This blog deep dives into the famous consensus algorithm called Raft and how it helps in achieving consensus in a distributed system consisting of multiple nodes.
when the network partition is recovered, the uncommitted entries of the minority group will be discarded and the logs from the majority group leader node will be communicated to all the follower nodes in the system and thus consensus will be achieved again.
```
Whatever write requests would have come to minority group leader, they will be discarded.
Which justifies the definition of Network paritition,
during network partition, client might face the data loss.
One question here is during log replication what if majority of the followers fail to commit the entry. Does the leader also reverts in this case or this is all under a transaction boundary?
```
when the network partition is recovered, the uncommitted entries of the minority group will be discarded and the logs from the majority group leader node will be communicated to all the follower nodes in the system and thus consensus will be achieved again.
```
Whatever write requests would have come to minority group leader, they will be discarded.
Which justifies the definition of Network paritition,
during network partition, client might face the data loss.
One question here is during log replication what if majority of the followers fail to commit the entry. Does the leader also reverts in this case or this is all under a transaction boundary?