I am learning the consensus supported by Apache IoTDB. I wonder when the consensus protocol of data copy is set to RatisConsensus, when the cluster has 3 data nodes and the number of data copies is 3, because RatisConsensus has strong consistency, is it impossible to write data to the cluster when one of the datanodes goes wrong? Also, is IoTConsensus (also supported by IoTDB) the same for this question?
I checked the official website about consensus, but it did not answer availability under specific number of cluster datanodes. I need an answer for this specific circumstance.
At present, if more than half of the datanodes are normal, then the cluster can normally operate for the three consensus protocols supported by Apache IoTDB, that is to say, if two nodes went wrong, writing cannot be performed. However, IoTConsensus is self-developed by IoTDB and has higher availability, so in your scenario, even if only one datanode survives when using IoTConsensus, new data can also be inserted into the cluster.