Yesterday we saw an event that couple of Shards
in our Kinesis data streams were not processed at all.
Setup
We have a kinesis stream and we have a fargate service which is consuming the records from kinesis shards.
We have total 16 shards and 16 ECS tasks consuming from these shards.
What could be the possible root causes for stuck shards? What all things we should lookout for to debug this as we are seeing this issue continuously