fresh-borzoni commented on code in PR #3260:
URL: https://github.com/apache/fluss/pull/3260#discussion_r3344934114
##########
fluss-spark/fluss-spark-common/src/main/scala/org/apache/fluss/spark/read/FlussBatch.scala:
##########
@@ -129,26 +130,58 @@ class FlussAppendBatch(
}
override def planInputPartitions(): Array[InputPartition] = {
- val bucketOffsetsRetrieverImpl = new BucketOffsetsRetrieverImpl(admin,
tablePath)
+ val maxRecordsPerPartition: Option[Long] = {
+ val value =
flussConfig.getLong(SparkFlussConf.SCAN_MAX_RECORDS_PER_PARTITION, 0)
+ if (value > 0) Some(value) else None
+ }
+
+ val bucketOffsetsRetrieverImpl = maxRecordsPerPartition match {
+ case Some(_) => new BucketOffsetsRetrieverImpl(admin, tablePath, true)
+ case _ => new BucketOffsetsRetrieverImpl(admin, tablePath)
+ }
val buckets = (0 until tableInfo.getNumBuckets).toSeq
+ def splitOffsetRange(
+ tableBucket: TableBucket,
+ startOffset: Long,
+ stopOffset: Long,
+ maxRecords: Long): Seq[InputPartition] = {
+ if (
Review Comment:
what about emoty tables and buckets?
##########
fluss-spark/fluss-spark-common/src/main/scala/org/apache/fluss/spark/read/FlussBatch.scala:
##########
@@ -91,26 +92,64 @@ class FlussAppendBatch(
}
override def planInputPartitions(): Array[InputPartition] = {
- val bucketOffsetsRetrieverImpl = new BucketOffsetsRetrieverImpl(admin,
tablePath)
+ val maxRecordsPerPartition: Option[Long] = {
+ val opt =
flussConfig.getOptional(SparkFlussConf.SCAN_MAX_RECORDS_PER_PARTITION)
+ if (opt.isPresent) Some(opt.get().longValue()) else None
+ }
+
+ val bucketOffsetsRetrieverImpl = maxRecordsPerPartition match {
+ case Some(_) => new BucketOffsetsRetrieverImpl(admin, tablePath, true)
+ case None => new BucketOffsetsRetrieverImpl(admin, tablePath)
+ }
val buckets = (0 until tableInfo.getNumBuckets).toSeq
+ def splitOffsetRange(
+ tableBucket: TableBucket,
+ startOffset: Long,
+ stopOffset: Long,
+ maxRecords: Long): Seq[InputPartition] = {
+ if (
+ startOffset < 0 || stopOffset <= startOffset || stopOffset <=
(startOffset + maxRecords)
Review Comment:
let's emit a warn or file an issue then 👍
##########
fluss-spark/fluss-spark-common/src/main/scala/org/apache/fluss/spark/SparkFlussConf.scala:
##########
@@ -50,4 +50,14 @@ object SparkFlussConf {
.durationType()
.defaultValue(Duration.ofMillis(10000L))
.withDescription("The timeout for log scanner to poll records.")
+
+ val SCAN_MAX_RECORDS_PER_PARTITION: ConfigOption[java.lang.Long] =
+ ConfigBuilder
+ .key("scan.maxRecordsPerPartition")
Review Comment:
now we have a mix of CamelCase and dot-hypen style. Shall we converge to one
format?
--
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]
