Closed Hixon10 closed 6 months ago
It looks like that I am wrong, based on - https://github.com/ihippik/wal-listener/blob/ef2311852ef336250bad81630450d573eca33783/listener/listener.go#L234-L256
However, I still wonder, are there any well-known use-cases, when we can lose data.
Sorry for delay, I'll do some research on this. but in any case I should add some metrics with potential data problems
so, I added straightforward behavior - an error occurred while processing a message, then we forward it to the top and restart the service (without ack). And in this case we will get it again (at least once). I also added a metric by which you can evaluate the number of problems.
Hi there,
I want to confirm my understanding of this library. In case, when we read a new event from PostgreSQL replication log, but isn't able to write it to Kafka/NATS, we are going to lose this data. Hence, it is at-most-once delivery semantic. Is it correct, or am I wrong?