vinothchandar commented on code in PR #13489:
URL: https://github.com/apache/hudi/pull/13489#discussion_r2178808468
##########
hudi-common/src/main/java/org/apache/hudi/metadata/HoodieBackedTableMetadata.java:
##########
@@ -277,61 +268,291 @@ private Predicate
transformKeyPrefixesToPredicate(List<String> keyPrefixes) {
return Predicates.startsWithAny(null, right);
}
- @Override
- protected Map<String, HoodieRecord<HoodieMetadataPayload>>
getRecordsByKeys(List<String> keys, String partitionName) {
- if (keys.isEmpty()) {
- return Collections.emptyMap();
+ private HoodiePairData<String, HoodieRecord<HoodieMetadataPayload>>
doLookupWithMapping(
+ HoodieData<String> keys, String partitionName, List<FileSlice>
fileSlices, boolean isSecondaryIndex,
+ Option<SerializableFunction<String, String>> keyEncodingFn) {
+
+ final int numFileSlices = fileSlices.size();
+ if (numFileSlices == 1) {
+ TreeSet<String> distinctSortedKeys = new TreeSet<>(keys.collectAsList());
+ return lookupRecordsWithMapping(partitionName, new
ArrayList<>(distinctSortedKeys), fileSlices.get(0), !isSecondaryIndex,
keyEncodingFn);
}
- Map<String, HoodieRecord<HoodieMetadataPayload>> result;
+ SerializableBiFunction<String, Integer, Integer> mappingFunction =
HoodieTableMetadataUtil::mapRecordKeyToFileGroupIndex;
Review Comment:
by hash function : do we mean the sharding function that maps a secondary
index entry to the right shard/filegroup within the MDT partition?
>looking for unescaped impacts perf non-trivially.
any micro-benchmarks? relative to other costs of shuffling/writing.. we
think this is high impact still?
--
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]
