[
https://issues.apache.org/jira/browse/IMPALA-11813?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17652243#comment-17652243
]
Quanlong Huang commented on IMPALA-11813:
-----------------------------------------
[~rongchuan.jin] Thanks for reporting this!
The doc is stale. The behavior was changed from invalidation to refresh in
IMPALA-9101 since we found it more helpful in reality. Before we have
EventProcessor, users do AlterTable in Hive and manually run REFRESH in Impala
to sync the metadata. The EventProcessor just make the refresh automatic.
However, we'd better provide a flag to make this configurable. This will be a
good task for ramping up. Do you want to take it?
> Handling AlterTableEvent operation in Automatic Invalidate/Refresh of
> metadata should be invalidate rather than reload
> ----------------------------------------------------------------------------------------------------------------------
>
> Key: IMPALA-11813
> URL: https://issues.apache.org/jira/browse/IMPALA-11813
> Project: IMPALA
> Issue Type: Bug
> Components: Catalog
> Affects Versions: Impala 4.1.0, Impala 4.2.0, Impala 4.1.1
> Reporter: rongchuan.jin
> Priority: Major
>
> As the
> [docs|https://impala.apache.org/docs/build/html/topics/impala_metadata.html]
> said, metastoreEventProcessor invalidates the tables when it receives the
> ALTER TABLE event.
> But in fact, we still use reloadTable to handle alterTableEvent, see in
> [code|https://github.com/apache/impala/blob/branch-4.2.0/fe/src/main/java/org/apache/impala/catalog/events/MetastoreEvents.java#L1448]，which
> is inconsistent and may leads to unexpected metastore sync thread shutdown.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
