kafka-go
kafka-go copied to clipboard
WriteMessages can force metadata updates when occur NotLeaderForPartition error
WriteMessages will retry , when occur NotLeaderForPartition error, this mean the client's metadata are likely out of date.
Is it possible to force trigger metadata updates? Instead of relying only on metadataTTL for refresh
@zydovech this seems like a reasonable change but may be quite complex. Feel free to submit a PR with the fix if you have the time. If not a workaround may be to lower your metadata TTL