sync

package
v1.0.0-wilson Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: May 27, 2024 License: BSD-3-Clause Imports: 33 Imported by: 0

README

sync package

Overview

This package implements a client and server that allows for the syncing of a MerkleDB. The servers have an up to date version of the database, and the clients have an out of date version of the database or an empty database.

It's planned that these client and server implementations will eventually be compatible with Firewood.

Messages

There are four message types sent between the client and server:

  1. SyncGetRangeProofRequest
  2. RangeProof
  3. SyncGetChangeProofRequest
  4. SyncGetChangeProofResponse

These message types are defined in avalanchego/proto/sync.proto. For more information on range proofs and change proofs, see their definitions in avalanchego/merkledb/proof.go.

SyncGetRangeProofRequest

This message is sent from the client to the server to request a range proof for a given key range and root hash. That is, the client says, "Give me the key-value pairs that were in this key range when the database had this root." This request includes a limit on the number of key-value pairs to return, and the size of the response.

RangeProof

This message is sent from the server to the client in response to a SyncGetRangeProofRequest. It contains the key-value pairs that were in the requested key range when the database had the requested root, as well as a proof that the key-value pairs are correct. If a server can't serve the entire requested key range in one response, its response will omit keys from the end of the range rather than the start. For example, if a client requests a range proof for range [requested_start, requested_end] but the server can't fit all the key-value pairs in one response, it'll send a range proof for [requested_start, proof_end] where proof_end < requested_end, as opposed to sending a range proof for [proof_start, requested_end] where proof_start > requested_start.

SyncGetChangeProofRequest

This message is sent from the client to the server to request a change proof between the given root hashes. That is, the client says, "Give me the key-value pairs that changed between the time the database had this root and that root." This request includes a limit on the number of key-value pairs to return, and the size of the response.

SyncGetChangeProofResponse

This message is sent from the server to the client in response to a SyncGetChangeProofRequest. If the server had sufficient history to generate a change proof, it contains a change proof that contains the key-value pairs that changed between the requested roots. If the server did not have sufficient history to generate a change proof, it contains a range proof that contains the key-value pairs that were in the database when the database had the latter root. Like range proofs, if a client requests a change proof for range [requested_start, requested_end] but the server can't fit all the key-value pairs in one response, it'll send a change proof for [requested_start, proof_end] where proof_end < requested_end, as opposed to sending a change proof for [proof_start, requested_end] where proof_start > requested_start.

Algorithm

For each proof it receives, the sync client tracks the root hash of the revision associated with the proof's key-value pairs. For example, it will store information that says something like, "I have all of the key-value pairs that are in range [start, end] for the revision with root root_hash" for some keys start and end. Note that root_hash is the root hash of the revision that the client is trying to sync to, not the root hash of its own (incomplete) database. Tracking the revision associated with each downloaded key range, as well as using data in its own (incomplete) database, allows the client to figure out which key ranges are not up to date and need to be synced. The hash of the incomplete database on a client is never sent anywhere because it does not represent a root hash of any revision.

When the client is created, it is given the root hash of the revision to sync to. When it starts syncing, it requests from a server a range proof for the entire database. (To indicate that it wants no lower bound on the key range, the client doesn't provide a lower bound in the request. To indicate that it wants no upper bound, the client doesn't provide an upper bound. Thus, to request the entire database, the client omits both the lower and upper bounds in its request.) The server replies with a range proof, which the client verifies. If it's valid, the key-value pairs in the proof are written to the database. If it's not, the client drops the proof and requests the proof from another server.

A range proof sent by a server must return a continuous range of the key-value pairs, but may not return the full range that was requested. For example, a client might request all the key-value pairs in [requested_start, requested_end] but only receive those in range [requested_start, proof_end] where proof_end < requested_end. There might be too many key-value pairs to include in one message, or the server may be too busy to provide any more in its response. Unless the database is very small, this means that the range proof the client receives in response to its range proof request for the entire database will not contain all of the key-value pairs in the database.

If a client requests a range proof for range [requested_start, requested_end] but only receives a range proof for [requested_start, proof_end] where proof_end < requested_end it recognizes that it must still fetch all of the keys in [proof_end, requested_end]. It repeatedly requests range proofs for chunks of the remaining key range until it has all of the key-value pairs in [requested_start, requested_end]. The client may split the remaining key range into chunks and fetch chunks of key-value pairs in parallel, possibly even from different servers.

Additional commits to the database may occur while the client is syncing. The sync client can be notified that the root hash of the database it's trying to sync to has changed. Detecting that the root hash to sync to has changed is done outside this package. For example, if the database is being used to store blockchain state then the sync client would be notified when a new block is accepted because that implies a commit to the database. If this occurs, the key-value pairs the client has learned about via range proofs may no longer be up to date.

We use change proofs as an optimization to correct the out of date key-value pairs. When the sync client is notified that the root hash to sync to has changed, it requests a change proof from a server for a given key range. For example, if a client has the key-value pairs in range [start, end] that were in the database when it had root_hash, then it will request a change proof that provides all of the key-value changes in range [start, end] from the database version with root hash root_hash to the database version with root hash new_root_hash. The client verifies the change proof, and if it's valid, it applies the changes to its database. If it's not, the client drops the proof and requests the proof from another server.

A server needs to have history in order to serve a change proof. Namely, it needs to know all of the database changes between two roots. If the server does not have sufficient history to generate a change proof, it will send a range proof for the requested range at revision new_root_hash instead. The client will verify and apply the range proof. (Note that change proofs are just an optimization for bandwidth and speed. A range proof for a given key range and revision has the same information as a change proof from old_root_hash to new_root_hash for the key range, assuming the client has the key-value pairs for the key range at the revision with old_root_hash.) Change proofs, like range proofs, may not contain all of the key-value pairs in the requested range. This is OK because as mentioned above, the client tracks the root hash associated with each range of key-value pairs it has, so it knows which key-value pairs are out of date. Similar to range proofs, if a client requests the changes in range [requested_start, requested_end], but the server replies with all of the changes in [requested_start, proof_end] for some proof_end < requested_end, the client will repeatedly request change proofs until it gets remaining key-value pairs (namely in [proof_end, requested_end]).

Eventually, by repeatedly requesting, receiving, verifying and applying range and change proofs, the client will have all of the key-value pairs in the database. At this point, it's synced.

Diagram

Assuming you have Root Hash r1 which has many keys, some of which are k25, k50, k75, approximately 25%, 50%, and 75% of the way into the sorted set of keys, respectively, this diagram shows an example flow from client to server:

sequenceDiagram
    box Client/Server
        participant Server
        participant Client
    end
    box New Revision Notifier
        participant Notifier
    end

    Note right of Client: Normal sync flow
    Notifier->>Client: CurrentRoot(r1)
    Client->>Server: RangeProofRequest(r1, all)
    Server->>Client: RangeProofResponse(r1, ..k25)
    Client->>Server: RangeProofRequest(r1, k25..)
    Server->>Client: RangeProofResponse(r1, k25..k75)
    Notifier-)Client: NewRootHash(r2)
    Client->>Server: ChangeProofRequest(r1, r2, 0..k75)
    Server->>Client: ChangeProofResponse(r1, r2, 0..k50)
    Client->>Server: ChangeProofRequest(r1, r2, k50..k75)
    Server->>Client: ChangeProofResponse(r1, r2, k50..k75)
    Note right of Client: client is @r2 through (..k75)
    Client->>Server: RangeProofRequest(r2, k75..)
    Server->>Client: RangeProofResponse(r2, k75..k100)

TODOs

  • Handle errors on proof requests. Currently, any errors that occur server side are not sent back to the client.

Documentation

Overview

Package sync is a generated GoMock package.

Package sync is a generated GoMock package.

Index

Constants

This section is empty.

Variables

View Source 
var (
	ErrAlreadyStarted             = errors.New("cannot start a Manager that has already been started")
	ErrAlreadyClosed              = errors.New("Manager is closed")
	ErrNoClientProvided           = errors.New("client is a required field of the sync config")
	ErrNoDatabaseProvided         = errors.New("sync database is a required field of the sync config")
	ErrNoLogProvided              = errors.New("log is a required field of the sync config")
	ErrZeroWorkLimit              = errors.New("simultaneous work limit must be greater than 0")
	ErrFinishedWithUnexpectedRoot = errors.New("finished syncing with an unexpected root")
)
View Source 
var (
	ErrMinProofSizeIsTooLarge = errors.New("cannot generate any proof within the requested limit")
)

Functions

This section is empty.

Types

type Client 

type Client interface {
	// GetRangeProof synchronously sends the given request
	// and returns the parsed response.
	// This method verifies the range proof before returning it.
	GetRangeProof(
		ctx context.Context,
		request *pb.SyncGetRangeProofRequest,
	) (*merkledb.RangeProof, error)

	// GetChangeProof synchronously sends the given request
	// and returns the parsed response.
	// This method verifies the change proof / range proof
	// before returning it.
	// If the server responds with a change proof,
	// it's verified using [verificationDB].
	GetChangeProof(
		ctx context.Context,
		request *pb.SyncGetChangeProofRequest,
		verificationDB DB,
	) (*merkledb.ChangeOrRangeProof, error)
}

Client synchronously fetches data from the network to fulfill state sync requests. Repeatedly retries failed requests until the context is canceled.

func NewClient 

func NewClient(config *ClientConfig) (Client, error)

type ClientConfig 

type ClientConfig struct {
	NetworkClient    NetworkClient
	StateSyncNodeIDs []ids.NodeID
	Log              logging.Logger
	Metrics          SyncMetrics
	BranchFactor     merkledb.BranchFactor
	// If not specified, [merkledb.DefaultHasher] will be used.
	Hasher merkledb.Hasher
}

type DB 

type DB interface {
	merkledb.Clearer
	merkledb.MerkleRootGetter
	merkledb.ProofGetter
	merkledb.ChangeProofer
	merkledb.RangeProofer
}

type Manager 

type Manager struct {
	// contains filtered or unexported fields
}

func NewManager 

func NewManager(config ManagerConfig) (*Manager, error)

func (*Manager) Close 

func (m *Manager) Close()

Close will stop the syncing process

func (*Manager) Error 

func (m *Manager) Error() error

func (*Manager) Start 

func (m *Manager) Start(ctx context.Context) error

func (*Manager) UpdateSyncTarget 

func (m *Manager) UpdateSyncTarget(syncTargetRoot ids.ID) error

func (*Manager) Wait 

func (m *Manager) Wait(ctx context.Context) error

Wait blocks until one of the following occurs: - sync is complete. - sync fatally errored. - [ctx] is canceled. If [ctx] is canceled, returns [ctx].Err().

type ManagerConfig 

type ManagerConfig struct {
	DB                    DB
	Client                Client
	SimultaneousWorkLimit int
	Log                   logging.Logger
	TargetRoot            ids.ID
	BranchFactor          merkledb.BranchFactor
}

type MockClient 

type MockClient struct {
	// contains filtered or unexported fields
}

MockClient is a mock of Client interface.

func NewMockClient 

func NewMockClient(ctrl *gomock.Controller) *MockClient

NewMockClient creates a new mock instance.

func (*MockClient) EXPECT 

func (m *MockClient) EXPECT() *MockClientMockRecorder

EXPECT returns an object that allows the caller to indicate expected use.

func (*MockClient) GetChangeProof 

func (m *MockClient) GetChangeProof(arg0 context.Context, arg1 *sync.SyncGetChangeProofRequest, arg2 DB) (*merkledb.ChangeOrRangeProof, error)

GetChangeProof mocks base method.

func (*MockClient) GetRangeProof 

func (m *MockClient) GetRangeProof(arg0 context.Context, arg1 *sync.SyncGetRangeProofRequest) (*merkledb.RangeProof, error)

GetRangeProof mocks base method.

type MockClientMockRecorder 

type MockClientMockRecorder struct {
	// contains filtered or unexported fields
}

MockClientMockRecorder is the mock recorder for MockClient.

func (*MockClientMockRecorder) GetChangeProof 

func (mr *MockClientMockRecorder) GetChangeProof(arg0, arg1, arg2 any) *gomock.Call

GetChangeProof indicates an expected call of GetChangeProof.

func (*MockClientMockRecorder) GetRangeProof 

func (mr *MockClientMockRecorder) GetRangeProof(arg0, arg1 any) *gomock.Call

GetRangeProof indicates an expected call of GetRangeProof.

type MockNetworkClient 

type MockNetworkClient struct {
	// contains filtered or unexported fields
}

MockNetworkClient is a mock of NetworkClient interface.

func NewMockNetworkClient 

func NewMockNetworkClient(ctrl *gomock.Controller) *MockNetworkClient

NewMockNetworkClient creates a new mock instance.

func (*MockNetworkClient) AppRequestFailed 

func (m *MockNetworkClient) AppRequestFailed(arg0 context.Context, arg1 ids.NodeID, arg2 uint32) error

AppRequestFailed mocks base method.

func (*MockNetworkClient) AppResponse 

func (m *MockNetworkClient) AppResponse(arg0 context.Context, arg1 ids.NodeID, arg2 uint32, arg3 []byte) error

AppResponse mocks base method.

func (*MockNetworkClient) Connected 

func (m *MockNetworkClient) Connected(arg0 context.Context, arg1 ids.NodeID, arg2 *version.Application) error

Connected mocks base method.

func (*MockNetworkClient) Disconnected 

func (m *MockNetworkClient) Disconnected(arg0 context.Context, arg1 ids.NodeID) error

Disconnected mocks base method.

func (*MockNetworkClient) EXPECT 

func (m *MockNetworkClient) EXPECT() *MockNetworkClientMockRecorder

EXPECT returns an object that allows the caller to indicate expected use.

func (*MockNetworkClient) Request 

func (m *MockNetworkClient) Request(arg0 context.Context, arg1 ids.NodeID, arg2 []byte) ([]byte, error)

Request mocks base method.

func (*MockNetworkClient) RequestAny 

func (m *MockNetworkClient) RequestAny(arg0 context.Context, arg1 []byte) (ids.NodeID, []byte, error)

RequestAny mocks base method.

type MockNetworkClientMockRecorder 

type MockNetworkClientMockRecorder struct {
	// contains filtered or unexported fields
}

MockNetworkClientMockRecorder is the mock recorder for MockNetworkClient.

func (*MockNetworkClientMockRecorder) AppRequestFailed 

func (mr *MockNetworkClientMockRecorder) AppRequestFailed(arg0, arg1, arg2 any) *gomock.Call

AppRequestFailed indicates an expected call of AppRequestFailed.

func (*MockNetworkClientMockRecorder) AppResponse 

func (mr *MockNetworkClientMockRecorder) AppResponse(arg0, arg1, arg2, arg3 any) *gomock.Call

AppResponse indicates an expected call of AppResponse.

func (*MockNetworkClientMockRecorder) Connected 

func (mr *MockNetworkClientMockRecorder) Connected(arg0, arg1, arg2 any) *gomock.Call

Connected indicates an expected call of Connected.

func (*MockNetworkClientMockRecorder) Disconnected 

func (mr *MockNetworkClientMockRecorder) Disconnected(arg0, arg1 any) *gomock.Call

Disconnected indicates an expected call of Disconnected.

func (*MockNetworkClientMockRecorder) Request 

func (mr *MockNetworkClientMockRecorder) Request(arg0, arg1, arg2 any) *gomock.Call

Request indicates an expected call of Request.

func (*MockNetworkClientMockRecorder) RequestAny 

func (mr *MockNetworkClientMockRecorder) RequestAny(arg0, arg1 any) *gomock.Call

RequestAny indicates an expected call of RequestAny.

type NetworkClient 

type NetworkClient interface {
	// RequestAny synchronously sends request to an arbitrary peer with a
	// node version greater than or equal to minVersion.
	// Returns response bytes, the ID of the chosen peer, and ErrRequestFailed if
	// the request should be retried.
	RequestAny(
		ctx context.Context,
		request []byte,
	) (ids.NodeID, []byte, error)

	// Sends [request] to [nodeID] and returns the response.
	// Blocks until the number of outstanding requests is
	// below the limit before sending the request.
	Request(
		ctx context.Context,
		nodeID ids.NodeID,
		request []byte,
	) ([]byte, error)

	// Always returns nil because the engine considers errors
	// returned from this function as fatal.
	AppResponse(context.Context, ids.NodeID, uint32, []byte) error

	// Always returns nil because the engine considers errors
	// returned from this function as fatal.
	AppRequestFailed(context.Context, ids.NodeID, uint32) error

	// Adds the given [nodeID] to the peer
	// list so that it can receive messages.
	// If [nodeID] is this node's ID, this is a no-op.
	Connected(context.Context, ids.NodeID, *version.Application) error

	// Removes given [nodeID] from the peer list.
	Disconnected(context.Context, ids.NodeID) error
}

NetworkClient defines ability to send request / response through the Network

func NewNetworkClient 

func NewNetworkClient(
	appSender common.AppSender,
	myNodeID ids.NodeID,
	maxActiveRequests int64,
	log logging.Logger,
	metricsNamespace string,
	registerer prometheus.Registerer,
	minVersion *version.Application,
) (NetworkClient, error)

type NetworkServer 

type NetworkServer struct {
	// contains filtered or unexported fields
}

func NewNetworkServer 

func NewNetworkServer(appSender common.AppSender, db DB, log logging.Logger) *NetworkServer

func (*NetworkServer) AppRequest 

func (s *NetworkServer) AppRequest(
	ctx context.Context,
	nodeID ids.NodeID,
	requestID uint32,
	deadline time.Time,
	request []byte,
) error

AppRequest is called by avalanchego -> VM when there is an incoming AppRequest from a peer. Returns a non-nil error iff we fail to send an app message. This is a fatal error. Sends a response back to the sender if length of response returned by the handler > 0.

func (*NetworkServer) HandleChangeProofRequest 

func (s *NetworkServer) HandleChangeProofRequest(
	ctx context.Context,
	nodeID ids.NodeID,
	requestID uint32,
	req *pb.SyncGetChangeProofRequest,
) error

Generates a change proof and sends it to [nodeID]. If [errAppSendFailed] is returned, this should be considered fatal.

func (*NetworkServer) HandleRangeProofRequest 

func (s *NetworkServer) HandleRangeProofRequest(
	ctx context.Context,
	nodeID ids.NodeID,
	requestID uint32,
	req *pb.SyncGetRangeProofRequest,
) error

Generates a range proof and sends it to [nodeID]. If [errAppSendFailed] is returned, this should be considered fatal.

type ResponseHandler 

type ResponseHandler interface {
	// Called when [response] is received.
	OnResponse(response []byte)
	// Called when the request failed or timed out.
	OnFailure()
}

Handles responses/failure notifications for a sent request. Exactly one of OnResponse or OnFailure is eventually called.

type SyncMetrics 

type SyncMetrics interface {
	RequestFailed()
	RequestMade()
	RequestSucceeded()
}

func NewMetrics 

func NewMetrics(namespace string, reg prometheus.Registerer) (SyncMetrics, error)

Source Files

View all Source files

Directories

Path Synopsis

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.