Csaba Ringhofer created IMPALA-13778:
----------------------------------------

             Summary: Recursive file listing should ignore more directories
                 Key: IMPALA-13778
                 URL: https://issues.apache.org/jira/browse/IMPALA-13778
             Project: IMPALA
          Issue Type: Bug
          Components: Catalog
            Reporter: Csaba Ringhofer


The paths to ignore is governed by ignored_dir_prefix_list (default= 
".,_tmp.,_spark_metadata").

Hive recently started writing tmp files to folders starting with _tmp instead 
of -tmp, leading to errors if Impala loads the table when these temporary files 
still exist.

Another prefix missing in ignored_dir_prefix_list is Impala's own tmp dir, 
_impala_insert_staging. This can only cause issues in non-partitioned tables as 
Impala writes this dir to the root folder of the table.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to