nevzheng opened a new pull request, #12259:
URL: https://github.com/apache/gravitino/pull/12259

   ### What changes were proposed in this pull request?
   
   > **Iceberg REST purge stack (Stack 30), review 1 of 8.** Depends on #12227. 
Files changed are cumulative until earlier stack layers land. Review only 
commit `4f3b4d857`. This PR is published as a draft for design alignment.
   
   - Keep provider retry policy inside Iceberg `FileIO` and its storage SDK.
   - Propagate a final bulk-delete failure to the existing durable job retry 
path.
   - For non-bulk `FileIO`, treat only a confirmed `NotFoundException` as 
already complete.
   - Drain submitted sibling batches before failure returns so shared `FileIO` 
is not closed early.
   
   **After this PR:** cleanup cannot report false success after a final 
provider failure.
   
   ### Why are the changes needed?
   
   Iceberg and provider SDKs already own in-call retry. Gravitino should 
preserve their final outcome
   and let the durable job state machine own cross-attempt retry.
   
   Part of #12258.
   
   ### Does this PR introduce _any_ user-facing change?
   
   No API change.
   
   ### How was this patch tested?
   
   Focused tests cover bulk failure, per-file missing paths, ordinary errors, 
sibling draining,
   interruption, and a real worker/store retry. `spotlessJavaCheck`,
   `TestIcebergCleanupFailureHandling`, and `TestIcebergCleanupManager` 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]

Reply via email to