You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues-all@impala.apache.org by "Adam Holley (JIRA)" <ji...@apache.org> on 2018/10/31 15:48:00 UTC

[jira] [Assigned] (IMPALA-6385) Inconsistencies in alter table implementations

     [ https://issues.apache.org/jira/browse/IMPALA-6385?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Adam Holley reassigned IMPALA-6385:
-----------------------------------

    Assignee:     (was: Adam Holley)

> Inconsistencies in alter table implementations
> ----------------------------------------------
>
>                 Key: IMPALA-6385
>                 URL: https://issues.apache.org/jira/browse/IMPALA-6385
>             Project: IMPALA
>          Issue Type: Bug
>          Components: Frontend
>            Reporter: Adam Holley
>            Priority: Minor
>
> In CatalogOpExecutor, several of the alter statements have inconsistent behavior.  For example, if passing partition parameters, after some alter statements, the metadata for the partition is just marked dirty and will be reloaded on the next read.  If not using partition parameters, the table metadata and default partition metadata is reloaded. Specifically, in alterTableSetFileFormat, if the partitionSet is null, the method will return true which results in HdfsTable.load being called with loadPartitionFileMetadata true which reloads the metadata.  However if the partitionSet is not null, the metadata is not reloaded and instead the partition is instead just marked dirty (catalog_.getHdfsPartition(...).markDirty()).



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-all-unsubscribe@impala.apache.org
For additional commands, e-mail: issues-all-help@impala.apache.org