rangareddy commented on issue #17394:
URL: https://github.com/apache/hudi/issues/17394#issuecomment-5351244380

   This issue was reviewed as part of the JIRA-migrated backlog triage 
(HUDI-9131).
   
   **Findings: cannot be executed yet - its dependency is only half delivered. 
Routing to the performance validation pass.**
   
   The two halves of this benchmark are in different states:
   
   - **"new File group reader"** - not ready either. The metadata table does 
read through `HoodieFileGroupReader` / `HoodieLsmFileGroupReader` on `master` 
(`HoodieBackedTableMetadata.java:617`, `:629`) via PR #13544, but only for key 
lookups - there is no full-file-slice read path. Tracked as #17393, which is 
still open.
   - **"w/ expression filters"** - not ready. #17392 (HUDI-9129) is only 
partially delivered: a predicate is plumbed through, but 
`HoodieBackedTableMetadata.java:603` still asserts `predicate instanceof 
Predicates.In` with the message "For Metadata Table Reuse, key filter should be 
based on full keys". General expression filters do not exist to benchmark.
   
   So the comparison described here - new FGR with expression filters versus 
the older file-slice-based MDT reads with filters - cannot be run as specified 
until both HUDI-9130 (#17393) and HUDI-9129 (#17392) land.
   
   Suggest marking this as blocked on #17393 and #17392 rather than leaving it 
unqualified in the backlog. When it is unblocked, the results belong in the 
ticket with the hardware and dataset stated, since there is no in-repo 
benchmark harness for this path today.
   
   Keeping this open.
   


-- 
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