[jira] [Updated] (HIVE-6787) ORC+ACID assumes all missing buckets are in ACID structure

["Owen O'Malley (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HIVE-6787?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Owen O'Malley updated HIVE-6787:
--------------------------------

    Attachment: HIVE-6787.patch

This fixes the problem, adds a test case, and also makes a log message that was confusing users a debug level.

> ORC+ACID assumes all missing buckets are in ACID structure
> ----------------------------------------------------------
>
>                 Key: HIVE-6787
>                 URL: https://issues.apache.org/jira/browse/HIVE-6787
>             Project: Hive
>          Issue Type: Bug
>          Components: File Formats
>    Affects Versions: 0.13.0
>            Reporter: Gopal V
>            Assignee: Owen O'Malley
>            Priority: Blocker
>             Fix For: 0.13.0
>
>         Attachments: HIVE-6787.patch
>
>
> ORC+ACID creates ACID structure splits for all missing buckets in a table
> {code}
> java.lang.RuntimeException: java.io.IOException: java.io.IOException: Vectorization and ACID tables are incompatible.
> at org.apache.hadoop.hive.ql.io.orc.OrcInputFormat.getRecordReader(OrcInputFormat.java:996)
> 	at org.apache.hadoop.hive.ql.io.HiveInputFormat.getRecordReader(HiveInputFormat.java:240)
> 	... 15 more
> {code}
> The tables are normal ORC tables and are not using ACID structure at all.
> {code}
> @@ -539,7 +539,7 @@ public void run() {
>          for(int b=0; b < context.numBuckets; ++b) {
>            if (!covered[b]) {
>              context.splits.add(new OrcSplit(dir, b, 0, new String[0], null,
> -                               false, false, deltas));
> +                               isOriginal, false, deltas));
>            }
>          }
> {code}
> seems to fix the issue. [~owen.omalley], please confirm if that is what I should be doing.



--
This message was sent by Atlassian JIRA
(v6.2#6252)