[
https://issues.apache.org/jira/browse/IMPALA-6994?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Pranay Singh updated IMPALA-6994:
---------------------------------
Comment: was deleted
(was: [~jeszyb] I have made the below change, that is currently in review, here
is a brief description of the change made.
The problem is that while inserting a new row to an unpartitioned HDFS
table,
or to an existing partition, the catalogd makes unecessary call to Hive
Meta
Store to do a getTable() and list the partitions of the table, when in
fact
only the file metadata for HDFS tables needs to be updated.
This extra call to HMS introduces a point of failure, we need to handle
for error scenario when Hive MetaStore crashes. This change removes the
extra call to Hive Meta Store for the case when a row is inserted to an
existing partition in HDFS table or when a row is added to unpartitioned
table. Thus, an optimization as it reduces the call to Hive Meta Store
during Update of Catalog)
> Avoid reloading a table's HMS data for file-only operations
> -----------------------------------------------------------
>
> Key: IMPALA-6994
> URL: https://issues.apache.org/jira/browse/IMPALA-6994
> Project: IMPALA
> Issue Type: Improvement
> Components: Catalog
> Affects Versions: Impala 2.12.0
> Reporter: Balazs Jeszenszky
> Assignee: Pranay Singh
> Priority: Major
>
> Reloading file metadata for HDFS tables (e.g. as a final step in an 'insert')
> is done via
> https://github.com/apache/impala/blob/branch-2.12.0/fe/src/main/java/org/apache/impala/service/CatalogOpExecutor.java#L628
> , which calls
> https://github.com/apache/impala/blob/branch-2.12.0/fe/src/main/java/org/apache/impala/catalog/HdfsTable.java#L1243
> HdfsTable.load has no option to only load file metadata. HMS metadata will
> also be reloaded every time, which is an unnecessary overhead (and potential
> point of failure) when adding files to existing locations.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-all-unsubscr...@impala.apache.org
For additional commands, e-mail: issues-all-h...@impala.apache.org
