Slow Performance Connecting To Kinesis #28
Description
Appreciate this is likely environment related rather than code, but I am seeing our Spark app having intermittent delays of approximately 10 seconds when connecting to Kinesis.
24/09/08 16:20:09 INFO ShardConsumer: ShardConsumer init on StreamShard(kinesisstream-agentevents, shardId-000000000000), startingPosition: KinesisPosition(AFTER_SEQUENCE_NUMBER,
49654215856880772161764109510791531192601896088858787842, -1, true)
24/09/08 16:20:17 INFO KinesisV2PartitionReader: getNext emptyCnt 2 >= 2. Stop fetchNext.
I have tried all of the settings relating to batch size, but nothing seems to improve this. There is nothing in cloudtrail/cloudwatch about contention or quotas on the Kinesis topics and nothing obvious in debug level logging apart from a few of these:
logDebug(s"${streamShard} - millis behind latest: ${batch.millisBehindLatest}," +
Any help appreciated!