You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@iceberg.apache.org by GitBox <gi...@apache.org> on 2021/12/14 05:03:15 UTC
[GitHub] [iceberg] xloya commented on a change in pull request #3724: [Spark][Core]: Support RewriteDataFiles when Files are Completely Eliminated by Deletes
xloya commented on a change in pull request #3724:
URL: https://github.com/apache/iceberg/pull/3724#discussion_r768259735
##########
File path: spark/v3.2/spark/src/test/java/org/apache/iceberg/spark/actions/TestRewriteDataFilesAction.java
##########
@@ -234,12 +236,220 @@ public void testBinPackWithDeletes() throws Exception {
Assert.assertEquals("7 rows are removed", total - 7, actualRecords.size());
}
+ @Test
+ public void testBinPackWithDeleteAllDataAndDataFileHasGroupOffsets() {
+ Map<String, String> options = new HashMap<>();
+ options.put(TableProperties.FORMAT_VERSION, "2");
+ options.put(TableProperties.PARQUET_ROW_GROUP_SIZE_BYTES, "1024");
+ options.put(TableProperties.PARQUET_PAGE_SIZE_BYTES, "256");
+ options.put(TableProperties.PARQUET_DICT_SIZE_BYTES, "512");
+ Table table = createTablePartitioned(1, 1, options);
+ shouldHaveFiles(table, 1);
+ table.refresh();
+
+ CloseableIterable<FileScanTask> tasks = table.newScan().planFiles();
+ List<DataFile> dataFiles = Lists.newArrayList(CloseableIterable.transform(tasks, FileScanTask::file));
+ int total = (int) dataFiles.stream().mapToLong(ContentFile::recordCount).sum();
+
+ RowDelta rowDelta = table.newRowDelta();
+ // remove all data
+ writePosDeletesToFile(table, dataFiles.get(0), total)
+ .forEach(rowDelta::addDeletes);
+
+ rowDelta.commit();
+ table.refresh();
+
+ AssertHelpers.assertThrows("Expected an exception",
+ RuntimeException.class,
+ () -> actions().rewriteDataFiles(table)
+ .option(BinPackStrategy.MIN_FILE_SIZE_BYTES, "0")
+ .option(RewriteDataFiles.TARGET_FILE_SIZE_BYTES, Long.toString(Long.MAX_VALUE - 1))
+ .option(BinPackStrategy.MAX_FILE_SIZE_BYTES, Long.toString(Long.MAX_VALUE))
+ .option(BinPackStrategy.DELETE_FILE_THRESHOLD, "1")
+ .option(RewriteDataFiles.USE_STARTING_SEQUENCE_NUMBER, "false")
+ .execute());
+ }
+
+ @Test
+ public void testBinPackWithDeleteAllDataAndDataFileHasGroupOffsetsWithSeqNum() {
+ Map<String, String> options = new HashMap<>();
+ options.put(TableProperties.FORMAT_VERSION, "2");
+ options.put(TableProperties.PARQUET_ROW_GROUP_SIZE_BYTES, "1024");
+ options.put(TableProperties.PARQUET_PAGE_SIZE_BYTES, "256");
+ options.put(TableProperties.PARQUET_DICT_SIZE_BYTES, "512");
+ Table table = createTablePartitioned(1, 1, options);
+ shouldHaveFiles(table, 1);
+ table.refresh();
+
+ CloseableIterable<FileScanTask> tasks = table.newScan().planFiles();
+ List<DataFile> dataFiles = Lists.newArrayList(CloseableIterable.transform(tasks, FileScanTask::file));
+ int total = (int) dataFiles.stream().mapToLong(ContentFile::recordCount).sum();
+
+ RowDelta rowDelta = table.newRowDelta();
+ // remove all data
+ writePosDeletesToFile(table, dataFiles.get(0), total)
+ .forEach(rowDelta::addDeletes);
+
+ rowDelta.commit();
+ table.refresh();
+
+ AssertHelpers.assertThrows("Expected an exception",
+ RuntimeException.class,
Review comment:
According to my current implementation method, it is actually unable to handle the situation that multiple row groups of a parquet data file may be split into multiple tasks when reading, so I hope to keep throwing exceptions first. If we can be sure that the data file in a single rewrite group must be a complete file when rewriting, then I think we can directly remove it without using `split_offsets` to judge
--
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