geyanggang opened a new pull request, #13021:
URL: https://github.com/apache/gravitino/pull/13021
### What changes were proposed in this pull request?
For Hive-catalog `alterTable` requests that contain only property or comment
changes, pass an `EnvironmentContext` with `DO_NOT_UPDATE_STATS=true` to the
Hive Metastore, so it does not recompute table statistics and therefore does
not access the table's storage location.
- Added an `alterTable(..., boolean skipStatsUpdate)` overload to
`HiveClient`
(the existing overload is kept as a default delegating with
`skipStatsUpdate=false`), `HiveClientImpl`, and `HiveShim`.
- `HiveShimV2` uses `alter_table_with_environmentContext`; `HiveShimV3` uses
the
catalog-aware `alter_table(..., EnvironmentContext)`. The context is built
internally so the shared client interface stays free of Hive-specific
types.
- `HiveCatalogOperations.alterTable` sets `skipStatsUpdate=true` only when
every
change is a set/remove property or update comment. Column changes and
renames
keep the previous behavior.
### Why are the changes needed?
A property-only or comment-only alter does not change table data, so the
metastore statistics recomputation (and the storage-location access it
triggers,
e.g. an HDFS `getFileInfo`) is unnecessary. When the filesystem is slow or
unavailable, such a lightweight alter can block for minutes on the server
side.
Fix: #13020
### Does this PR introduce _any_ user-facing change?
No. There is no API or property change. Property-only and comment-only alters
avoid an unnecessary metastore stats update; all other alter behavior is
unchanged.
### How was this patch tested?
- Added a unit test `TestHiveCatalogOperations#testCanSkipStatsUpdate`
covering
the decision logic (property/comment-only changes skip stats; column
changes
and renames do not; empty/null changes fall back to the default).
- Ran `./gradlew :catalogs:hive-metastore-common:test
:catalogs:catalog-hive:test -PskipITs`
and `spotlessApply`; all pass.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
