topic Re: SAS interface to hadoop issue in Administration and Deployment

SAS interface to hadoop issue

SangaviS — Wed, 17 Jul 2019 16:09:21 GMT

We are getting the following error when trying to connect to hadoop adp

ERROR: Execute error: org.apache.hive.service.cli.HiveSQLException: Error while compiling statement: FAILED: SemanticException Line

1:17 Invalid path ''/tmp/sasdata-2019-07-17-09-49-56-164-e-00001.dlv'': No files matching path

hdfs://hcadps01hann/tmp/sasdata-2019-07-17-09-49-56-164-e-00001.dlv

Can someone help resolve this ?

Re: SAS interface to hadoop issue

AnandVyas — Thu, 18 Jul 2019 05:43:44 GMT

Hi @SangaviS ,

Errors during SAS/Access Interface Hadoop connection can be very generic sometimes so it is best to work with hadoop admin team to resolve this kind of issues. In order to further troubleshoot I would suggest you to try below options.

1) Check if the connection is working out of SAS using beeline. If it's working then try to form libname statement using that.

2) Check if the user trying to connect hive has write access on /tmp HDFS file system.

3) Add extra trace logging options to the code. -> options sastrace=',,d,' sastraceloc=saslog nostsuffix;

sample can be found here: https://go.documentation.sas.com/?docsetId=engfedsrv&docsetTarget=n167fxh8cc9h2an198cln7b46vb9.htm&docsetVersion=9.4&locale=en#p1mno9fc4cnea8n1nm70ojheqqmg

-AV

Re: SAS interface to hadoop issue

SangaviS — Fri, 19 Jul 2019 16:34:37 GMT

Thank you! Mapping folder was not set up properly on the Hadoop end. It got resolved after setting it up.

Re: SAS interface to hadoop issue

SimonDawson — Tue, 23 Jul 2019 10:57:30 GMT

@SangaviS it would be awesome if you could document a little bit of what you did and tag your post as the answer to make it easier for others to find the solution in this thread if they have a similar error. Thanks.

Re: SAS interface to hadoop issue

SangaviS — Fri, 16 Aug 2019 12:35:45 GMT

The actual issue was that the jar files were not set up for this environment. The issue got resolved after setting them up. It was also missing a mapping folder on Hadoop end where the datasets would sit but that was not the actual root cause.