prashantwason commented on code in PR #17462:
URL: https://github.com/apache/hudi/pull/17462#discussion_r2770697015
##########
hudi-client/hudi-client-common/src/main/java/org/apache/hudi/index/bloom/HoodieBloomIndex.java:
##########
@@ -137,6 +137,13 @@ private List<Pair<String, BloomIndexFileInfo>>
getBloomIndexFileInfoForPartition
List<String> affectedPartitionPathList) {
List<Pair<String, BloomIndexFileInfo>> fileInfoList = new ArrayList<>();
+ // Preload the partitions so that each parallel op does not have to
perform listing.
+ // This is only needed when the embedded timeline server is not enabled,
as TLS caches file listings.
+ if (!config.isEmbeddedTimelineServerEnabled()) {
+ hoodieTable.getHoodieView().sync();
Review Comment:
You're right, the `sync()` call is unnecessary here. The FileSystemView is
created fresh for each write operation with the current timeline, and index
operations are early in the write flow before any commits happen. I've removed
the `sync()` call - only keeping the preloading of base files for each
partition.
--
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]
