We are using the async operator to make API calls and sometimes, they fail. We would like to retry the same event again in some custom time intervals. The failure could also be because of some post-call validation, hence triggered from code. We found this: https://issues.apache.org/jira/browse/FLINK-11909 Not sure if this has been implemented. If not, is there a way out? Perhaps using an operator down the chain to push the events back to the async operator after a delay? Or simply push the events that need a retry back to the input stream after filtering?
0
votes
1 Answers
1
votes
At the moment, Flink does not support retriable asynchronous operations. The best solution at the moment is to implement the retry logic in your user code. You could take a look at FutureUtils.retry for an example how to do it naively.
