Struct paxakos::node_builder::NodeBuilderWithNodeIdAndCommunicator

pub struct NodeBuilderWithNodeIdAndCommunicator<I: Invocation, C: Communicator> { /* private fields */ }

Node builder with node id and communicator already set.

Implementations

impl<I, C> NodeBuilderWithNodeIdAndCommunicator<I, C>where
    I: Invocation,
    C: Communicator<Node = NodeOf<I>, RoundNum = RoundNumOf<I>, CoordNum = CoordNumOf<I>, LogEntry = LogEntryOf<I>, Error = CommunicationErrorOf<I>, Yea = YeaOf<I>, Nay = NayOf<I>, Abstain = AbstainOf<I>>,

pub fn with(
    self,
    starter: Starter<I>
) -> NodeBuilder<Core<I, C>, impl Finisher<Node = Core<I, C>>>

Starts the node as specified by the given starter.

pub fn without_state(
    self
) -> NodeBuilder<Core<I, C>, impl Finisher<Node = Core<I, C>>>

Starts the node without any state and in passive mode.

pub fn with_initial_state<S: Into<Option<FrozenStateOf<I>>>>(
    self,
    initial_state: S
) -> NodeBuilder<Core<I, C>, impl Finisher<Node = Core<I, C>>>

Starts a new cluster with the given initial state.

The round number will be zero.

pub fn resuming_from<S: Into<SnapshotFor<I>>>(
    self,
    snapshot: S
) -> NodeBuilder<Core<I, C>, impl Finisher<Node = Core<I, C>>>

Resume operation from the given snapshot.

Soundness

It is assumed that the given snapshot was yielded from the Final event of a clean shutdown and that the node hasn’t run in the meantime. As such the node will start in active participation mode. This is unsound if the assumptions are violated.

Use recovering_with to have a failed node recover.

pub fn recovering_with<S: Into<Option<SnapshotFor<I>>>>(
    self,
    snapshot: S
) -> NodeBuilder<Core<I, C>, impl Finisher<Node = Core<I, C>>>

Resume operation from the given snapshot.

The node will participate passively until it can be certain that it is not breaking any previous commitments.

pub fn recovering_without_state(
    self
) -> NodeBuilder<Core<I, C>, impl Finisher<Node = Core<I, C>>>

Resume operation without a snapshot.

The node will participate passively until it can be certain that it is not breaking any previous commitments.

pub fn joining_with<S: Into<Option<SnapshotFor<I>>>>(
    self,
    snapshot: S
) -> NodeBuilder<Core<I, C>, impl Finisher<Node = Core<I, C>>>

Commence operation from the given snapshot.

Soundness

This method assumes that the node is (re-)joining the Paxos cluster. Use recovering_with to have a failed node recover.

A node is considered to rejoin iff there is a previous round r such that this node

• was not considered a member of the cluster for r and
• it did not participate in any rounds since r.

pub fn joining_without_state(
    self
) -> NodeBuilder<Core<I, C>, impl Finisher<Node = Core<I, C>>>

Commence operation without a snapshot.

Soundness

This method assumes that the node is (re-)joining the Paxos cluster. Use recovering_with to have a failed node recover.

A node is considered to rejoin iff there is a previous round r such that this node

• was not considered a member of the cluster for r and
• it did not participate in any rounds since r.