Davis-Zhang-Onehouse opened a new issue, #19725:
URL: https://github.com/apache/hudi/issues/19725

   HUDI-7494 (#10841) taught the meta-sync clients to persist and read 
`last_commit_completion_time_sync`, so incremental sync can register a commit 
that completes after the instant-time watermark has already moved past it (for 
example a long-running INSERT_OVERWRITE overtaken by a later-instant commit). 
If that mechanism regresses, the affected partition is silently and permanently 
dropped from the catalog: every later cycle either short-circuits on 
`isAlreadySynced` or resolves zero partitions and advances the watermark again. 
Nothing in the test suites fails today if the completion watermark stops being 
written, is computed with the wrong ordering, or is never consulted on read.
   
   Separately, `GlueTestUtil.createHoodieTable` writes its fixture commit into 
`.hoodie/`, while the version-8 timeline reads `.hoodie/timeline/`, so 
`TestAWSGlueSyncClient` runs every test against an empty active timeline 
without noticing.
   
   Both are addressed by adding sync-client coverage: falsifiable unit tests on 
`AWSGlueCatalogSyncClient`, an end-to-end `HiveSyncTool` test of the 
hollow-instant rescue, and a fixture that actually lands in the timeline folder.
   


-- 
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