Distributed POST verification (#5390)

Part of #5185, closes #5376

- [x] Local randomness seed to select K3 indices using p2p peer ID
- [x] Verifying K3 indices of POST
- [x] Verifying ALL indices for initial POST
- [x] Publishing POST malfeasance proofs
- [x] Verifying only the invalid index in POST malfeasance proofs
- [x] Verify the candidate for a positioning ATX and its chain
- [x] Verify the candidate for a commitment ATX and its chain
- [x] Wait "some time" for malfeasance proofs before creating an active set
        ☝️ this is covered by ATX grading
- [x] configurable duration after which ATXs (and their chains) are considered valid (in terms of POST labels)
- [x] system tests

CHANGELOG.md

@@ -86,6 +86,14 @@ configuration is as follows:
   The node will automatically migrate the data from disk and store it in the database. The migration will take place at the
   first startup after the upgrade.
 
+* [#5390](https://github.com/spacemeshos/go-spacemesh/pull/5390)
+  Distributed PoST verification.
+
+  The nodes on the network can now choose to verify
+  only a subset of labels in PoST proofs by choosing a K3 value lower than K2.
+  If a node finds a proof invalid, it will report it to the network by
+  creating a malfeasance proof. The malicious node will then be blacklisted by the network.
+
 ### Features
 
 ### Improvements
@@ -346,7 +354,7 @@ for more information on how to configure the node to work with the PoST service.
 
 ### Improvements
 
-* further increased cache sizes and and p2p timeouts to compensate for the increased number of nodes on the network.
+* further increased cache sizes and p2p timeouts to compensate for the increased number of nodes on the network.
 
 * [#5329](https://github.com/spacemeshos/go-spacemesh/pull/5329) P2P decentralization improvements. Added support for QUIC
   transport and DHT routing discovery for finding peers and relays. Also, added the `ping-peers` feature which is useful

Makefile-libs.Inc

@@ -50,7 +50,7 @@ else
 	endif
 endif
 
-POSTRS_SETUP_REV = 0.6.6
+POSTRS_SETUP_REV = 0.7.0
 POSTRS_SETUP_ZIP = libpost-$(platform)-v$(POSTRS_SETUP_REV).zip
 POSTRS_SETUP_URL_ZIP ?= https://github.com/spacemeshos/post-rs/releases/download/v$(POSTRS_SETUP_REV)/$(POSTRS_SETUP_ZIP)
 POSTRS_PROFILER_ZIP = profiler-$(platform)-v$(POSTRS_SETUP_REV).zip

activation/activation.go

@@ -58,7 +58,6 @@ const (
 // Config defines configuration for Builder.
 type Config struct {
 	GoldenATXID      types.ATXID
-	LayersPerEpoch   uint32
 	RegossipInterval time.Duration
 }
 
@@ -90,11 +89,18 @@ type Builder struct {
 	stop              context.CancelFunc
 	poetCfg           PoetConfig
 	poetRetryInterval time.Duration
+	// delay before PoST in ATX is considered valid (counting from the time it was received)
+	postValidityDelay time.Duration
 }
 
-// BuilderOption ...
 type BuilderOption func(*Builder)
 
+func WithPostValidityDelay(delay time.Duration) BuilderOption {
+	return func(b *Builder) {
+		b.postValidityDelay = delay
+	}
+}
+
 // WithPoetRetryInterval modifies time that builder will have to wait before retrying ATX build process
 // if it failed due to issues with PoET server.
 func WithPoetRetryInterval(interval time.Duration) BuilderOption {
@@ -149,6 +155,7 @@ func NewBuilder(
 		syncer:            syncer,
 		log:               log,
 		poetRetryInterval: defaultPoetRetryInterval,
+		postValidityDelay: 12 * time.Hour,
 	}
 	for _, opt := range opts {
 		opt(b)
@@ -403,7 +410,7 @@ func (b *Builder) buildNIPostChallenge(ctx context.Context) (*types.NIPostChalle
 		}
 	}
 
-	posAtx, err := b.GetPositioningAtx()
+	posAtx, err := b.getPositioningAtx(ctx)
 	if err != nil {
 		return nil, fmt.Errorf("failed to get positioning ATX: %w", err)
 	}
@@ -584,17 +591,24 @@ func (b *Builder) broadcast(ctx context.Context, atx *types.ActivationTx) (int,
 	return len(buf), nil
 }
 
-// GetPositioningAtx returns atx id with the highest tick height.
-func (b *Builder) GetPositioningAtx() (types.ATXID, error) {
-	id, err := atxs.GetIDWithMaxHeight(b.cdb, b.signer.NodeID())
-	if err != nil {
-		if errors.Is(err, sql.ErrNotFound) {
-			b.log.Info("using golden atx as positioning atx")
-			return b.goldenATXID, nil
-		}
-		return types.ATXID{}, fmt.Errorf("cannot find pos atx: %w", err)
+// getPositioningAtx returns atx id with the highest tick height.
+func (b *Builder) getPositioningAtx(ctx context.Context) (types.ATXID, error) {
+	id, err := findFullyValidHighTickAtx(
+		ctx,
+		b.cdb,
+		b.signer.NodeID(),
+		b.goldenATXID,
+		b.validator,
+		b.log,
+		VerifyChainOpts.AssumeValidBefore(time.Now().Add(-b.postValidityDelay)),
+		VerifyChainOpts.WithTrustedID(b.signer.NodeID()),
+		VerifyChainOpts.WithLogger(b.log),
+	)
+	if errors.Is(err, sql.ErrNotFound) {
+		b.log.Info("using golden atx as positioning atx")
+		return b.goldenATXID, nil
 	}
-	return id, nil
+	return id, err
 }
 
 func (b *Builder) Regossip(ctx context.Context) error {
@@ -630,3 +644,38 @@ func buildNipostChallengeStartDeadline(roundStart time.Time, gracePeriod time.Du
 	jitter := randomDurationInRange(time.Duration(0), gracePeriod*maxNipostChallengeBuildJitter/100.0)
 	return roundStart.Add(jitter).Add(-gracePeriod)
 }
+
+func findFullyValidHighTickAtx(
+	ctx context.Context,
+	db sql.Executor,
+	prefNodeID types.NodeID,
+	goldenATXID types.ATXID,
+	validator nipostValidator,
+	log *zap.Logger,
+	opts ...VerifyChainOption,
+) (types.ATXID, error) {
+	rejectedAtxs := make(map[types.ATXID]struct{})
+	filter := func(id types.ATXID) bool {
+		_, ok := rejectedAtxs[id]
+		return !ok
+	}
+
+	for {
+		select {
+		case <-ctx.Done():
+			return types.ATXID{}, ctx.Err()
+		default:
+		}
+		id, err := atxs.GetIDWithMaxHeight(db, prefNodeID, filter)
+		if err != nil {
+			return types.ATXID{}, err
+		}
+
+		if err := validator.VerifyChain(ctx, id, goldenATXID, opts...); err != nil {
+			log.Info("rejecting candidate for high-tick atx", zap.Error(err), zap.Stringer("atx_id", id))
+			rejectedAtxs[id] = struct{}{}
+		} else {
+			return id, nil
+		}
+	}
+}