rahil-c opened a new issue, #18850:
URL: https://github.com/apache/hudi/issues/18850

   Part of #18676. RFC-104 / [design 
PR](https://github.com/chrevanthreddy/hudi/pull/1).
   
   ## Scope
   
   Schema/registration plumbing only — no data flow yet. Lands the new MDT 
partition type so subsequent sub-tasks have something to write into.
   
   ## Tasks
   
   - Add `VECTOR_INDEX` enum constant in 
`hudi-common/src/main/java/org/apache/hudi/metadata/MetadataPartitionType.java` 
with:
     - `partitionPath = "vector_index_"` (multi-instance prefix, mirrors 
`secondary_index_` and `expr_index_`)
     - new `recordType` value
     - override `getPartitionPath(metaClient, indexName)` to suffix the user 
index name (see `SECONDARY_INDEX` lines ~234–243)
   - Add `PARTITION_NAME_VECTOR_INDEX_PREFIX = "vector_index_"` in 
`hudi-common/src/main/java/org/apache/hudi/metadata/HoodieTableMetadataUtil.java`
 near lines 210–212.
   - Add config knobs in 
`hudi-common/src/main/java/org/apache/hudi/common/config/HoodieMetadataConfig.java`
 (mirror `RECORD_INDEX_*` block, lines 229–370):
     - `hoodie.metadata.vector.index.enable` (default `false`)
     - `hoodie.metadata.vector.index.num.clusters` (default `256`)
     - `hoodie.metadata.vector.index.file.group.count.per.cluster` (default `1` 
— lets a cluster span N file groups)
     - `hoodie.metadata.vector.index.training.sample.size` (default `1_000_000` 
— caps KMeans training rows for large tables)
   
   ## Tests
   
   - Unit test in `TestMetadataPartitionType` confirming partition path 
derivation for `vector_index_myidx`.
   
   ## Out of scope
   
   Payload schema, file-group mapping, KMeans, write path — covered by 
follow-up sub-issues.
   


-- 
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]

Reply via email to