[GitHub] [iceberg] szehon-ho commented on a diff in pull request #5373: API, Spark: Update remove orphan files procedure to use bulk deletion if applicable

[GitBox <gi...@apache.org>]

szehon-ho commented on code in PR #5373:
URL: https://github.com/apache/iceberg/pull/5373#discussion_r934777491


##########
spark/v2.4/spark/src/main/java/org/apache/iceberg/spark/actions/BaseDeleteOrphanFilesSparkAction.java:
##########
@@ -182,12 +192,26 @@ private DeleteOrphanFiles.Result doExecute() {
     List<String> orphanFiles =
         actualFileDF.join(validFileDF, joinCond, "leftanti").as(Encoders.STRING()).collectAsList();
 
-    Tasks.foreach(orphanFiles)
-        .noRetry()
-        .executeWith(deleteExecutorService)
-        .suppressFailureWhenFinished()
-        .onFailure((file, exc) -> LOG.warn("Failed to delete file: {}", file, exc))
-        .run(deleteFunc::accept);
+    if (batchDeletionSize > 1) {
+      Preconditions.checkArgument(
+          table.io() instanceof SupportsBulkOperations,
+          "FileIO %s does not support bulk deletion",
+          table.io().getClass().getName());
+      SupportsBulkOperations bulkFileIO = (SupportsBulkOperations) table.io();
+      List<List<String>> fileBatches = Lists.partition(orphanFiles, batchDeletionSize);
+      Tasks.foreach(fileBatches)
+          .noRetry()
+          .executeWith(deleteExecutorService)
+          .suppressFailureWhenFinished()
+          .run(bulkFileIO::deleteFiles);

Review Comment:
   Was looking a bit with @dramaticlly at this earlier.  My concern here is the existing 'deleteFunc' plugin mechanism, which I know some users are using for dry-run purpose, ie printing out the list of files to delete instead of actually deleting them.  Im thinking:
   
   - do a precondition check, if the deleteFunc != defaultDeleteFunc.  If so, then dont allow bulkDelete
   - add a bulk deleteFunc?  
   @aokolnychyi any thoughts?



-- 
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: issues-unsubscribe@iceberg.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@iceberg.apache.org
For additional commands, e-mail: issues-help@iceberg.apache.org