Details
-
Sub-task
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
Description
Although just implementing 1-3 will likely be a significant improvement, we will eventually want to tackle the case of an exception that affects all tasks on that thread. I think in this case we would just need to keep retrying the specific call that’s failing, whatever that may be. This would almost certainly require solution #3/4 as a prerequisite as we would need to keep retrying on that thread’s Producer. Of course, with KIP-572 we’ll already retry most kinds of errors on the Producer, if not all. So this would most likely only apply to a few kinds of exceptions, such as any Consumer calls that fail for reasons besides having dropped out of the group.