minihippo commented on a change in pull request #3173:
URL: https://github.com/apache/hudi/pull/3173#discussion_r770494052
##########
File path:
hudi-client/hudi-client-common/src/main/java/org/apache/hudi/execution/CopyOnWriteInsertHandler.java
##########
@@ -85,7 +85,9 @@ public void
consumeOneRecord(HoodieInsertValueGenResult<HoodieRecord> payload) {
handles.put(partitionPath, handle);
}
- if (!handle.canWrite(payload.record)) {
+ // If index requires a customized partitioner, cannot open a new file here
since it might violate
+ // the distribution required by index.
+ if (!handle.canWrite(payload.record) &&
!hoodieTable.getIndex().needCustomizedPartitioner()) {
Review comment:
> this is a little opaque. but get why you are doing this. I ll think of
better ways if possible
Bucket Index requires only one parquet file in a bucket. It‘s a bucket
layout limitation for writer, so i abstract the constraint and move it to the
storage layout
--
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]
