You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hive.apache.org by "Naveen Gangam (Jira)" <ji...@apache.org> on 2020/05/06 22:26:00 UTC

[jira] [Created] (HIVE-23388) CTAS queries should use target's location for staging.

Naveen Gangam created HIVE-23388:
------------------------------------

             Summary: CTAS queries should use target's location for staging.
                 Key: HIVE-23388
                 URL: https://issues.apache.org/jira/browse/HIVE-23388
             Project: Hive
          Issue Type: Sub-task
          Components: HiveServer2
    Affects Versions: 4.0.0
            Reporter: Naveen Gangam
            Assignee: Naveen Gangam


In cloud based storage systems, renaming files across different root level buckets seem to be disallowed. The S3AFileSystem throws the following exception. This appears to be bug in S3FS impl.

Failed with exception Wrong FS s3a://hive-managed/clusters/env-xxxxx/warehouse-xxxx-xxxx/warehouse/tablespace/managed/hive/tpch.db/customer/delta_0000001_0000001_0000 -expected s3a://hive-external
2020-04-27T19:34:27,573 INFO  [Thread-6] jdbc.TestDriver: java.lang.IllegalArgumentException: Wrong FS s3a://hive-managed//clusters/env-xxxx/warehouse-xxxx-xxxx/warehouse/tablespace/managed/hive/tpch.db/customer/delta_0000001_0000001_0000 -expected s3a://hive-external

But we should fix our query plans to use the target table's directory for staging as well. That should resolve this issue and it is the right thing to do as well (in case there are different encryption zones/keys for these buckets).

Fix in HIVE-22995 probably changed this behavior.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)