Skip to main content

Leadership Control

Kommander includes leadership-control support in the runtime and tests.

Stable Leader Waiting​

WaitForLeaderStableAsync waits until the same non-empty leader endpoint has remained stable for a minimum duration:

string stableLeader = await raft.WaitForLeaderStableAsync(
    partitionId: 1,
    minStableFor: TimeSpan.FromMilliseconds(500),
    cancellationToken: cancellationToken
);

This is useful when you do not want to react to a leader that is still flapping during startup or failover.

Step Down​

StepDownAsync asks the local leader to step down while keeping the node online as a follower:

RaftOperationStatus status = await raft.StepDownAsync(
    partitionId: 1,
    cancellationToken: cancellationToken
);

The runtime can notify other nodes with a step-down notice so a new election can start promptly.

Leadership Transfer​

TransferLeadershipAsync asks the local leader to hand leadership to a specific up-to-date follower:

RaftOperationStatus status = await raft.TransferLeadershipAsync(
    partitionId: 1,
    targetEndpoint: "node-b:2070",
    cancellationToken: cancellationToken
);

If the target is stale or unknown, the transfer can fail with a non-success RaftOperationStatus.

Heartbeat Suspension​

Two additional hooks exist for fault-injection tests:

  • SuspendHeartbeatsAsync
  • ResumeHeartbeatsAsync

These let tests pause and resume periodic outbound heartbeats for a partition without disabling all Raft traffic.

Scope​

These APIs are exposed on IRaft, but they are marked as advanced test hooks in the codebase. Treat them as deterministic-test and controlled-experiment tools, not normal application write APIs.