n3nash commented on a change in pull request #1194: [HUDI-326] Add support to
delete records with only record_key
URL: https://github.com/apache/incubator-hudi/pull/1194#discussion_r363554374
##########
File path: hudi-spark/src/main/scala/org/apache/hudi/HoodieSparkSqlWriter.scala
##########
@@ -161,7 +162,17 @@ private[hudi] object HoodieSparkSqlWriter {
// Convert to RDD[HoodieKey]
val keyGenerator =
DataSourceUtils.createKeyGenerator(toProperties(parameters))
val genericRecords: RDD[GenericRecord] =
AvroConversionUtils.createRdd(df, structName, nameSpace)
- val hoodieKeysToDelete = genericRecords.map(gr =>
keyGenerator.getKey(gr)).toJavaRDD()
+ val hoodieKeysToDelete = if
(HoodieIndex.GLOBAL_INDICES.contains(parameters(HoodieIndexConfig.INDEX_TYPE_PROP)))
{
Review comment:
There is a NonPartitionedKeyGenerator that you can pass and use for
deletions. Since as a client you are aware of 2 facts a) Either you are using
GLOBAL_INDEX or not b) If this batch is for Deletions, you can configure the
hudi client accordingly with a correct partition generator ->
https://github.com/apache/incubator-hudi/blob/master/hudi-spark/src/main/java/org/apache/hudi/NonpartitionedKeyGenerator.java
May be we can extend this class or rename it to reflect the correct key
generation strategy for deletions to be explicit
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
