Skip to content

[HUDI-4601] Read error from MOR table after compaction with timestamp partitioning #6365

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
danny0405 merged 6 commits into from
Aug 17, 2022
Merged

[HUDI-4601] Read error from MOR table after compaction with timestamp partitioning #6365

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
86f98b8
read error from mor after compaction
Aug 11, 2022
02cbe72
fix checkstyle
Aug 11, 2022
4ed8bd8
update
Aug 14, 2022
3b63d1c
update
Aug 15, 2022
62209a9
update from review
Aug 16, 2022
800e7ef
rollback unnecessary change
Aug 16, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 8 additions & 3 deletions ...rce/hudi-flink/src/main/java/org/apache/hudi/table/format/cow/CopyOnWriteInputFormat.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -108,9 +108,14 @@ public void open(FileInputSplit fileSplit) throws IOException {
LinkedHashMap<String, String> partSpec = PartitionPathUtils.extractPartitionSpecFromPath(
fileSplit.getPath());
LinkedHashMap<String, Object> partObjects = new LinkedHashMap<>();
partSpec.forEach((k, v) -> partObjects.put(k, DataTypeUtils.resolvePartition(
partDefaultName.equals(v) ? null : v,
fullFieldTypes[fieldNameList.indexOf(k)])));
partSpec.forEach((k, v) -> {
DataType fieldType = fullFieldTypes[fieldNameList.indexOf(k)];
if (!DataTypeUtils.isDatetimeType(fieldType)) {
// date time type partition field is formatted specifically,
// read directly from the data file to avoid format mismatch or precision loss
partObjects.put(k, DataTypeUtils.resolvePartition(partDefaultName.equals(v) ? null : v, fieldType));
}
});

this.reader = ParquetSplitReaderUtil.genPartColumnarRowReader(
utcTimestamp,
Expand Down
11 changes: 8 additions & 3 deletions ...rce/hudi-flink/src/main/java/org/apache/hudi/table/format/mor/MergeOnReadInputFormat.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -299,9 +299,14 @@ private ParquetColumnarRowSplitReader getReader(String path, int[] requiredPos)
this.conf.getBoolean(FlinkOptions.HIVE_STYLE_PARTITIONING),
FilePathUtils.extractPartitionKeys(this.conf));
LinkedHashMap<String, Object> partObjects = new LinkedHashMap<>();
partSpec.forEach((k, v) -> partObjects.put(k, DataTypeUtils.resolvePartition(
defaultPartName.equals(v) ? null : v,
fieldTypes.get(fieldNames.indexOf(k)))));
partSpec.forEach((k, v) -> {
DataType fieldType = fieldTypes.get(fieldNames.indexOf(k));
if (!DataTypeUtils.isDatetimeType(fieldType)) {
// date time type partition field is formatted specifically,
// read directly from the data file to avoid format mismatch or precision loss
partObjects.put(k, DataTypeUtils.resolvePartition(defaultPartName.equals(v) ? null : v, fieldType));
}
});

return ParquetSplitReaderUtil.genPartColumnarRowReader(
this.conf.getBoolean(FlinkOptions.UTC_TIMEZONE),
Expand Down
29 changes: 23 additions & 6 deletions ...ink-datasource/hudi-flink/src/test/java/org/apache/hudi/table/ITTestHoodieDataSource.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1017,15 +1017,12 @@ void testAppendWrite(boolean clustering) {
}

@ParameterizedTest
@EnumSource(value = ExecMode.class)
void testWriteAndReadWithTimestampPartitioning(ExecMode execMode) {
// can not read the hive style and timestamp based partitioning table
// in batch mode, the code path in CopyOnWriteInputFormat relies on
// the value on the partition path to recover the partition value,
// but the date format has changed(milliseconds switch to hours).
@MethodSource("executionModeAndPartitioningParams")
void testWriteAndReadWithTimestampPartitioning(ExecMode execMode, boolean hiveStylePartitioning) {
TableEnvironment tableEnv = execMode == ExecMode.BATCH ? batchTableEnv : streamTableEnv;
String hoodieTableDDL = sql("t1")
.option(FlinkOptions.PATH, tempFile.getAbsolutePath())
.option(FlinkOptions.HIVE_STYLE_PARTITIONING, hiveStylePartitioning)
.partitionField("ts") // use timestamp as partition path field
.end();
tableEnv.executeSql(hoodieTableDDL);
Expand All @@ -1044,6 +1041,26 @@ void testWriteAndReadWithTimestampPartitioning(ExecMode execMode) {
+ "+I[id8, Han, 56, 1970-01-01T00:00:08, par4]]");
}

@Test
void testMergeOnReadCompactionWithTimestampPartitioning() {
TableEnvironment tableEnv = batchTableEnv;

String hoodieTableDDL = sql("t1")
.option(FlinkOptions.PATH, tempFile.getAbsolutePath())
.option(FlinkOptions.TABLE_TYPE, FlinkOptions.TABLE_TYPE_MERGE_ON_READ)
.option(FlinkOptions.COMPACTION_DELTA_COMMITS, 1)
.option(FlinkOptions.COMPACTION_TASKS, 1)
.partitionField("ts")
.end();
tableEnv.executeSql(hoodieTableDDL);
execInsertSql(tableEnv, TestSQL.INSERT_T1);

List<Row> rows = CollectionUtil.iterableToList(
() -> tableEnv.sqlQuery("select * from t1").execute().collect());

assertRowsEquals(rows, TestData.DATA_SET_SOURCE_INSERT);
}

@ParameterizedTest
@ValueSource(strings = {FlinkOptions.PARTITION_FORMAT_DAY, FlinkOptions.PARTITION_FORMAT_DASHED_DAY})
void testWriteAndReadWithDatePartitioning(String partitionFormat) {
Expand Down