You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Worker 2 is assigned to partition 0,1. But worker 1 is still working meanwhile. So when worker 2 tries to write new version, it faces the conflict on partition 1. On conflict the worker 2, gets latest offsets from delta and seeks. But there's no stored offset for partition 0, so it does not seek it, however worker 2 already consumed some messages from partition 0, which result in their absence.
The text was updated successfully, but these errors were encountered:
The fix would be: If there's no stored offset for a partition in delta, we go into auto.offset.reset if it's earliest then seek to the beginning, if latest then to the end.
Worker 2 is assigned to partition 0,1. But worker 1 is still working meanwhile. So when worker 2 tries to write new version, it faces the conflict on partition 1. On conflict the worker 2, gets latest offsets from delta and seeks. But there's no stored offset for partition 0, so it does not seek it, however worker 2 already consumed some messages from partition 0, which result in their absence.
The text was updated successfully, but these errors were encountered: