[
https://issues.apache.org/jira/browse/KYLIN-5854?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17865838#comment-17865838
]
pengfei.zhan commented on KYLIN-5854:
-------------------------------------
h1. Design
1. After parsing the view to get the table, save the full name with
`catalogName` to `viewMapping`.
2. During the auto-refresh detection, parse the `viewMapping` file. If
`tableIdentity` contains more than two elements, use the first one as the
catalog parameter to determine if a refresh is needed.
3. When calculating whether to refresh, add a condition: if the request
contains a catalog parameter, use the corresponding catalog implementation
class to get the correct location.
4. Preserve the original case of the table (Iceberg tables are case sensitive)
when saving and fetching from `viewMapping`.
5. If the metadata catalog of Iceberg changes, it can be used as a basis for
updating the table data.
6. Determine if there is a schema change to decide whether to switch the
filesystem (e.g., `kylin` deployed on HDFS, Iceberg data source may be on S3).
> In a read/write cluster, snapshots are automatically refreshed to maintain
> the original behavior
> ------------------------------------------------------------------------------------------------
>
> Key: KYLIN-5854
> URL: https://issues.apache.org/jira/browse/KYLIN-5854
> Project: Kylin
> Issue Type: Bug
> Components: Query Engine
> Affects Versions: 5.0.0
> Reporter: pengfei.zhan
> Assignee: pengfei.zhan
> Priority: Major
> Fix For: 5.0.0
>
>
> In a read/write cluster, snapshots are automatically refreshed to maintain
> the original behavior
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
