Joshua Kim

See the docs on [Record](https://docs.aws.amazon.com/kinesis/latest/APIReference/API_Record.html). In short, the blob is always encoded in base64, so you'll need to deserialize in order to reconstruct the original object.

It's possible, but it's not possible to determine with just the data in the logs. This could be due to: 1. A network-related error 2. Churn/high load in your KCL...

Looks ilke you also meant to remove L124?

I haven't touched/looked at this code-base in a while but I personally think a cleaner solution would be to add a configuration flag (`skipRecordsOnApplicationException=true` by default) that just allows the...

FWIW I don't maintain this library and the flag I'm referring to doesn't actually exist, it's just what I personally think would make sense from my understanding of how the...

You might want to take a heap dump of the process to see what's taking up so much memory, try adding this JVM flag `-XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=/path/to/your/heap/dump/file.hprof"`. I think it's more...

Are you using `WebIdentityTokenFileCredentialsProvider` (in the default credential chain)? There's some known memory-leaks in the current release of the java sdk - https://github.com/aws/aws-sdk-java-v2/issues/3270.

Until the AWS SDK guys fix this bug you'll probably have to fall back to another credentials provider as a temporary workaround. @gaddam1987 Can you try that and see if...

Fix in https://github.com/aws/aws-sdk-java-v2/pull/3440, this issue should be fixed upon the next SDK upgrade.

> Cannot find the shard given the shardId chgenvulgfjlejltgvglhecbucrihrcbbclfj