Cedar Dispatch cannot acknowledge writes with only one synchronized replica. Recover the missing copies while preserving the two-copy requirement.
Two brokers are down, but the remaining broker process is alive. The topic still requires two synchronized copies before acks=all writes can succeed. Investigate the original health evidence, restore one broker, let it catch up and test an actual write. Then restore the full replica set and state the recovery recommendation. The sequence distinguishes a live process from an in-sync replica. Lowering the minimum would remove the visible rejection while changing the guarantee. The model performs broker state changes and replication immediately when requested; it does not measure production recovery time. Preserve the original durability setting throughout this carried incident.
Inspect the topics, partition logs and group offsets before changing the cluster.