danny0405 commented on code in PR #18826:
URL: https://github.com/apache/hudi/pull/18826#discussion_r3426151437
##########
hudi-client/hudi-client-common/src/main/java/org/apache/hudi/metadata/HoodieBackedTableMetadataWriter.java:
##########
@@ -824,28 +824,54 @@ private Pair<Integer, HoodieData<HoodieRecord>>
initializeRecordIndexPartition(
dataMetaClient,
dataWriteConfig);
- // Initialize the file groups
- final int fileGroupCount = estimateFileGroupCount(records);
+ Pair<Integer, Integer> bounds = getRLIFileGroupCountBounds();
+ int minFileGroupCount = bounds.getLeft();
+ int maxFileGroupCount = bounds.getRight();
+
+ int fileGroupCount;
+ if (minFileGroupCount != maxFileGroupCount) {
+ // Estimate file group count based on record count read from base file
footer metadata.
+ // Avoids the expensive records.persist() + records.count() pass over
materialized records.
+ List<Pair<String, HoodieBaseFile>> partitionBaseFilePairs =
latestMergedPartitionFileSliceList.stream()
+ .filter(p -> p.getRight().getBaseFile().isPresent())
+ .map(p -> Pair.of(p.getLeft(), p.getRight().getBaseFile().get()))
+ .collect(Collectors.toList());
+ fileGroupCount = estimateFileGroupCountFromBaseFiles(
+ partitionBaseFilePairs, minFileGroupCount, maxFileGroupCount);
+ LOG.info("Estimated {} file groups from base file footer metadata",
fileGroupCount);
+ } else {
+ // min == max: skip estimation, use the fixed value directly.
+ fileGroupCount = minFileGroupCount;
+ LOG.info("Using user-configured file group count: {}", fileGroupCount);
+ }
+
LOG.info("Initializing record index with {} file groups.", fileGroupCount);
return Pair.of(fileGroupCount, records);
}
- private int estimateFileGroupCount(HoodieData<HoodieRecord> records) {
- int minFileGroupCount;
- int maxFileGroupCount;
+ /**
+ * Returns the (min, max) file group count bounds for RLI based on which RLI
variant is enabled.
+ */
+ private Pair<Integer, Integer> getRLIFileGroupCountBounds() {
if (dataWriteConfig.isRecordLevelIndexEnabled()) {
- minFileGroupCount =
dataWriteConfig.getRecordLevelIndexMinFileGroupCount();
- maxFileGroupCount =
dataWriteConfig.getRecordLevelIndexMaxFileGroupCount();
- } else {
- minFileGroupCount =
dataWriteConfig.getGlobalRecordLevelIndexMinFileGroupCount();
- maxFileGroupCount =
dataWriteConfig.getGlobalRecordLevelIndexMaxFileGroupCount();
- }
- Supplier<Long> recordCountSupplier = () -> {
- records.persist("MEMORY_AND_DISK_SER");
- long count = records.count();
- LOG.info("Initializing record index with {} mappings", count);
- return count;
- };
+ return Pair.of(dataWriteConfig.getRecordLevelIndexMinFileGroupCount(),
Review Comment:
in validateRecordIndex, we also got a count on the records, do we need to
call `records.persist("MEMORY_AND_DISK_SER")` from there?
--
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]
