Manually commit offset to prevent restart loop

Hello,

does someone know if it is possible to manually commit a specific offset for a consumer on a specific partition when running a streams topology in a service?

Background: We were wondering what we could do from operations perspective if our service is in a restart loop because it can’t process a message and we decide we want to skip it, because the service needs to keep processing other messages and implementing a fix would take too long.

Or is there even some general way which we didn’t find yet to catch all exceptions on a high level and prevent the service from re-reading the failed message at all?

Best regards,
Ramona