperationHandle: OperationHandle [opType=EXECUTE_STATEMENT,
getHandleIdentifier()=aef23648-028f-4252-9769-97b4997f7d2d]'.
Any help is higly appriciated!!
Created 07-12-2017 12:12 PM
Created 07-12-2017 01:04 PM
Hi Sravan,
code is pretty straight farward with simple libname and proc sql statements :
libname impala odbc dsn="XXXXXXX" user=XXXXXXX password='XXXXX' schema=XXXX;
libname tgtlib "XXXXX";
proc sql;
create table tgtlib.xyz as
select * from impala.xyz;
quit;
Created 07-12-2017 01:19 PM
Do you have kerberos enabled in your environment?
Also pls check that your issue is related to the one mentioned in this link
https://issues.apache.org/jira/browse/IMPALA-5020
Created 07-12-2017 01:31 PM
Saranvisa,
I have active ticket from kerberos.
proc sql (executing from PC SAS) runs for about 15-20min and then it fails with above error saying resource unavaialble.
fyi, this is a huge table with 4k+ columns and 2 million+ records.
Tried with filtering data to different chunks and it works fine, but issue is I get duplicate records in each indivudual chuck i.e. count mismtach with source.
Created 07-12-2017 01:53 PM
Can you execute your query directly in Impala/Hive?
it may need detailed analysis, becuase
1. If your answer is yes for my above question, the similar issue "Resource temporarily unavailable error message from Impala shell " has already been discussed in this link
2. if your answer is no then it might be an issue with ODBC (or) your network capacity issue for the table with huge size.