Github user glaksh100 commented on a diff in the pull request:
https://github.com/apache/flink/pull/6300#discussion_r202156901
--- Diff:
flink-connectors/flink-connector-kinesis/src/main/java/org/apache/flink/streaming/connectors/kinesis/internals/ShardConsumer.java
---
@@ -330,4 +347,24 @@ private GetRecordsResult getRecords(String shardItr,
int maxNumberOfRecords) thr
protected static List<UserRecord> deaggregateRecords(List<Record>
records, String startingHashKey, String endingHashKey) {
return UserRecord.deaggregate(records, new
BigInteger(startingHashKey), new BigInteger(endingHashKey));
}
+
+ /**
+ * Adapts the maxNumberOfRecordsPerFetch based on the current average
record size
+ * to optimize 2 Mb / sec read limits.
+ *
+ * @param averageRecordSizeBytes
+ * @return adaptedMaxRecordsPerFetch
+ */
+
+ private int getAdaptiveMaxRecordsPerFetch(long averageRecordSizeBytes) {
--- End diff --
Makes sense. Done.
---
