Graceful step-down API + membership-change integration tests#8
Open
antondalgren wants to merge 2 commits into
Open
Graceful step-down API + membership-change integration tests#8antondalgren wants to merge 2 commits into
antondalgren wants to merge 2 commits into
Conversation
Wraps transfer_leadership with target selection: picks the voter peer with the highest match_index (most caught-up) and initiates the transfer. Returns the chosen target NodeID on success or nil if not leader, no eligible peers exist, or transfer_leadership rejects the target. Useful for rolling restarts — the application calls step_down before shutting down the leader, giving the new leader a head start instead of waiting out an election timeout. LavinMQ integration's most likely caller. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Four new tests in integration_spec.cr exercising single-server membership changes end-to-end via the manual-delivery harness: - scales out 1 → 3 nodes via add_server with learner auto-promotion: bootstrap node 1, add_server(2), drive until learner auto-promotes to voter via maybe_promote_learner, repeat for node 3. Asserts final 3-voter state and that a propose replicates to all three. - scales in 3 → 1 nodes via remove_server: 3-node cluster, remove_server(3), remove_server(2), asserts single-voter cluster still functional. - remove_server refuses to remove the last voter: bootstraps a single-voter cluster, asserts remove_server(self) returns false. - rejects concurrent membership changes while one is in flight: documents the single-voter quick-commit case (subsequent add_server succeeds immediately because the previous one already committed via self-ack). Two new file-level helpers: make_lone_node (empty-peer single node for scale-out scenarios) and drive_until (run tick + deliver cycles until a predicate becomes true or rounds elapse). 112 examples passing (108 prior + 4 new). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Two additions motivated by the upcoming LavinMQ leader-election integration.
Node#step_down```crystal
def step_down : NodeID?
```
Wraps
transfer_leadership(to:)with target selection: picks the voter peer with the highestmatch_indexand initiates the transfer. Returns the chosen NodeID on success,nilon any of: not leader, no other voters, transfer refused. Same best-effort semantics astransfer_leadership— caller pollsNode#leader_idto observe completion.Primary use case: graceful rolling restarts.
Membership-change integration tests
Four new tests in
spec/raft/integration_spec.cr. Previouslyadd_server/remove_server/promote_learnerwere covered only at the unit level; nothing exercised the full bootstrap → add → auto-promote → replicate flow.add_serverwith learner auto-promotion.remove_server.Plus two new file-level helpers:
make_lone_node(id)(empty-peer single node) anddrive_until(nodes, rounds, &predicate)(tick + deliver convergence loop).Test plan
crystal spec spec/raft/ -Dpreview_mt -Dexecution_context— 112 examples, 0 failures (108 prior + 6 new).🤖 Generated with Claude Code