Hi. I've been noticing in my graphs that every so often I'm receiving
some anomalous spikes in data that contain almost twice the amount of
data that I'm expecting. I believe this has something to do with my
batches being double counted.

I am using an OpaqueTridentKafkaSpout. This supposedly guarantees
exactly once processing.
I am using DynamoDB as an external datastore.

Does anybody have any suggestions on how I can go about debugging this ?

Thanks

-- 
Raphael Hsieh

Reply via email to