Hi! There are no acks for individual messages in Flink. All messages that an operator receives between two checkpoint barriers fail or succeed together.
That's why dropped messages in filters need no dedicated acks. If the next checkpoint barrier passes through the whole topology, the set of messages from before that checkpoint is "acked" as a whole. Greetings, Stephan On Tue, Jan 19, 2016 at 1:42 PM, Don Frascuchon <frascuc...@gmail.com> wrote: > Hello, > > Working with flink checkpointing i see the messages discarded by a filter > function are no acked for the checkpoint ack. > > How can i auto confirm those discarded messages? > The ack notification is trigged by a sink ? > > Thanks in advance! > >
