How Raft Commits a Log Entry
The lesson animates a three-node cluster. A client sends a command to the leader, the leader appends it to its log, and AppendEntries messages carry it to followers. The timeline highlights the majority condition and shows why commitment depends on Raft's term and log rules rather than any single acknowledgement.
The second half introduces a divergent follower log and shows the leader backing up to repair it. The conclusion separates safety from availability: without a quorum, the cluster cannot commit new entries. Companion material: partial-failure article, Raft trace, and the Raft paper.
Related articles
Partial Failure Changes the Design of a Distributed System
Understand why timeouts, retries, ordering, and consistency matter when remote machines can fail independently.
Consensus: How Nodes Agree Despite Failures
Why distributed systems need consensus, how Raft uses terms and quorums to elect a leader, and when the cost is worth paying.
Raft Uses Terms and Quorums to Choose a Leader
Consensus lets a group of processes agree on an ordered sequence of decisions despite specified failures.
New lessons by email
Get new articles and notes on the systems behind everyday software.
One technical dispatch per week. No noise.
Not started
Sign in to save your learning progress.