Hello Flink community , We are currently working on a Flink job that consumes messages from RabbitMQ, with checkpointing configured to at-least-once mode.
In our job, we make external API requests to retrieve information. If the external api is down or a timeout is occured, we currently throw an exception to avoid acknowledging the message in RabbitMQ as we aim to replay it. However, this approach causes all tasks to be redeployed. I'm reaching out to inquire if there are alternative solutions available in Flink to avoid throwing an exception. We are interested in a method that allows us to instruct Flink not to acknowledge a message if a problem occurred during its processing. Any suggestions would be appreciated. Best regards,
