While testing kafka connect in distributed mode, we found that it works on standby mode rather than active active mode.
If we have a two member cluster, only one works at a time for HDFS sink connector and the other waits in standby. If the first one dies or is interrupted, then the second waiting one picks up the operation.
Does it support a way that both work together through different sink names for HDFS under one single cluster group id and if one goes down, flush operation is not interrupted at all as the second one is anyway running and pushing data to HDFS?
