AppendEntries

AppendEntries is a remote procedure call used in the Raft consensus algorithm. It is sent by the leader to followers to replicate log entries and to maintain authority through heartbeats. Each AppendEntries message includes the leader’s term, the leader’s identifier, the index and term of the log entry immediately preceding new entries (prevLogIndex and prevLogTerm), a set of new log entries to store (entries), and the leader’s commit index (leaderCommit).

When a follower receives AppendEntries, it first compares terms. If the leader’s term is older than the follower’s current term, the follower rejects the request. If the term is current, the follower checks that the log contains an entry at prevLogIndex with term equal to prevLogTerm. If this check passes, the follower appends any new entries not already present and deletes any conflicting entries beyond prevLogIndex. The follower then updates its commitIndex to the minimum of leaderCommit and the index of the last new entry. If the log does not contain the required entry, the follower rejects the request.

The follower responds with its current term and a success flag. A successful response allows the leader to advance its own matchIndex and nextIndex for that follower. A failed response prompts the leader to adjust its indexing, typically by retrying with an earlier nextIndex. If the entries field is empty, AppendEntries serves as a heartbeat to maintain leadership authority.

AppendEntries is central to log replication and leader performance in Raft, ensuring that followers converge toward a consistent, replicated log and, therefore, a consistent state machine across the cluster.