AGENT SKILL

understand-replicated_log

暂无简介

License 许可
MIT
最近更新
2026-10-09
Stars
★ 0
Forks
0

03 SKILL.md 正文

understand-replicated_log

CALL NUMBER: distributed_consensus_and_replication_protocols.replicated_log : deep_raft(6) DEFINITION: Ordered log of commands maintained across replicas; foundation of state machine replication.

Invoke this skill to understand replicated_log down to its primitives. The RELATIVE ROOT below is the least-fixed-point closure of everything it bundles from — the full import cone, grouped by the lib each prim comes from. Projected from a prover-typed KB (MAP/SWI-Prolog consistency gate): every reference below resolves.

THE RELATIVE ROOT (the import cone, by lib)

from ?

  • agreement (d3): The property that a set of distributed processes reach the same decision outcome on a given value. Uniform agreement requires all processes including those that later crash to decide identically; asymmetric agreement permits different decision classes for different process roles such as primary versus backup.

from deep_raft

  • a005 (d1): Uniform agreement: all processes including those that later fail must decide on the same value, stricter than ordinary agreement which only binds correct processes.
  • a007 (d3): Quorum in agreement: a subset of processes sufficient to guarantee agreement can be reached, typically intersecting any valid quorum to ensure overlap.
  • a008 (d3): Threshold in agreement: the minimum number of correct processes required to guarantee safety and liveness under given failure assumptions.
  • a009 (d4): Asymmetric agreement: one class of processes (e.g., primary) must agree while another class (e.g., backups) may reach different decisions, used in primary-backup protocols.
  • a002 (d4): Validity in agreement: the decided value must have been proposed by some process in the system, guaranteeing agreements are grounded in actual proposals.
  • a004 (d5): Termination in agreement: every correct process eventually decides some value, ensuring the protocol makes progress and does not stall indefinitely.

from distributed_consensus_and_replication_protocols

  • consensus (d2): Agreement on a single value among a set of distributed processes; the core problem consensus protocols solve.
  • fault_tolerance (d3): System's ability to continue operating correctly despite component failures.

CONSUMERS (what needs this)

a007, a008, append_entries, commit_index, follower, ft004, ft015, leader, multi_paxos, raft, smr, state_machine_replication


Projected from the distributed consensus and replication protocols KB (196 concepts / 187 relations) — consistency-typed by MAP; the facet list after the colon IS the cross-lib dependency web.

04 同仓库更多 Skills