[
https://issues.apache.org/jira/browse/TAJO-1673?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14653058#comment-14653058
]
ASF GitHub Bot commented on TAJO-1673:
--------------------------------------
Github user blrunner commented on the pull request:
https://github.com/apache/tajo/pull/626#issuecomment-127478487
I applied CatalogStore::addPartitions to DDLExecutor and implemented some
unit test cases.
When I implementing unit test cases, I found that drop partition operate on
different way than expected. I think that drop partition need to delete data of
table with `PURGE` option. But currently, when dropping partition on managed
table without `PURGE` option, all data had been deleted. So, I've fixed this
bug.
> Implement recover partitions
> ----------------------------
>
> Key: TAJO-1673
> URL: https://issues.apache.org/jira/browse/TAJO-1673
> Project: Tajo
> Issue Type: Sub-task
> Components: Catalog
> Reporter: Jaehwa Jung
> Assignee: Jaehwa Jung
>
> Tajo stores a list of partitions for each table in its CatalogStore. If,
> however, new partitions are directly added to HDFS, the CatalogStore will not
> be aware of these partitions unless the user runs ALTER TABLE table_name ADD
> PARTITION commands on each of the newly added partitions.
> However, users can run a CatalogStore check command with the repair table
> option:
> {code:xml}
> MSCK REPAIR TABLE <table_name>
> {code}
> For the reference, I've referenced hive msck command.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
