[IMPALA-8581] Switch hdfs commandline calls in dataload and tests to use "-d" on s3 where applicable

• Type: Bug
• Status: Resolved
• Priority: Minor
• Resolution: Duplicate
• Affects Version/s: Impala 3.3.0
• Fix Version/s: Not Applicable
• Component/s: Infrastructure
• Labels: newbie
• Epic Color: ghx-label-9
• Assignee: Unassigned
• Reporter: Joe McDonnell
• Created: 23/May/19 20:51
• Updated: 04/Dec/20 19:26
• Resolved: 04/Dec/20 19:26

Description

When doing a put or copyFromLocal via the HDFS commandline, it puts the file into a temporary file and then moves it into place. The "-d" option skips the temporary file and copies the file directly into place. On S3, this avoids an extra copy and can also avoid consistency issues. We should look into using the "-d" option across our tests. Our usages of the HDFS commandline generally don't require atomicity of this file copy, so using "-d" is unlikely to impact HDFS tests.

Issue Links

• duplicates: IMPALA-8950 Add -d and -f option to copyFromLocal and re-enable disabled S3 tests