There was an issue raised in #131. Doing further investigation I have found:
With EnablePartitionEof set to true, when the partition ended, SMB was committing too advanced a partition index.
The SMB-driven manual commit with the CheckpointAfter(TimeSpan) might not happen after the desired time when there are no protocol events that happen (partition assignment or partition EOF).
Moreover, the Confluent.Kafka library has being upgraded from 2.3.0 to 2.5.2.
There was an issue raised in #131. Doing further investigation I have found:
EnablePartitionEof
set totrue
, when the partition ended, SMB was committing too advanced a partition index.CheckpointAfter(TimeSpan)
might not happen after the desired time when there are no protocol events that happen (partition assignment or partition EOF).Moreover, the Confluent.Kafka library has being upgraded from 2.3.0 to 2.5.2.