lokeshj1703 commented on code in PR #19717:
URL: https://github.com/apache/hudi/pull/19717#discussion_r4037893237


##########
hudi-client/hudi-client-common/src/main/java/org/apache/hudi/table/action/index/RunIndexActionExecutor.java:
##########
@@ -227,8 +229,9 @@ private void abort(HoodieInstant indexInstant, Set<String> 
requestedPartitions)
       }
     });
 
-    // delete inflight instant
+    // delete the pending instant, both the inflight and the requested file
     
table.getMetaClient().reloadActiveTimeline().deleteInstantFileIfExists(instantGenerator.getIndexInflightInstant(indexInstant.requestedTime()));
+    
table.getMetaClient().getActiveTimeline().deleteInstantFileIfExists(instantGenerator.getIndexRequestedInstant(indexInstant.requestedTime()));

Review Comment:
   `updateTableConfigAndTimeline` now reloads the table config before adding to 
it, but `abort` (line 213) still reads `table.getMetaClient().getTableConfig()` 
directly, and the metadata writer builds its own meta client, so this copy can 
be stale here too. Removing the requested partitions works out the same either 
way, but a partition another writer completed during the action would be 
dropped from `hoodie.table.metadata.partitions` on the write-back — should 
`abort` take the same reload?



-- 
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