mock fixes

Co-authored-by: Kasey Kirkham <kasey@users.noreply.github.com>

.bazelrc

@@ -6,6 +6,12 @@ import %workspace%/build/bazelrc/debug.bazelrc
 import %workspace%/build/bazelrc/hermetic-cc.bazelrc
 import %workspace%/build/bazelrc/performance.bazelrc
 
+# hermetic_cc_toolchain v3.0.1 required changes.
+common --enable_platform_specific_config
+build:linux --sandbox_add_mount_pair=/tmp
+build:macos --sandbox_add_mount_pair=/var/tmp
+build:windows --sandbox_add_mount_pair=C:\Temp
+
 # E2E run with debug gotag
 test:e2e --define gotags=debug
 

.bazelversion

@@ -1 +1 @@
-7.0.0
+7.1.0

.gitignore

@@ -41,3 +41,6 @@ jwt.hex
 
 # manual testing
 tmp
+
+# spectest coverage reports
+report.txt

README.md

@@ -2,7 +2,7 @@
 
 [![Build status](https://badge.buildkite.com/b555891daf3614bae4284dcf365b2340cefc0089839526f096.svg?branch=master)](https://buildkite.com/prysmatic-labs/prysm)
 [![Go Report Card](https://goreportcard.com/badge/github.com/prysmaticlabs/prysm)](https://goreportcard.com/report/github.com/prysmaticlabs/prysm)
-[![Consensus_Spec_Version 1.3.0](https://img.shields.io/badge/Consensus%20Spec%20Version-v1.3.0-blue.svg)](https://github.com/ethereum/consensus-specs/tree/v1.3.0)
+[![Consensus_Spec_Version 1.4.0](https://img.shields.io/badge/Consensus%20Spec%20Version-v1.4.0-blue.svg)](https://github.com/ethereum/consensus-specs/tree/v1.4.0)
 [![Execution_API_Version 1.0.0-beta.2](https://img.shields.io/badge/Execution%20API%20Version-v1.0.0.beta.2-blue.svg)](https://github.com/ethereum/execution-apis/tree/v1.0.0-beta.2/src/engine)
 [![Discord](https://user-images.githubusercontent.com/7288322/34471967-1df7808a-efbb-11e7-9088-ed0b04151291.png)](https://discord.gg/prysmaticlabs)
 [![GitPOAP Badge](https://public-api.gitpoap.io/v1/repo/prysmaticlabs/prysm/badge)](https://www.gitpoap.io/gh/prysmaticlabs/prysm)

WORKSPACE

@@ -16,12 +16,14 @@ load("@rules_pkg//:deps.bzl", "rules_pkg_dependencies")
 
 rules_pkg_dependencies()
 
+HERMETIC_CC_TOOLCHAIN_VERSION = "v3.0.1"
+
 http_archive(
     name = "hermetic_cc_toolchain",
-    sha256 = "973ab22945b921ef45b8e1d6ce01ca7ce1b8a462167449a36e297438c4ec2755",
-    strip_prefix = "hermetic_cc_toolchain-5098046bccc15d2962f3cc8e7e53d6a2a26072dc",
+    sha256 = "3bc6ec127622fdceb4129cb06b6f7ab098c4d539124dde96a6318e7c32a53f7a",
     urls = [
-        "https://github.com/uber/hermetic_cc_toolchain/archive/5098046bccc15d2962f3cc8e7e53d6a2a26072dc.tar.gz",  # 2023-06-28
+        "https://mirror.bazel.build/github.com/uber/hermetic_cc_toolchain/releases/download/{0}/hermetic_cc_toolchain-{0}.tar.gz".format(HERMETIC_CC_TOOLCHAIN_VERSION),
+        "https://github.com/uber/hermetic_cc_toolchain/releases/download/{0}/hermetic_cc_toolchain-{0}.tar.gz".format(HERMETIC_CC_TOOLCHAIN_VERSION),
     ],
 )
 
@@ -81,10 +83,10 @@ bazel_skylib_workspace()
 
 http_archive(
     name = "bazel_gazelle",
-    sha256 = "d3fa66a39028e97d76f9e2db8f1b0c11c099e8e01bf363a923074784e451f809",
+    integrity = "sha256-MpOL2hbmcABjA1R5Bj2dJMYO2o15/Uc5Vj9Q0zHLMgk=",
     urls = [
-        "https://mirror.bazel.build/github.com/bazelbuild/bazel-gazelle/releases/download/v0.33.0/bazel-gazelle-v0.33.0.tar.gz",
-        "https://github.com/bazelbuild/bazel-gazelle/releases/download/v0.33.0/bazel-gazelle-v0.33.0.tar.gz",
+        "https://mirror.bazel.build/github.com/bazelbuild/bazel-gazelle/releases/download/v0.35.0/bazel-gazelle-v0.35.0.tar.gz",
+        "https://github.com/bazelbuild/bazel-gazelle/releases/download/v0.35.0/bazel-gazelle-v0.35.0.tar.gz",
     ],
 )
 
@@ -113,6 +115,13 @@ http_archive(
     url = "https://github.com/GoogleContainerTools/distroless/archive/9dc924b9fe812eec2fa0061824dcad39eb09d0d6.tar.gz",  # 2024-01-24
 )
 
+http_archive(
+    name = "aspect_bazel_lib",
+    sha256 = "f5ea76682b209cc0bd90d0f5a3b26d2f7a6a2885f0c5f615e72913f4805dbb0d",
+    strip_prefix = "bazel-lib-2.5.0",
+    url = "https://github.com/aspect-build/bazel-lib/releases/download/v2.5.0/bazel-lib-v2.5.0.tar.gz",
+)
+
 load("@aspect_bazel_lib//lib:repositories.bzl", "aspect_bazel_lib_dependencies", "aspect_bazel_lib_register_toolchains")
 
 aspect_bazel_lib_dependencies()
@@ -121,9 +130,9 @@ aspect_bazel_lib_register_toolchains()
 
 http_archive(
     name = "rules_oci",
-    sha256 = "c71c25ed333a4909d2dd77e0b16c39e9912525a98c7fa85144282be8d04ef54c",
-    strip_prefix = "rules_oci-1.3.4",
-    url = "https://github.com/bazel-contrib/rules_oci/releases/download/v1.3.4/rules_oci-v1.3.4.tar.gz",
+    sha256 = "4a276e9566c03491649eef63f27c2816cc222f41ccdebd97d2c5159e84917c3b",
+    strip_prefix = "rules_oci-1.7.4",
+    url = "https://github.com/bazel-contrib/rules_oci/releases/download/v1.7.4/rules_oci-v1.7.4.tar.gz",
 )
 
 load("@rules_oci//oci:dependencies.bzl", "rules_oci_dependencies")
@@ -144,17 +153,13 @@ http_archive(
         # Expose internals of go_test for custom build transitions.
         "//third_party:io_bazel_rules_go_test.patch",
     ],
-    sha256 = "d6ab6b57e48c09523e93050f13698f708428cfd5e619252e369d377af6597707",
+    sha256 = "80a98277ad1311dacd837f9b16db62887702e9f1d1c4c9f796d0121a46c8e184",
     urls = [
-        "https://mirror.bazel.build/github.com/bazelbuild/rules_go/releases/download/v0.43.0/rules_go-v0.43.0.zip",
-        "https://github.com/bazelbuild/rules_go/releases/download/v0.43.0/rules_go-v0.43.0.zip",
+        "https://mirror.bazel.build/github.com/bazelbuild/rules_go/releases/download/v0.46.0/rules_go-v0.46.0.zip",
+        "https://github.com/bazelbuild/rules_go/releases/download/v0.46.0/rules_go-v0.46.0.zip",
     ],
 )
 
-load("//:distroless_deps.bzl", "distroless_deps")
-
-distroless_deps()
-
 # Override default import in rules_go with special patch until
 # https://github.com/gogo/protobuf/pull/582 is merged.
 git_repository(
@@ -193,10 +198,14 @@ load("@io_bazel_rules_go//go:deps.bzl", "go_register_toolchains", "go_rules_depe
 go_rules_dependencies()
 
 go_register_toolchains(
-    go_version = "1.21.6",
+    go_version = "1.21.8",
     nogo = "@//:nogo",
 )
 
+load("//:distroless_deps.bzl", "distroless_deps")
+
+distroless_deps()
+
 http_archive(
     name = "io_kubernetes_build",
     sha256 = "b84fbd1173acee9d02a7d3698ad269fdf4f7aa081e9cecd40e012ad0ad8cfa2a",
@@ -234,9 +243,7 @@ filegroup(
     url = "https://github.com/ethereum/EIPs/archive/5480440fe51742ed23342b68cf106cefd427e39d.tar.gz",
 )
 
-consensus_spec_version = "v1.4.0-beta.7"
-
-consensus_spec_test_version = "v1.4.0-beta.7-hotfix"
+consensus_spec_version = "v1.4.0"
 
 bls_test_version = "v0.1.1"
 
@@ -253,7 +260,7 @@ filegroup(
 )
     """,
     sha256 = "c282c0f86f23f3d2e0f71f5975769a4077e62a7e3c7382a16bd26a7e589811a0",
-    url = "https://github.com/ethereum/consensus-spec-tests/releases/download/%s/general.tar.gz" % consensus_spec_test_version,
+    url = "https://github.com/ethereum/consensus-spec-tests/releases/download/%s/general.tar.gz" % consensus_spec_version,
 )
 
 http_archive(
@@ -269,7 +276,7 @@ filegroup(
 )
     """,
     sha256 = "4649c35aa3b8eb0cfdc81bee7c05649f90ef36bede5b0513e1f2e8baf37d6033",
-    url = "https://github.com/ethereum/consensus-spec-tests/releases/download/%s/minimal.tar.gz" % consensus_spec_test_version,
+    url = "https://github.com/ethereum/consensus-spec-tests/releases/download/%s/minimal.tar.gz" % consensus_spec_version,
 )
 
 http_archive(
@@ -285,7 +292,7 @@ filegroup(
 )
     """,
     sha256 = "c5a03f724f757456ffaabd2a899992a71d2baf45ee4db65ca3518f2b7ee928c8",
-    url = "https://github.com/ethereum/consensus-spec-tests/releases/download/%s/mainnet.tar.gz" % consensus_spec_test_version,
+    url = "https://github.com/ethereum/consensus-spec-tests/releases/download/%s/mainnet.tar.gz" % consensus_spec_version,
 )
 
 http_archive(
@@ -299,7 +306,7 @@ filegroup(
     visibility = ["//visibility:public"],
 )
     """,
-    sha256 = "049c29267310e6b88280f4f834a75866c2f5b9036fa97acb9d9c6db8f64d9118",
+    sha256 = "cd1c9d97baccbdde1d2454a7dceb8c6c61192a3b581eee12ffc94969f2db8453",
     strip_prefix = "consensus-specs-" + consensus_spec_version[1:],
     url = "https://github.com/ethereum/consensus-specs/archive/refs/tags/%s.tar.gz" % consensus_spec_version,
 )
@@ -335,22 +342,6 @@ filegroup(
     url = "https://github.com/eth-clients/eth2-networks/archive/934c948e69205dcf2deb87e4ae6cc140c335f94d.tar.gz",
 )
 
-http_archive(
-    name = "goerli_testnet",
-    build_file_content = """
-filegroup(
-    name = "configs",
-    srcs = [
-        "prater/config.yaml",
-    ],
-    visibility = ["//visibility:public"],
-)
-    """,
-    sha256 = "43fc0f55ddff7b511713e2de07aa22846a67432df997296fb4fc09cd8ed1dcdb",
-    strip_prefix = "goerli-6522ac6684693740cd4ddcc2a0662e03702aa4a1",
-    url = "https://github.com/eth-clients/goerli/archive/6522ac6684693740cd4ddcc2a0662e03702aa4a1.tar.gz",
-)
-
 http_archive(
     name = "holesky_testnet",
     build_file_content = """

api/client/beacon/BUILD.bazel

@@ -6,11 +6,14 @@ go_library(
         "checkpoint.go",
         "client.go",
         "doc.go",
+        "health.go",
+        "log.go",
     ],
     importpath = "github.com/prysmaticlabs/prysm/v5/api/client/beacon",
     visibility = ["//visibility:public"],
     deps = [
         "//api/client:go_default_library",
+        "//api/client/beacon/iface:go_default_library",
         "//api/server:go_default_library",
         "//api/server/structs:go_default_library",
         "//beacon-chain/core/helpers:go_default_library",
@@ -36,10 +39,12 @@ go_test(
     srcs = [
         "checkpoint_test.go",
         "client_test.go",
+        "health_test.go",
     ],
     embed = [":go_default_library"],
     deps = [
         "//api/client:go_default_library",
+        "//api/client/beacon/testing:go_default_library",
         "//beacon-chain/state:go_default_library",
         "//config/params:go_default_library",
         "//consensus-types/blocks:go_default_library",
@@ -53,5 +58,6 @@ go_test(
         "//testing/util:go_default_library",
         "//time/slots:go_default_library",
         "@com_github_pkg_errors//:go_default_library",
+        "@org_uber_go_mock//gomock:go_default_library",
     ],
 )

api/client/beacon/checkpoint.go

@@ -17,7 +17,7 @@ import (
 	"github.com/prysmaticlabs/prysm/v5/io/file"
 	"github.com/prysmaticlabs/prysm/v5/runtime/version"
 	"github.com/prysmaticlabs/prysm/v5/time/slots"
-	log "github.com/sirupsen/logrus"
+	"github.com/sirupsen/logrus"
 	"golang.org/x/mod/semver"
 )
 
@@ -74,7 +74,12 @@ func DownloadFinalizedData(ctx context.Context, client *Client) (*OriginData, er
 	if err != nil {
 		return nil, errors.Wrap(err, "error detecting chain config for finalized state")
 	}
-	log.Printf("detected supported config in remote finalized state, name=%s, fork=%s", vu.Config.ConfigName, version.String(vu.Fork))
+
+	log.WithFields(logrus.Fields{
+		"name": vu.Config.ConfigName,
+		"fork": version.String(vu.Fork),
+	}).Info("Detected supported config in remote finalized state")
+
 	s, err := vu.UnmarshalBeaconState(sb)
 	if err != nil {
 		return nil, errors.Wrap(err, "error unmarshaling finalized state to correct version")
@@ -108,10 +113,10 @@ func DownloadFinalizedData(ctx context.Context, client *Client) (*OriginData, er
 	}
 
 	log.
-		WithField("block_slot", b.Block().Slot()).
-		WithField("state_slot", s.Slot()).
-		WithField("state_root", hexutil.Encode(sr[:])).
-		WithField("block_root", hexutil.Encode(br[:])).
+		WithField("blockSlot", b.Block().Slot()).
+		WithField("stateSlot", s.Slot()).
+		WithField("stateRoot", hexutil.Encode(sr[:])).
+		WithField("blockRoot", hexutil.Encode(br[:])).
 		Info("Downloaded checkpoint sync state and block.")
 	return &OriginData{
 		st: s,

api/client/beacon/client.go

@@ -22,7 +22,7 @@ import (
 	"github.com/prysmaticlabs/prysm/v5/encoding/bytesutil"
 	"github.com/prysmaticlabs/prysm/v5/network/forks"
 	ethpb "github.com/prysmaticlabs/prysm/v5/proto/prysm/v1alpha1"
-	log "github.com/sirupsen/logrus"
+	"github.com/sirupsen/logrus"
 )
 
 const (
@@ -309,9 +309,9 @@ func (c *Client) SubmitChangeBLStoExecution(ctx context.Context, request []*stru
 		}
 		for _, failure := range errorJson.Failures {
 			w := request[failure.Index].Message
-			log.WithFields(log.Fields{
-				"validator_index":    w.ValidatorIndex,
-				"withdrawal_address": w.ToExecutionAddress,
+			log.WithFields(logrus.Fields{
+				"validatorIndex":    w.ValidatorIndex,
+				"withdrawalAddress": w.ToExecutionAddress,
 			}).Error(failure.Message)
 		}
 		return errors.Errorf("POST error %d: %s", errorJson.Code, errorJson.Message)
@@ -341,9 +341,9 @@ type forkScheduleResponse struct {
 func (fsr *forkScheduleResponse) OrderedForkSchedule() (forks.OrderedSchedule, error) {
 	ofs := make(forks.OrderedSchedule, 0)
 	for _, d := range fsr.Data {
-		epoch, err := strconv.Atoi(d.Epoch)
+		epoch, err := strconv.ParseUint(d.Epoch, 10, 64)
 		if err != nil {
-			return nil, err
+			return nil, errors.Wrapf(err, "error parsing epoch %s", d.Epoch)
 		}
 		vSlice, err := hexutil.Decode(d.CurrentVersion)
 		if err != nil {
@@ -355,7 +355,7 @@ func (fsr *forkScheduleResponse) OrderedForkSchedule() (forks.OrderedSchedule, e
 		version := bytesutil.ToBytes4(vSlice)
 		ofs = append(ofs, forks.ForkScheduleEntry{
 			Version: version,
-			Epoch:   primitives.Epoch(uint64(epoch)),
+			Epoch:   primitives.Epoch(epoch),
 		})
 	}
 	sort.Sort(ofs)

api/client/beacon/health.go

@@ -0,0 +1,55 @@
+package beacon
+
+import (
+	"context"
+	"sync"
+
+	"github.com/prysmaticlabs/prysm/v5/api/client/beacon/iface"
+)
+
+type NodeHealthTracker struct {
+	isHealthy  *bool
+	healthChan chan bool
+	node       iface.HealthNode
+	sync.RWMutex
+}
+
+func NewNodeHealthTracker(node iface.HealthNode) *NodeHealthTracker {
+	return &NodeHealthTracker{
+		node:       node,
+		healthChan: make(chan bool, 1),
+	}
+}
+
+// HealthUpdates provides a read-only channel for health updates.
+func (n *NodeHealthTracker) HealthUpdates() <-chan bool {
+	return n.healthChan
+}
+
+func (n *NodeHealthTracker) IsHealthy() bool {
+	n.RLock()
+	defer n.RUnlock()
+	if n.isHealthy == nil {
+		return false
+	}
+	return *n.isHealthy
+}
+
+func (n *NodeHealthTracker) CheckHealth(ctx context.Context) bool {
+	n.RLock()
+	newStatus := n.node.IsHealthy(ctx)
+	if n.isHealthy == nil {
+		n.isHealthy = &newStatus
+	}
+	isStatusChanged := newStatus != *n.isHealthy
+	n.RUnlock()
+
+	if isStatusChanged {
+		n.Lock()
+		// Double-check the condition to ensure it hasn't changed since the first check.
+		n.isHealthy = &newStatus
+		n.Unlock() // It's better to unlock as soon as the protected section is over.
+		n.healthChan <- newStatus
+	}
+	return newStatus
+}

api/client/beacon/health_test.go

@@ -0,0 +1,118 @@
+package beacon
+
+import (
+	"context"
+	"sync"
+	"testing"
+
+	healthTesting "github.com/prysmaticlabs/prysm/v5/api/client/beacon/testing"
+	"go.uber.org/mock/gomock"
+)
+
+func TestNodeHealth_IsHealthy(t *testing.T) {
+	tests := []struct {
+		name      string
+		isHealthy bool
+		want      bool
+	}{
+		{"initially healthy", true, true},
+		{"initially unhealthy", false, false},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			n := &NodeHealthTracker{
+				isHealthy:  &tt.isHealthy,
+				healthChan: make(chan bool, 1),
+			}
+			if got := n.IsHealthy(); got != tt.want {
+				t.Errorf("IsHealthy() = %v, want %v", got, tt.want)
+			}
+		})
+	}
+}
+
+func TestNodeHealth_UpdateNodeHealth(t *testing.T) {
+	tests := []struct {
+		name       string
+		initial    bool // Initial health status
+		newStatus  bool // Status to update to
+		shouldSend bool // Should a message be sent through the channel
+	}{
+		{"healthy to unhealthy", true, false, true},
+		{"unhealthy to healthy", false, true, true},
+		{"remain healthy", true, true, false},
+		{"remain unhealthy", false, false, false},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			ctrl := gomock.NewController(t)
+			defer ctrl.Finish()
+			client := healthTesting.NewMockHealthClient(ctrl)
+			client.EXPECT().IsHealthy(gomock.Any()).Return(tt.newStatus)
+			n := &NodeHealthTracker{
+				isHealthy:  &tt.initial,
+				node:       client,
+				healthChan: make(chan bool, 1),
+			}
+
+			s := n.CheckHealth(context.Background())
+			// Check if health status was updated
+			if s != tt.newStatus {
+				t.Errorf("UpdateNodeHealth() failed to update isHealthy from %v to %v", tt.initial, tt.newStatus)
+			}
+
+			select {
+			case status := <-n.HealthUpdates():
+				if !tt.shouldSend {
+					t.Errorf("UpdateNodeHealth() unexpectedly sent status %v to HealthCh", status)
+				} else if status != tt.newStatus {
+					t.Errorf("UpdateNodeHealth() sent wrong status %v, want %v", status, tt.newStatus)
+				}
+			default:
+				if tt.shouldSend {
+					t.Error("UpdateNodeHealth() did not send any status to HealthCh when expected")
+				}
+			}
+		})
+	}
+}
+
+func TestNodeHealth_Concurrency(t *testing.T) {
+	ctrl := gomock.NewController(t)
+	defer ctrl.Finish()
+	client := healthTesting.NewMockHealthClient(ctrl)
+	n := NewNodeHealthTracker(client)
+	var wg sync.WaitGroup
+
+	// Number of goroutines to spawn for both reading and writing
+	numGoroutines := 6
+
+	go func() {
+		for range n.HealthUpdates() {
+			// Consume values to avoid blocking on channel send.
+		}
+	}()
+
+	wg.Add(numGoroutines * 2) // for readers and writers
+
+	// Concurrently update health status
+	for i := 0; i < numGoroutines; i++ {
+		go func() {
+			defer wg.Done()
+			client.EXPECT().IsHealthy(gomock.Any()).Return(false)
+			n.CheckHealth(context.Background())
+			client.EXPECT().IsHealthy(gomock.Any()).Return(true)
+			n.CheckHealth(context.Background())
+		}()
+	}
+
+	// Concurrently read health status
+	for i := 0; i < numGoroutines; i++ {
+		go func() {
+			defer wg.Done()
+			_ = n.IsHealthy() // Just read the value
+		}()
+	}
+
+	wg.Wait() // Wait for all goroutines to finish
+}

api/client/beacon/iface/BUILD.bazel

@@ -0,0 +1,8 @@
+load("@prysm//tools/go:def.bzl", "go_library")
+
+go_library(
+    name = "go_default_library",
+    srcs = ["health.go"],
+    importpath = "github.com/prysmaticlabs/prysm/v5/api/client/beacon/iface",
+    visibility = ["//visibility:public"],
+)

api/client/beacon/iface/health.go

@@ -0,0 +1,13 @@
+package iface
+
+import "context"
+
+type HealthTracker interface {
+	HealthUpdates() <-chan bool
+	IsHealthy() bool
+	CheckHealth(ctx context.Context) bool
+}
+
+type HealthNode interface {
+	IsHealthy(ctx context.Context) bool
+}

api/client/beacon/log.go

@@ -0,0 +1,5 @@
+package beacon
+
+import "github.com/sirupsen/logrus"
+
+var log = logrus.WithField("prefix", "beacon")

api/client/beacon/testing/BUILD.bazel

@@ -0,0 +1,12 @@
+load("@prysm//tools/go:def.bzl", "go_library")
+
+go_library(
+    name = "go_default_library",
+    srcs = ["mock.go"],
+    importpath = "github.com/prysmaticlabs/prysm/v5/api/client/beacon/testing",
+    visibility = ["//visibility:public"],
+    deps = [
+        "//api/client/beacon/iface:go_default_library",
+        "@org_uber_go_mock//gomock:go_default_library",
+    ],
+)

api/client/beacon/testing/mock.go

@@ -0,0 +1,53 @@
+package testing
+
+import (
+	"context"
+	"reflect"
+
+	"github.com/prysmaticlabs/prysm/v5/api/client/beacon/iface"
+	"go.uber.org/mock/gomock"
+)
+
+var (
+	_ = iface.HealthNode(&MockHealthClient{})
+)
+
+// MockHealthClient is a mock of HealthClient interface.
+type MockHealthClient struct {
+	ctrl     *gomock.Controller
+	recorder *MockHealthClientMockRecorder
+}
+
+// MockHealthClientMockRecorder is the mock recorder for MockHealthClient.
+type MockHealthClientMockRecorder struct {
+	mock *MockHealthClient
+}
+
+// IsHealthy mocks base method.
+func (m *MockHealthClient) IsHealthy(arg0 context.Context) bool {
+	m.ctrl.T.Helper()
+	ret := m.ctrl.Call(m, "IsHealthy", arg0)
+	ret0, ok := ret[0].(bool)
+	if !ok {
+		return false
+	}
+	return ret0
+}
+
+// EXPECT returns an object that allows the caller to indicate expected use.
+func (m *MockHealthClient) EXPECT() *MockHealthClientMockRecorder {
+	return m.recorder
+}
+
+// IsHealthy indicates an expected call of IsHealthy.
+func (mr *MockHealthClientMockRecorder) IsHealthy(arg0 any) *gomock.Call {
+	mr.mock.ctrl.T.Helper()
+	return mr.mock.ctrl.RecordCallWithMethodType(mr.mock, "IsHealthy", reflect.TypeOf((*MockHealthClient)(nil).IsHealthy), arg0)
+}
+
+// NewMockHealthClient creates a new mock instance.
+func NewMockHealthClient(ctrl *gomock.Controller) *MockHealthClient {
+	mock := &MockHealthClient{ctrl: ctrl}
+	mock.recorder = &MockHealthClientMockRecorder{mock}
+	return mock
+}

api/client/builder/client.go

@@ -57,8 +57,8 @@ func (*requestLogger) observe(r *http.Request) (e error) {
 	b := bytes.NewBuffer(nil)
 	if r.Body == nil {
 		log.WithFields(log.Fields{
-			"body-base64": "(nil value)",
-			"url":         r.URL.String(),
+			"bodyBase64": "(nil value)",
+			"url":        r.URL.String(),
 		}).Info("builder http request")
 		return nil
 	}
@@ -74,8 +74,8 @@ func (*requestLogger) observe(r *http.Request) (e error) {
 	}
 	r.Body = io.NopCloser(b)
 	log.WithFields(log.Fields{
-		"body-base64": string(body),
-		"url":         r.URL.String(),
+		"bodyBase64": string(body),
+		"url":        r.URL.String(),
 	}).Info("builder http request")
 
 	return nil
@@ -304,6 +304,8 @@ func (c *Client) SubmitBlindedBlock(ctx context.Context, sb interfaces.ReadOnlyS
 		}
 		versionOpt := func(r *http.Request) {
 			r.Header.Add("Eth-Consensus-Version", version.String(version.Bellatrix))
+			r.Header.Set("Content-Type", "application/json")
+			r.Header.Set("Accept", "application/json")
 		}
 		rb, err := c.do(ctx, http.MethodPost, postBlindedBeaconBlockPath, bytes.NewBuffer(body), versionOpt)
 
@@ -341,6 +343,8 @@ func (c *Client) SubmitBlindedBlock(ctx context.Context, sb interfaces.ReadOnlyS
 		}
 		versionOpt := func(r *http.Request) {
 			r.Header.Add("Eth-Consensus-Version", version.String(version.Capella))
+			r.Header.Set("Content-Type", "application/json")
+			r.Header.Set("Accept", "application/json")
 		}
 		rb, err := c.do(ctx, http.MethodPost, postBlindedBeaconBlockPath, bytes.NewBuffer(body), versionOpt)
 
@@ -379,6 +383,8 @@ func (c *Client) SubmitBlindedBlock(ctx context.Context, sb interfaces.ReadOnlyS
 
 		versionOpt := func(r *http.Request) {
 			r.Header.Add("Eth-Consensus-Version", version.String(version.Deneb))
+			r.Header.Set("Content-Type", "application/json")
+			r.Header.Set("Accept", "application/json")
 		}
 		rb, err := c.do(ctx, http.MethodPost, postBlindedBeaconBlockPath, bytes.NewBuffer(body), versionOpt)
 		if err != nil {

api/client/builder/client_test.go

@@ -321,6 +321,8 @@ func TestSubmitBlindedBlock(t *testing.T) {
 			Transport: roundtrip(func(r *http.Request) (*http.Response, error) {
 				require.Equal(t, postBlindedBeaconBlockPath, r.URL.Path)
 				require.Equal(t, "bellatrix", r.Header.Get("Eth-Consensus-Version"))
+				require.Equal(t, "application/json", r.Header.Get("Content-Type"))
+				require.Equal(t, "application/json", r.Header.Get("Accept"))
 				return &http.Response{
 					StatusCode: http.StatusOK,
 					Body:       io.NopCloser(bytes.NewBufferString(testExampleExecutionPayload)),
@@ -347,6 +349,8 @@ func TestSubmitBlindedBlock(t *testing.T) {
 			Transport: roundtrip(func(r *http.Request) (*http.Response, error) {
 				require.Equal(t, postBlindedBeaconBlockPath, r.URL.Path)
 				require.Equal(t, "capella", r.Header.Get("Eth-Consensus-Version"))
+				require.Equal(t, "application/json", r.Header.Get("Content-Type"))
+				require.Equal(t, "application/json", r.Header.Get("Accept"))
 				return &http.Response{
 					StatusCode: http.StatusOK,
 					Body:       io.NopCloser(bytes.NewBufferString(testExampleExecutionPayloadCapella)),
@@ -376,6 +380,8 @@ func TestSubmitBlindedBlock(t *testing.T) {
 			Transport: roundtrip(func(r *http.Request) (*http.Response, error) {
 				require.Equal(t, postBlindedBeaconBlockPath, r.URL.Path)
 				require.Equal(t, "deneb", r.Header.Get("Eth-Consensus-Version"))
+				require.Equal(t, "application/json", r.Header.Get("Content-Type"))
+				require.Equal(t, "application/json", r.Header.Get("Accept"))
 				var req structs.SignedBlindedBeaconBlockDeneb
 				err := json.NewDecoder(r.Body).Decode(&req)
 				require.NoError(t, err)

api/client/errors.go

@@ -21,6 +21,9 @@ var ErrNotFound = errors.Wrap(ErrNotOK, "recv 404 NotFound response from API")
 // ErrInvalidNodeVersion indicates that the /eth/v1/node/version API response format was not recognized.
 var ErrInvalidNodeVersion = errors.New("invalid node version response")
 
+// ErrConnectionIssue represents a connection problem.
+var ErrConnectionIssue = errors.New("could not connect")
+
 // Non200Err is a function that parses an HTTP response to handle responses that are not 200 with a formatted error.
 func Non200Err(response *http.Response) error {
 	bodyBytes, err := io.ReadAll(response.Body)

api/client/event/BUILD.bazel

@@ -0,0 +1,24 @@
+load("@prysm//tools/go:def.bzl", "go_library", "go_test")
+
+go_library(
+    name = "go_default_library",
+    srcs = ["event_stream.go"],
+    importpath = "github.com/prysmaticlabs/prysm/v5/api/client/event",
+    visibility = ["//visibility:public"],
+    deps = [
+        "//api:go_default_library",
+        "//api/client:go_default_library",
+        "@com_github_pkg_errors//:go_default_library",
+        "@com_github_sirupsen_logrus//:go_default_library",
+    ],
+)
+
+go_test(
+    name = "go_default_test",
+    srcs = ["event_stream_test.go"],
+    embed = [":go_default_library"],
+    deps = [
+        "//testing/require:go_default_library",
+        "@com_github_sirupsen_logrus//:go_default_library",
+    ],
+)

api/client/event/event_stream.go

@@ -0,0 +1,148 @@
+package event
+
+import (
+	"bufio"
+	"context"
+	"net/http"
+	"net/url"
+	"strings"
+
+	"github.com/pkg/errors"
+	"github.com/prysmaticlabs/prysm/v5/api"
+	"github.com/prysmaticlabs/prysm/v5/api/client"
+	log "github.com/sirupsen/logrus"
+)
+
+const (
+	EventHead                        = "head"
+	EventBlock                       = "block"
+	EventAttestation                 = "attestation"
+	EventVoluntaryExit               = "voluntary_exit"
+	EventBlsToExecutionChange        = "bls_to_execution_change"
+	EventProposerSlashing            = "proposer_slashing"
+	EventAttesterSlashing            = "attester_slashing"
+	EventFinalizedCheckpoint         = "finalized_checkpoint"
+	EventChainReorg                  = "chain_reorg"
+	EventContributionAndProof        = "contribution_and_proof"
+	EventLightClientFinalityUpdate   = "light_client_finality_update"
+	EventLightClientOptimisticUpdate = "light_client_optimistic_update"
+	EventPayloadAttributes           = "payload_attributes"
+	EventBlobSidecar                 = "blob_sidecar"
+	EventError                       = "error"
+	EventConnectionError             = "connection_error"
+)
+
+var (
+	_ = EventStreamClient(&EventStream{})
+)
+
+var DefaultEventTopics = []string{EventHead}
+
+type EventStreamClient interface {
+	Subscribe(eventsChannel chan<- *Event)
+}
+
+type Event struct {
+	EventType string
+	Data      []byte
+}
+
+// EventStream is responsible for subscribing to the Beacon API events endpoint
+// and dispatching received events to subscribers.
+type EventStream struct {
+	ctx        context.Context
+	httpClient *http.Client
+	host       string
+	topics     []string
+}
+
+func NewEventStream(ctx context.Context, httpClient *http.Client, host string, topics []string) (*EventStream, error) {
+	// Check if the host is a valid URL
+	_, err := url.ParseRequestURI(host)
+	if err != nil {
+		return nil, err
+	}
+	if len(topics) == 0 {
+		return nil, errors.New("no topics provided")
+	}
+
+	return &EventStream{
+		ctx:        ctx,
+		httpClient: httpClient,
+		host:       host,
+		topics:     topics,
+	}, nil
+}
+
+func (h *EventStream) Subscribe(eventsChannel chan<- *Event) {
+	allTopics := strings.Join(h.topics, ",")
+	log.WithField("topics", allTopics).Info("Listening to Beacon API events")
+	fullUrl := h.host + "/eth/v1/events?topics=" + allTopics
+	req, err := http.NewRequestWithContext(h.ctx, http.MethodGet, fullUrl, nil)
+	if err != nil {
+		eventsChannel <- &Event{
+			EventType: EventConnectionError,
+			Data:      []byte(errors.Wrap(err, "failed to create HTTP request").Error()),
+		}
+	}
+	req.Header.Set("Accept", api.EventStreamMediaType)
+	req.Header.Set("Connection", api.KeepAlive)
+	resp, err := h.httpClient.Do(req)
+	if err != nil {
+		eventsChannel <- &Event{
+			EventType: EventConnectionError,
+			Data:      []byte(errors.Wrap(err, client.ErrConnectionIssue.Error()).Error()),
+		}
+	}
+
+	defer func() {
+		if closeErr := resp.Body.Close(); closeErr != nil {
+			log.WithError(closeErr).Error("Failed to close events response body")
+		}
+	}()
+	// Create a new scanner to read lines from the response body
+	scanner := bufio.NewScanner(resp.Body)
+
+	var eventType, data string // Variables to store event type and data
+
+	// Iterate over lines of the event stream
+	for scanner.Scan() {
+		select {
+		case <-h.ctx.Done():
+			log.Info("Context canceled, stopping event stream")
+			close(eventsChannel)
+			return
+		default:
+			line := scanner.Text() // TODO(13730): scanner does not handle /r and does not fully adhere to https://html.spec.whatwg.org/multipage/server-sent-events.html#the-eventsource-interface
+			// Handle the event based on your specific format
+			if line == "" {
+				// Empty line indicates the end of an event
+				if eventType != "" && data != "" {
+					// Process the event when both eventType and data are set
+					eventsChannel <- &Event{EventType: eventType, Data: []byte(data)}
+				}
+
+				// Reset eventType and data for the next event
+				eventType, data = "", ""
+				continue
+			}
+			et, ok := strings.CutPrefix(line, "event: ")
+			if ok {
+				// Extract event type from the "event" field
+				eventType = et
+			}
+			d, ok := strings.CutPrefix(line, "data: ")
+			if ok {
+				// Extract data from the "data" field
+				data = d
+			}
+		}
+	}
+
+	if err := scanner.Err(); err != nil {
+		eventsChannel <- &Event{
+			EventType: EventConnectionError,
+			Data:      []byte(errors.Wrap(err, errors.Wrap(client.ErrConnectionIssue, "scanner failed").Error()).Error()),
+		}
+	}
+}

api/client/event/event_stream_test.go

@@ -0,0 +1,80 @@
+package event
+
+import (
+	"context"
+	"fmt"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+	"time"
+
+	"github.com/prysmaticlabs/prysm/v5/testing/require"
+	log "github.com/sirupsen/logrus"
+)
+
+func TestNewEventStream(t *testing.T) {
+	validURL := "http://localhost:8080"
+	invalidURL := "://invalid"
+	topics := []string{"topic1", "topic2"}
+
+	tests := []struct {
+		name    string
+		host    string
+		topics  []string
+		wantErr bool
+	}{
+		{"Valid input", validURL, topics, false},
+		{"Invalid URL", invalidURL, topics, true},
+		{"No topics", validURL, []string{}, true},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			_, err := NewEventStream(context.Background(), &http.Client{}, tt.host, tt.topics)
+			if (err != nil) != tt.wantErr {
+				t.Errorf("NewEventStream() error = %v, wantErr %v", err, tt.wantErr)
+			}
+		})
+	}
+}
+
+func TestEventStream(t *testing.T) {
+	mux := http.NewServeMux()
+	mux.HandleFunc("/eth/v1/events", func(w http.ResponseWriter, r *http.Request) {
+		flusher, ok := w.(http.Flusher)
+		require.Equal(t, true, ok)
+		for i := 1; i <= 2; i++ {
+			_, err := fmt.Fprintf(w, "event: head\ndata: data%d\n\n", i)
+			require.NoError(t, err)
+			flusher.Flush()                    // Trigger flush to simulate streaming data
+			time.Sleep(100 * time.Millisecond) // Simulate delay between events
+		}
+	})
+	server := httptest.NewServer(mux)
+	defer server.Close()
+
+	topics := []string{"head"}
+	eventsChannel := make(chan *Event, 1)
+	stream, err := NewEventStream(context.Background(), http.DefaultClient, server.URL, topics)
+	require.NoError(t, err)
+	go stream.Subscribe(eventsChannel)
+
+	// Collect events
+	var events []*Event
+
+	for len(events) != 2 {
+		select {
+		case event := <-eventsChannel:
+			log.Info(event)
+			events = append(events, event)
+		}
+	}
+
+	// Assertions to verify the events content
+	expectedData := []string{"data1", "data2"}
+	for i, event := range events {
+		if string(event.Data) != expectedData[i] {
+			t.Errorf("Expected event data %q, got %q", expectedData[i], string(event.Data))
+		}
+	}
+}

async/abool/abool_test.go

@@ -88,7 +88,7 @@ func TestToggle(t *testing.T) {
 	}
 }
 
-func TestToogleMultipleTimes(t *testing.T) {
+func TestToggleMultipleTimes(t *testing.T) {
 	t.Parallel()
 
 	v := New()
@@ -101,16 +101,16 @@ func TestToogleMultipleTimes(t *testing.T) {
 
 		expected := i%2 != 0
 		if v.IsSet() != expected {
-			t.Fatalf("AtomicBool.Toogle() doesn't work after %d calls, expected: %v, got %v", i, expected, v.IsSet())
+			t.Fatalf("AtomicBool.Toggle() doesn't work after %d calls, expected: %v, got %v", i, expected, v.IsSet())
 		}
 
 		if pre == v.IsSet() {
-			t.Fatalf("AtomicBool.Toogle() returned wrong value at the %dth calls, expected: %v, got %v", i, !v.IsSet(), pre)
+			t.Fatalf("AtomicBool.Toggle() returned wrong value at the %dth calls, expected: %v, got %v", i, !v.IsSet(), pre)
 		}
 	}
 }
 
-func TestToogleAfterOverflow(t *testing.T) {
+func TestToggleAfterOverflow(t *testing.T) {
 	t.Parallel()
 
 	var value int32 = math.MaxInt32
@@ -122,7 +122,7 @@ func TestToogleAfterOverflow(t *testing.T) {
 	v.Toggle()
 	expected := math.MaxInt32%2 == 0
 	if v.IsSet() != expected {
-		t.Fatalf("AtomicBool.Toogle() doesn't work after overflow, expected: %v, got %v", expected, v.IsSet())
+		t.Fatalf("AtomicBool.Toggle() doesn't work after overflow, expected: %v, got %v", expected, v.IsSet())
 	}
 
 	// make sure overflow happened
@@ -135,7 +135,7 @@ func TestToogleAfterOverflow(t *testing.T) {
 	v.Toggle()
 	expected = !expected
 	if v.IsSet() != expected {
-		t.Fatalf("AtomicBool.Toogle() doesn't work after the second call after overflow, expected: %v, got %v", expected, v.IsSet())
+		t.Fatalf("AtomicBool.Toggle() doesn't work after the second call after overflow, expected: %v, got %v", expected, v.IsSet())
 	}
 }
 

async/event/feed.go

@@ -20,6 +20,7 @@ package event
 import (
 	"errors"
 	"reflect"
+	"slices"
 	"sync"
 )
 
@@ -219,12 +220,9 @@ type caseList []reflect.SelectCase
 
 // find returns the index of a case containing the given channel.
 func (cs caseList) find(channel interface{}) int {
-	for i, cas := range cs {
-		if cas.Chan.Interface() == channel {
-			return i
-		}
-	}
-	return -1
+	return slices.IndexFunc(cs, func(selectCase reflect.SelectCase) bool {
+		return selectCase.Chan.Interface() == channel
+	})
 }
 
 // delete removes the given case from cs.

async/scatter.go

@@ -63,7 +63,7 @@ func Scatter(inputLen int, sFunc func(int, int, *sync.RWMutex) (interface{}, err
 	return results, nil
 }
 
-// calculateChunkSize calculates a suitable chunk size for the purposes of parallelisation.
+// calculateChunkSize calculates a suitable chunk size for the purposes of parallelization.
 func calculateChunkSize(items int) int {
 	// Start with a simple even split
 	chunkSize := items / runtime.GOMAXPROCS(0)

bazel.sh

@@ -2,7 +2,7 @@
 
 # This script serves as a wrapper around bazel to limit the scope of environment variables that
 # may change the action output. Using this script should result in a higher cache hit ratio for
-# cached actions with a more heremtic build.
+# cached actions with a more hermetic build.
 
 env -i \
  PATH=/usr/bin:/bin \

beacon-chain/blockchain/chain_info.go

@@ -11,7 +11,6 @@ import (
 	"github.com/prysmaticlabs/prysm/v5/beacon-chain/core/helpers"
 	f "github.com/prysmaticlabs/prysm/v5/beacon-chain/forkchoice"
 	doublylinkedtree "github.com/prysmaticlabs/prysm/v5/beacon-chain/forkchoice/doubly-linked-tree"
-	forkchoicetypes "github.com/prysmaticlabs/prysm/v5/beacon-chain/forkchoice/types"
 	"github.com/prysmaticlabs/prysm/v5/beacon-chain/state"
 	fieldparams "github.com/prysmaticlabs/prysm/v5/config/fieldparams"
 	"github.com/prysmaticlabs/prysm/v5/config/params"
@@ -399,14 +398,6 @@ func (s *Service) InForkchoice(root [32]byte) bool {
 	return s.cfg.ForkChoiceStore.HasNode(root)
 }
 
-// IsViableForCheckpoint returns whether the given checkpoint is a checkpoint in any
-// chain known to forkchoice
-func (s *Service) IsViableForCheckpoint(cp *forkchoicetypes.Checkpoint) (bool, error) {
-	s.cfg.ForkChoiceStore.RLock()
-	defer s.cfg.ForkChoiceStore.RUnlock()
-	return s.cfg.ForkChoiceStore.IsViableForCheckpoint(cp)
-}
-
 // IsOptimisticForRoot takes the root as argument instead of the current head
 // and returns true if it is optimistic.
 func (s *Service) IsOptimisticForRoot(ctx context.Context, root [32]byte) (bool, error) {
@@ -563,3 +554,9 @@ func (s *Service) RecentBlockSlot(root [32]byte) (primitives.Slot, error) {
 func (s *Service) inRegularSync() bool {
 	return s.cfg.SyncChecker.Synced()
 }
+
+// validating returns true if the beacon is tracking some validators that have
+// registered for proposing.
+func (s *Service) validating() bool {
+	return s.cfg.TrackedValidatorsCache.Validating()
+}

beacon-chain/blockchain/head_sync_committee_info_test.go

@@ -61,7 +61,7 @@ func TestService_headNextSyncCommitteeIndices(t *testing.T) {
 	indices, err := c.headNextSyncCommitteeIndices(context.Background(), 0, primitives.Slot(slot))
 	require.NoError(t, err)
 
-	// NextSyncCommittee should be be empty after `ProcessSyncCommitteeUpdates`. Validator should get indices.
+	// NextSyncCommittee should be empty after `ProcessSyncCommitteeUpdates`. Validator should get indices.
 	require.NotEqual(t, 0, len(indices))
 }
 

beacon-chain/blockchain/head_test.go

@@ -63,7 +63,7 @@ func TestSaveHead_Different(t *testing.T) {
 	wsb := util.SaveBlock(t, context.Background(), service.cfg.BeaconDB, newHeadSignedBlock)
 	newRoot, err := newHeadBlock.HashTreeRoot()
 	require.NoError(t, err)
-	state, blkRoot, err = prepareForkchoiceState(ctx, wsb.Block().Slot()-1, wsb.Block().ParentRoot(), service.cfg.ForkChoiceStore.CachedHeadRoot(), [32]byte{}, ojc, ofc)
+	state, blkRoot, err = prepareForkchoiceState(ctx, slots.PrevSlot(wsb.Block().Slot()), wsb.Block().ParentRoot(), service.cfg.ForkChoiceStore.CachedHeadRoot(), [32]byte{}, ojc, ofc)
 	require.NoError(t, err)
 	require.NoError(t, service.cfg.ForkChoiceStore.InsertNode(ctx, state, blkRoot))
 
@@ -238,7 +238,7 @@ func TestRetrieveHead_ReadOnly(t *testing.T) {
 	wsb := util.SaveBlock(t, context.Background(), service.cfg.BeaconDB, newHeadSignedBlock)
 	newRoot, err := newHeadBlock.HashTreeRoot()
 	require.NoError(t, err)
-	state, blkRoot, err := prepareForkchoiceState(ctx, wsb.Block().Slot()-1, wsb.Block().ParentRoot(), service.cfg.ForkChoiceStore.CachedHeadRoot(), [32]byte{}, ojc, ofc)
+	state, blkRoot, err := prepareForkchoiceState(ctx, slots.PrevSlot(wsb.Block().Slot()), wsb.Block().ParentRoot(), service.cfg.ForkChoiceStore.CachedHeadRoot(), [32]byte{}, ojc, ofc)
 	require.NoError(t, err)
 	require.NoError(t, service.cfg.ForkChoiceStore.InsertNode(ctx, state, blkRoot))
 

beacon-chain/blockchain/log.go

@@ -82,19 +82,20 @@ func logBlockSyncStatus(block interfaces.ReadOnlyBeaconBlock, blockRoot [32]byte
 	if level >= logrus.DebugLevel {
 		parentRoot := block.ParentRoot()
 		lf := logrus.Fields{
-			"slot":                      block.Slot(),
-			"slotInEpoch":               block.Slot() % params.BeaconConfig().SlotsPerEpoch,
-			"block":                     fmt.Sprintf("0x%s...", hex.EncodeToString(blockRoot[:])[:8]),
-			"epoch":                     slots.ToEpoch(block.Slot()),
-			"justifiedEpoch":            justified.Epoch,
-			"justifiedRoot":             fmt.Sprintf("0x%s...", hex.EncodeToString(justified.Root)[:8]),
-			"finalizedEpoch":            finalized.Epoch,
-			"finalizedRoot":             fmt.Sprintf("0x%s...", hex.EncodeToString(finalized.Root)[:8]),
-			"parentRoot":                fmt.Sprintf("0x%s...", hex.EncodeToString(parentRoot[:])[:8]),
-			"version":                   version.String(block.Version()),
-			"sinceSlotStartTime":        prysmTime.Now().Sub(startTime),
-			"chainServiceProcessedTime": prysmTime.Now().Sub(receivedTime) - daWaitedTime,
-			"deposits":                  len(block.Body().Deposits()),
+			"slot":                       block.Slot(),
+			"slotInEpoch":                block.Slot() % params.BeaconConfig().SlotsPerEpoch,
+			"block":                      fmt.Sprintf("0x%s...", hex.EncodeToString(blockRoot[:])[:8]),
+			"epoch":                      slots.ToEpoch(block.Slot()),
+			"justifiedEpoch":             justified.Epoch,
+			"justifiedRoot":              fmt.Sprintf("0x%s...", hex.EncodeToString(justified.Root)[:8]),
+			"finalizedEpoch":             finalized.Epoch,
+			"finalizedRoot":              fmt.Sprintf("0x%s...", hex.EncodeToString(finalized.Root)[:8]),
+			"parentRoot":                 fmt.Sprintf("0x%s...", hex.EncodeToString(parentRoot[:])[:8]),
+			"version":                    version.String(block.Version()),
+			"sinceSlotStartTime":         prysmTime.Now().Sub(startTime),
+			"chainServiceProcessedTime":  prysmTime.Now().Sub(receivedTime) - daWaitedTime,
+			"dataAvailabilityWaitedTime": daWaitedTime,
+			"deposits":                   len(block.Body().Deposits()),
 		}
 		log.WithFields(lf).Debug("Synced new block")
 	} else {

beacon-chain/blockchain/pow_block_test.go

@@ -163,7 +163,7 @@ func Test_getBlkParentHashAndTD(t *testing.T) {
 	parentHash, totalDifficulty, err := service.getBlkParentHashAndTD(ctx, h[:])
 	require.NoError(t, err)
 	require.Equal(t, p, bytesutil.ToBytes32(parentHash))
-	require.Equal(t, td, totalDifficulty.String())
+	require.Equal(t, td, totalDifficulty.Hex())
 
 	_, _, err = service.getBlkParentHashAndTD(ctx, []byte{'c'})
 	require.ErrorContains(t, "could not get pow block: block not found", err)

beacon-chain/blockchain/process_attestation_helpers.go

@@ -18,17 +18,63 @@ import (
 	"github.com/prysmaticlabs/prysm/v5/time/slots"
 )
 
+func (s *Service) getRecentPreState(ctx context.Context, c *ethpb.Checkpoint) state.ReadOnlyBeaconState {
+	headEpoch := slots.ToEpoch(s.HeadSlot())
+	if c.Epoch < headEpoch {
+		return nil
+	}
+	if !s.cfg.ForkChoiceStore.IsCanonical([32]byte(c.Root)) {
+		return nil
+	}
+	if c.Epoch == headEpoch {
+		targetSlot, err := s.cfg.ForkChoiceStore.Slot([32]byte(c.Root))
+		if err != nil {
+			return nil
+		}
+		if slots.ToEpoch(targetSlot)+1 < headEpoch {
+			return nil
+		}
+		st, err := s.HeadStateReadOnly(ctx)
+		if err != nil {
+			return nil
+		}
+		return st
+	}
+	slot, err := slots.EpochStart(c.Epoch)
+	if err != nil {
+		return nil
+	}
+	// Try if we have already set the checkpoint cache
+	epochKey := strconv.FormatUint(uint64(c.Epoch), 10 /* base 10 */)
+	lock := async.NewMultilock(string(c.Root) + epochKey)
+	lock.Lock()
+	defer lock.Unlock()
+	cachedState, err := s.checkpointStateCache.StateByCheckpoint(c)
+	if err != nil {
+		return nil
+	}
+	if cachedState != nil && !cachedState.IsNil() {
+		return cachedState
+	}
+	st, err := s.HeadState(ctx)
+	if err != nil {
+		return nil
+	}
+	st, err = transition.ProcessSlotsUsingNextSlotCache(ctx, st, c.Root, slot)
+	if err != nil {
+		return nil
+	}
+	if err := s.checkpointStateCache.AddCheckpointState(c, st); err != nil {
+		return nil
+	}
+	return st
+}
+
 // getAttPreState retrieves the att pre state by either from the cache or the DB.
 func (s *Service) getAttPreState(ctx context.Context, c *ethpb.Checkpoint) (state.ReadOnlyBeaconState, error) {
 	// If the attestation is recent and canonical we can use the head state to compute the shuffling.
-	headEpoch := slots.ToEpoch(s.HeadSlot())
-	if c.Epoch == headEpoch {
-		targetSlot, err := s.cfg.ForkChoiceStore.Slot([32]byte(c.Root))
-		if err == nil && slots.ToEpoch(targetSlot)+1 >= headEpoch {
-			if s.cfg.ForkChoiceStore.IsCanonical([32]byte(c.Root)) {
-				return s.HeadStateReadOnly(ctx)
-			}
-		}
+	if st := s.getRecentPreState(ctx, c); st != nil {
+		return st, nil
 	}
 	// Use a multilock to allow scoped holding of a mutex by a checkpoint root + epoch
 	// allowing us to behave smarter in terms of how this function is used concurrently.

beacon-chain/blockchain/process_attestation_test.go

@@ -146,6 +146,28 @@ func TestStore_OnAttestation_Ok_DoublyLinkedTree(t *testing.T) {
 	require.NoError(t, service.OnAttestation(ctx, att[0], 0))
 }
 
+func TestService_GetRecentPreState(t *testing.T) {
+	service, _ := minimalTestService(t)
+	ctx := context.Background()
+
+	s, err := util.NewBeaconState()
+	require.NoError(t, err)
+	ckRoot := bytesutil.PadTo([]byte{'A'}, fieldparams.RootLength)
+	cp0 := &ethpb.Checkpoint{Epoch: 0, Root: ckRoot}
+	err = s.SetFinalizedCheckpoint(cp0)
+	require.NoError(t, err)
+
+	st, root, err := prepareForkchoiceState(ctx, 31, [32]byte(ckRoot), [32]byte{}, [32]byte{'R'}, cp0, cp0)
+	require.NoError(t, err)
+	require.NoError(t, service.cfg.ForkChoiceStore.InsertNode(ctx, st, root))
+	service.head = &head{
+		root:  [32]byte(ckRoot),
+		state: s,
+		slot:  31,
+	}
+	require.NotNil(t, service.getRecentPreState(ctx, &ethpb.Checkpoint{Epoch: 1, Root: ckRoot}))
+}
+
 func TestService_GetAttPreState_Concurrency(t *testing.T) {
 	service, _ := minimalTestService(t)
 	ctx := context.Background()

beacon-chain/blockchain/process_block.go

@@ -6,6 +6,7 @@ import (
 	"time"
 
 	"github.com/pkg/errors"
+	"github.com/sirupsen/logrus"
 	"go.opencensus.io/trace"
 
 	"github.com/prysmaticlabs/prysm/v5/beacon-chain/core/blocks"
@@ -307,16 +308,16 @@ func (s *Service) updateEpochBoundaryCaches(ctx context.Context, st state.Beacon
 	if err := helpers.UpdateProposerIndicesInCache(ctx, st, e); err != nil {
 		return errors.Wrap(err, "could not update proposer index cache")
 	}
-	go func() {
+	go func(ep primitives.Epoch) {
 		// Use a custom deadline here, since this method runs asynchronously.
 		// We ignore the parent method's context and instead create a new one
 		// with a custom deadline, therefore using the background context instead.
 		slotCtx, cancel := context.WithTimeout(context.Background(), slotDeadline)
 		defer cancel()
-		if err := helpers.UpdateCommitteeCache(slotCtx, st, e+1); err != nil {
+		if err := helpers.UpdateCommitteeCache(slotCtx, st, ep+1); err != nil {
 			log.WithError(err).Warn("Could not update committee cache")
 		}
-	}()
+	}(e)
 	// The latest block header is from the previous epoch
 	r, err := st.LatestBlockHeader().HashTreeRoot()
 	if err != nil {
@@ -558,6 +559,20 @@ func (s *Service) isDataAvailable(ctx context.Context, root [32]byte, signed int
 	// The gossip handler for blobs writes the index of each verified blob referencing the given
 	// root to the channel returned by blobNotifiers.forRoot.
 	nc := s.blobNotifiers.forRoot(root)
+
+	// Log for DA checks that cross over into the next slot; helpful for debugging.
+	nextSlot := slots.BeginsAt(signed.Block().Slot()+1, s.genesisTime)
+	// Avoid logging if DA check is called after next slot start.
+	if nextSlot.After(time.Now()) {
+		nst := time.AfterFunc(time.Until(nextSlot), func() {
+			if len(missing) == 0 {
+				return
+			}
+			log.WithFields(daCheckLogFields(root, signed.Block().Slot(), expected, len(missing))).
+				Error("Still waiting for DA check at slot end.")
+		})
+		defer nst.Stop()
+	}
 	for {
 		select {
 		case idx := <-nc:
@@ -571,11 +586,20 @@ func (s *Service) isDataAvailable(ctx context.Context, root [32]byte, signed int
 			s.blobNotifiers.delete(root)
 			return nil
 		case <-ctx.Done():
-			return errors.Wrap(ctx.Err(), "context deadline waiting for blob sidecars")
+			return errors.Wrapf(ctx.Err(), "context deadline waiting for blob sidecars slot: %d, BlockRoot: %#x", block.Slot(), root)
 		}
 	}
 }
 
+func daCheckLogFields(root [32]byte, slot primitives.Slot, expected, missing int) logrus.Fields {
+	return logrus.Fields{
+		"slot":          slot,
+		"root":          fmt.Sprintf("%#x", root),
+		"blobsExpected": expected,
+		"blobsWaiting":  missing,
+	}
+}
+
 // lateBlockTasks  is called 4 seconds into the slot and performs tasks
 // related to late blocks. It emits a MissedSlot state feed event.
 // It calls FCU and sets the right attributes if we are proposing next slot

beacon-chain/blockchain/process_block_helpers.go

@@ -60,7 +60,7 @@ func (s *Service) getFCUArgsEarlyBlock(cfg *postBlockProcessConfig, fcuArgs *fcu
 
 // logNonCanonicalBlockReceived prints a message informing that the received
 // block is not the head of the chain. It requires the caller holds a lock on
-// Foprkchoice.
+// Forkchoice.
 func (s *Service) logNonCanonicalBlockReceived(blockRoot [32]byte, headRoot [32]byte) {
 	receivedWeight, err := s.cfg.ForkChoiceStore.Weight(blockRoot)
 	if err != nil {

beacon-chain/blockchain/process_block_test.go

@@ -1531,6 +1531,7 @@ func TestStore_NoViableHead_NewPayload(t *testing.T) {
 // 12 and recover. Notice that it takes two epochs to fully recover, and we stay
 // optimistic for the whole time.
 func TestStore_NoViableHead_Liveness(t *testing.T) {
+	t.Skip("Requires #13664 to be fixed")
 	params.SetupTestConfigCleanup(t)
 	config := params.BeaconConfig()
 	config.SlotsPerEpoch = 6
@@ -2114,7 +2115,7 @@ func TestMissingIndices(t *testing.T) {
 	for _, c := range cases {
 		bm, bs := filesystem.NewEphemeralBlobStorageWithMocker(t)
 		t.Run(c.name, func(t *testing.T) {
-			require.NoError(t, bm.CreateFakeIndices(c.root, c.present))
+			require.NoError(t, bm.CreateFakeIndices(c.root, 0, c.present...))
 			missing, err := missingIndices(bs, c.root, c.expected)
 			if c.err != nil {
 				require.ErrorIs(t, err, c.err)

beacon-chain/blockchain/receive_attestation.go

@@ -95,7 +95,9 @@ func (s *Service) spawnProcessAttestationsRoutine() {
 				return
 			case slotInterval := <-ticker.C():
 				if slotInterval.Interval > 0 {
-					s.UpdateHead(s.ctx, slotInterval.Slot+1)
+					if s.validating() {
+						s.UpdateHead(s.ctx, slotInterval.Slot+1)
+					}
 				} else {
 					s.cfg.ForkChoiceStore.Lock()
 					if err := s.cfg.ForkChoiceStore.NewSlot(s.ctx, slotInterval.Slot); err != nil {

beacon-chain/blockchain/receive_block.go

@@ -32,6 +32,9 @@ import (
 // This defines how many epochs since finality the run time will begin to save hot state on to the DB.
 var epochsSinceFinalitySaveHotStateDB = primitives.Epoch(100)
 
+// This defines how many epochs since finality the run time will begin to expand our respective cache sizes.
+var epochsSinceFinalityExpandCache = primitives.Epoch(4)
+
 // BlockReceiver interface defines the methods of chain service for receiving and processing new blocks.
 type BlockReceiver interface {
 	ReceiveBlock(ctx context.Context, block interfaces.ReadOnlySignedBeaconBlock, blockRoot [32]byte, avs das.AvailabilityStore) error
@@ -94,6 +97,7 @@ func (s *Service) ReceiveBlock(ctx context.Context, block interfaces.ReadOnlySig
 	eg, _ := errgroup.WithContext(ctx)
 	var postState state.BeaconState
 	eg.Go(func() error {
+		var err error
 		postState, err = s.validateStateTransition(ctx, preState, blockCopy)
 		if err != nil {
 			return errors.Wrap(err, "failed to validate consensus state transition function")
@@ -102,6 +106,7 @@ func (s *Service) ReceiveBlock(ctx context.Context, block interfaces.ReadOnlySig
 	})
 	var isValidPayload bool
 	eg.Go(func() error {
+		var err error
 		isValidPayload, err = s.validateExecutionOnBlock(ctx, preStateVersion, preStateHeader, blockCopy, blockRoot)
 		if err != nil {
 			return errors.Wrap(err, "could not notify the engine of the new payload")
@@ -165,7 +170,7 @@ func (s *Service) ReceiveBlock(ctx context.Context, block interfaces.ReadOnlySig
 	// Send finalized events and finalized deposits in the background
 	if newFinalized {
 		finalized := s.cfg.ForkChoiceStore.FinalizedCheckpoint()
-		go s.sendNewFinalizedEvent(blockCopy, postState)
+		go s.sendNewFinalizedEvent(ctx, postState)
 		depCtx, cancel := context.WithTimeout(context.Background(), depositDeadline)
 		go func() {
 			s.insertFinalizedDeposits(depCtx, finalized.Root)
@@ -188,6 +193,11 @@ func (s *Service) ReceiveBlock(ctx context.Context, block interfaces.ReadOnlySig
 		return err
 	}
 
+	// We apply the same heuristic to some of our more important caches.
+	if err := s.handleCaches(); err != nil {
+		return err
+	}
+
 	// Reports on block and fork choice metrics.
 	cp := s.cfg.ForkChoiceStore.FinalizedCheckpoint()
 	finalized := &ethpb.Checkpoint{Epoch: cp.Epoch, Root: bytesutil.SafeCopyBytes(cp.Root[:])}
@@ -361,6 +371,27 @@ func (s *Service) checkSaveHotStateDB(ctx context.Context) error {
 	return s.cfg.StateGen.DisableSaveHotStateToDB(ctx)
 }
 
+func (s *Service) handleCaches() error {
+	currentEpoch := slots.ToEpoch(s.CurrentSlot())
+	// Prevent `sinceFinality` going underflow.
+	var sinceFinality primitives.Epoch
+	finalized := s.cfg.ForkChoiceStore.FinalizedCheckpoint()
+	if finalized == nil {
+		return errNilFinalizedInStore
+	}
+	if currentEpoch > finalized.Epoch {
+		sinceFinality = currentEpoch - finalized.Epoch
+	}
+
+	if sinceFinality >= epochsSinceFinalityExpandCache {
+		helpers.ExpandCommitteeCache()
+		return nil
+	}
+
+	helpers.CompressCommitteeCache()
+	return nil
+}
+
 // This performs the state transition function and returns the poststate or an
 // error if the block fails to verify the consensus rules
 func (s *Service) validateStateTransition(ctx context.Context, preState state.BeaconState, signed interfaces.ReadOnlySignedBeaconBlock) (state.BeaconState, error) {
@@ -412,7 +443,7 @@ func (s *Service) updateFinalizationOnBlock(ctx context.Context, preState, postS
 
 // sendNewFinalizedEvent sends a new finalization checkpoint event over the
 // event feed. It needs to be called on the background
-func (s *Service) sendNewFinalizedEvent(signed interfaces.ReadOnlySignedBeaconBlock, postState state.BeaconState) {
+func (s *Service) sendNewFinalizedEvent(ctx context.Context, postState state.BeaconState) {
 	isValidPayload := false
 	s.headLock.RLock()
 	if s.head != nil {
@@ -420,8 +451,17 @@ func (s *Service) sendNewFinalizedEvent(signed interfaces.ReadOnlySignedBeaconBl
 	}
 	s.headLock.RUnlock()
 
+	blk, err := s.cfg.BeaconDB.Block(ctx, bytesutil.ToBytes32(postState.FinalizedCheckpoint().Root))
+	if err != nil {
+		log.WithError(err).Error("Could not retrieve block for finalized checkpoint root. Finalized event will not be emitted")
+		return
+	}
+	if blk == nil || blk.IsNil() || blk.Block() == nil || blk.Block().IsNil() {
+		log.WithError(err).Error("Block retrieved for finalized checkpoint root is nil. Finalized event will not be emitted")
+		return
+	}
+	stateRoot := blk.Block().StateRoot()
 	// Send an event regarding the new finalized checkpoint over a common event feed.
-	stateRoot := signed.Block().StateRoot()
 	s.cfg.StateNotifier.StateFeed().Send(&feed.Event{
 		Type: statefeed.FinalizedCheckpoint,
 		Data: &ethpbv1.EventFinalizedCheckpoint{

beacon-chain/blockchain/receive_block_test.go

@@ -8,12 +8,14 @@ import (
 
 	blockchainTesting "github.com/prysmaticlabs/prysm/v5/beacon-chain/blockchain/testing"
 	"github.com/prysmaticlabs/prysm/v5/beacon-chain/cache"
+	statefeed "github.com/prysmaticlabs/prysm/v5/beacon-chain/core/feed/state"
 	"github.com/prysmaticlabs/prysm/v5/beacon-chain/das"
 	"github.com/prysmaticlabs/prysm/v5/beacon-chain/operations/voluntaryexits"
 	"github.com/prysmaticlabs/prysm/v5/config/params"
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/blocks"
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
 	"github.com/prysmaticlabs/prysm/v5/encoding/bytesutil"
+	ethpbv1 "github.com/prysmaticlabs/prysm/v5/proto/eth/v1"
 	ethpb "github.com/prysmaticlabs/prysm/v5/proto/prysm/v1alpha1"
 	"github.com/prysmaticlabs/prysm/v5/testing/assert"
 	"github.com/prysmaticlabs/prysm/v5/testing/require"
@@ -308,6 +310,29 @@ func TestCheckSaveHotStateDB_Overflow(t *testing.T) {
 	assert.LogsDoNotContain(t, hook, "Entering mode to save hot states in DB")
 }
 
+func TestHandleCaches_EnablingLargeSize(t *testing.T) {
+	hook := logTest.NewGlobal()
+	s, _ := minimalTestService(t)
+	st := params.BeaconConfig().SlotsPerEpoch.Mul(uint64(epochsSinceFinalitySaveHotStateDB))
+	s.genesisTime = time.Now().Add(time.Duration(-1*int64(st)*int64(params.BeaconConfig().SecondsPerSlot)) * time.Second)
+
+	require.NoError(t, s.handleCaches())
+	assert.LogsContain(t, hook, "Expanding committee cache size")
+}
+
+func TestHandleCaches_DisablingLargeSize(t *testing.T) {
+	hook := logTest.NewGlobal()
+	s, _ := minimalTestService(t)
+
+	st := params.BeaconConfig().SlotsPerEpoch.Mul(uint64(epochsSinceFinalitySaveHotStateDB))
+	s.genesisTime = time.Now().Add(time.Duration(-1*int64(st)*int64(params.BeaconConfig().SecondsPerSlot)) * time.Second)
+	require.NoError(t, s.handleCaches())
+	s.genesisTime = time.Now()
+
+	require.NoError(t, s.handleCaches())
+	assert.LogsContain(t, hook, "Reducing committee cache size")
+}
+
 func TestHandleBlockBLSToExecutionChanges(t *testing.T) {
 	service, tr := minimalTestService(t)
 	pool := tr.blsPool
@@ -355,3 +380,38 @@ func TestHandleBlockBLSToExecutionChanges(t *testing.T) {
 		require.Equal(t, false, pool.ValidatorExists(idx))
 	})
 }
+
+func Test_sendNewFinalizedEvent(t *testing.T) {
+	s, _ := minimalTestService(t)
+	notifier := &blockchainTesting.MockStateNotifier{RecordEvents: true}
+	s.cfg.StateNotifier = notifier
+	finalizedSt, err := util.NewBeaconState()
+	require.NoError(t, err)
+	finalizedStRoot, err := finalizedSt.HashTreeRoot(s.ctx)
+	require.NoError(t, err)
+	b := util.NewBeaconBlock()
+	b.Block.StateRoot = finalizedStRoot[:]
+	sbb, err := blocks.NewSignedBeaconBlock(b)
+	require.NoError(t, err)
+	sbbRoot, err := sbb.Block().HashTreeRoot()
+	require.NoError(t, err)
+	require.NoError(t, s.cfg.BeaconDB.SaveBlock(s.ctx, sbb))
+	st, err := util.NewBeaconState()
+	require.NoError(t, err)
+	require.NoError(t, st.SetFinalizedCheckpoint(&ethpb.Checkpoint{
+		Epoch: 123,
+		Root:  sbbRoot[:],
+	}))
+
+	s.sendNewFinalizedEvent(s.ctx, st)
+
+	require.Equal(t, 1, len(notifier.ReceivedEvents()))
+	e := notifier.ReceivedEvents()[0]
+	assert.Equal(t, statefeed.FinalizedCheckpoint, int(e.Type))
+	fc, ok := e.Data.(*ethpbv1.EventFinalizedCheckpoint)
+	require.Equal(t, true, ok, "event has wrong data type")
+	assert.Equal(t, primitives.Epoch(123), fc.Epoch)
+	assert.DeepEqual(t, sbbRoot[:], fc.Block)
+	assert.DeepEqual(t, finalizedStRoot[:], fc.State)
+	assert.Equal(t, false, fc.ExecutionOptimistic)
+}

beacon-chain/blockchain/service.go

@@ -199,6 +199,7 @@ func NewService(ctx context.Context, opts ...Option) (*Service, error) {
 // Start a blockchain service's main event loop.
 func (s *Service) Start() {
 	saved := s.cfg.FinalizedStateAtStartUp
+	defer s.removeStartupState()
 
 	if saved != nil && !saved.IsNil() {
 		if err := s.StartFromSavedState(saved); err != nil {
@@ -418,7 +419,7 @@ func (s *Service) startFromExecutionChain() error {
 						log.Error("event data is not type *statefeed.ChainStartedData")
 						return
 					}
-					log.WithField("starttime", data.StartTime).Debug("Received chain start event")
+					log.WithField("startTime", data.StartTime).Debug("Received chain start event")
 					s.onExecutionChainStart(s.ctx, data.StartTime)
 					return
 				}
@@ -550,6 +551,10 @@ func (s *Service) hasBlock(ctx context.Context, root [32]byte) bool {
 	return s.cfg.BeaconDB.HasBlock(ctx, root)
 }
 
+func (s *Service) removeStartupState() {
+	s.cfg.FinalizedStateAtStartUp = nil
+}
+
 func spawnCountdownIfPreGenesis(ctx context.Context, genesisTime time.Time, db db.HeadAccessDatabase) {
 	currentTime := prysmTime.Now()
 	if currentTime.After(genesisTime) {

beacon-chain/cache/committee.go

@@ -17,12 +17,16 @@ import (
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
 	"github.com/prysmaticlabs/prysm/v5/container/slice"
 	mathutil "github.com/prysmaticlabs/prysm/v5/math"
+	log "github.com/sirupsen/logrus"
 )
 
 const (
 	// maxCommitteesCacheSize defines the max number of shuffled committees on per randao basis can cache.
 	// Due to reorgs and long finality, it's good to keep the old cache around for quickly switch over.
-	maxCommitteesCacheSize = int(32)
+	maxCommitteesCacheSize = int(4)
+	// expandedCommitteeCacheSize defines the expanded size of the committee cache in the event we
+	// do not have finality to deal with long forks better.
+	expandedCommitteeCacheSize = int(32)
 )
 
 var (
@@ -43,6 +47,7 @@ type CommitteeCache struct {
 	CommitteeCache *lru.Cache
 	lock           sync.RWMutex
 	inProgress     map[string]bool
+	size           int
 }
 
 // committeeKeyFn takes the seed as the key to retrieve shuffled indices of a committee in a given epoch.
@@ -67,6 +72,33 @@ func (c *CommitteeCache) Clear() {
 	defer c.lock.Unlock()
 	c.CommitteeCache = lruwrpr.New(maxCommitteesCacheSize)
 	c.inProgress = make(map[string]bool)
+	c.size = maxCommitteesCacheSize
+}
+
+// ExpandCommitteeCache expands the size of the committee cache.
+func (c *CommitteeCache) ExpandCommitteeCache() {
+	c.lock.Lock()
+	defer c.lock.Unlock()
+
+	if c.size == expandedCommitteeCacheSize {
+		return
+	}
+	c.CommitteeCache.Resize(expandedCommitteeCacheSize)
+	c.size = expandedCommitteeCacheSize
+	log.Warnf("Expanding committee cache size from %d to %d", maxCommitteesCacheSize, expandedCommitteeCacheSize)
+}
+
+// CompressCommitteeCache compresses the size of the committee cache.
+func (c *CommitteeCache) CompressCommitteeCache() {
+	c.lock.Lock()
+	defer c.lock.Unlock()
+
+	if c.size == maxCommitteesCacheSize {
+		return
+	}
+	c.CommitteeCache.Resize(maxCommitteesCacheSize)
+	c.size = maxCommitteesCacheSize
+	log.Warnf("Reducing committee cache size from %d to %d", expandedCommitteeCacheSize, maxCommitteesCacheSize)
 }
 
 // Committee fetches the shuffled indices by slot and committee index. Every list of indices

beacon-chain/cache/committee_disabled.go

@@ -74,3 +74,11 @@ func (c *FakeCommitteeCache) MarkNotInProgress(seed [32]byte) error {
 func (c *FakeCommitteeCache) Clear() {
 	return
 }
+
+func (c *FakeCommitteeCache) ExpandCommitteeCache() {
+	return
+}
+
+func (c *FakeCommitteeCache) CompressCommitteeCache() {
+	return
+}

beacon-chain/cache/depositcache/deposits_cache.go

@@ -74,10 +74,10 @@ func (dc *DepositCache) InsertDeposit(ctx context.Context, d *ethpb.Deposit, blo
 	defer span.End()
 	if d == nil {
 		log.WithFields(logrus.Fields{
-			"block":        blockNum,
-			"deposit":      d,
-			"index":        index,
-			"deposit root": hex.EncodeToString(depositRoot[:]),
+			"block":       blockNum,
+			"deposit":     d,
+			"index":       index,
+			"depositRoot": hex.EncodeToString(depositRoot[:]),
 		}).Warn("Ignoring nil deposit insertion")
 		return errors.New("nil deposit inserted into the cache")
 	}

beacon-chain/cache/depositcache/deposits_cache_test.go

@@ -804,7 +804,7 @@ func TestFinalizedDeposits_ReturnsTrieCorrectly(t *testing.T) {
 	depositTrie, err := trie.GenerateTrieFromItems(trieItems, params.BeaconConfig().DepositContractTreeDepth)
 	assert.NoError(t, err)
 
-	// Perform this in a non-sensical ordering
+	// Perform this in a nonsensical ordering
 	require.NoError(t, dc.InsertFinalizedDeposits(context.Background(), 10, [32]byte{}, 0))
 	require.NoError(t, dc.InsertFinalizedDeposits(context.Background(), 2, [32]byte{}, 0))
 	require.NoError(t, dc.InsertFinalizedDeposits(context.Background(), 3, [32]byte{}, 0))

beacon-chain/cache/depositsnapshot/deposit_cache_test.go

@@ -784,7 +784,7 @@ func TestFinalizedDeposits_ReturnsTrieCorrectly(t *testing.T) {
 	depositTrie, err := trie.GenerateTrieFromItems(trieItems, params.BeaconConfig().DepositContractTreeDepth)
 	assert.NoError(t, err)
 
-	// Perform this in a non-sensical ordering
+	// Perform this in a nonsensical ordering
 	err = dc.InsertFinalizedDeposits(context.Background(), 1, [32]byte{}, 0)
 	require.NoError(t, err)
 	err = dc.InsertFinalizedDeposits(context.Background(), 2, [32]byte{}, 0)
@@ -1189,11 +1189,3 @@ func BenchmarkDepositTree_HashTreeRootOldImplementation(b *testing.B) {
 		require.NoError(b, err)
 	}
 }
-
-func emptyEth1data() *ethpb.Eth1Data {
-	return &ethpb.Eth1Data{
-		DepositRoot:  make([]byte, 32),
-		DepositCount: 0,
-		BlockHash:    make([]byte, 32),
-	}
-}

beacon-chain/cache/depositsnapshot/deposit_inserter.go

@@ -33,10 +33,10 @@ func (c *Cache) InsertDeposit(ctx context.Context, d *ethpb.Deposit, blockNum ui
 	}
 	if d == nil {
 		log.WithFields(logrus.Fields{
-			"block":        blockNum,
-			"deposit":      d,
-			"index":        index,
-			"deposit root": hex.EncodeToString(depositRoot[:]),
+			"block":       blockNum,
+			"deposit":     d,
+			"index":       index,
+			"depositRoot": hex.EncodeToString(depositRoot[:]),
 		}).Warn("Ignoring nil deposit insertion")
 		return errors.New("nil deposit inserted into the cache")
 	}

beacon-chain/cache/depositsnapshot/deposit_tree.go

@@ -15,8 +15,6 @@ import (
 )
 
 var (
-	// ErrEmptyExecutionBlock occurs when the execution block is nil.
-	ErrEmptyExecutionBlock = errors.New("empty execution block")
 	// ErrInvalidSnapshotRoot occurs when the snapshot root does not match the calculated root.
 	ErrInvalidSnapshotRoot = errors.New("snapshot root is invalid")
 	// ErrInvalidDepositCount occurs when the value for mix in length is 0.

beacon-chain/cache/proposer_indices_type.go

@@ -1,15 +1,9 @@
 package cache
 
 import (
-	"errors"
-
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
 )
 
-// ErrNotProposerIndices will be returned when a cache object is not a pointer to
-// a ProposerIndices struct.
-var ErrNotProposerIndices = errors.New("object is not a proposer indices struct")
-
 // ProposerIndices defines the cached struct for proposer indices.
 type ProposerIndices struct {
 	BlockRoot       [32]byte

beacon-chain/cache/skip_slot_cache.go

@@ -109,10 +109,6 @@ func (c *SkipSlotCache) Get(ctx context.Context, r [32]byte) (state.BeaconState,
 // MarkInProgress a request so that any other similar requests will block on
 // Get until MarkNotInProgress is called.
 func (c *SkipSlotCache) MarkInProgress(r [32]byte) error {
-	if c.disabled {
-		return nil
-	}
-
 	c.lock.Lock()
 	defer c.lock.Unlock()
 
@@ -126,10 +122,6 @@ func (c *SkipSlotCache) MarkInProgress(r [32]byte) error {
 // MarkNotInProgress will release the lock on a given request. This should be
 // called after put.
 func (c *SkipSlotCache) MarkNotInProgress(r [32]byte) {
-	if c.disabled {
-		return
-	}
-
 	c.lock.Lock()
 	defer c.lock.Unlock()
 

beacon-chain/cache/skip_slot_cache_test.go

@@ -2,6 +2,7 @@ package cache_test
 
 import (
 	"context"
+	"sync"
 	"testing"
 
 	"github.com/prysmaticlabs/prysm/v5/beacon-chain/cache"
@@ -35,3 +36,28 @@ func TestSkipSlotCache_RoundTrip(t *testing.T) {
 	require.NoError(t, err)
 	assert.DeepEqual(t, res.ToProto(), s.ToProto(), "Expected equal protos to return from cache")
 }
+
+func TestSkipSlotCache_DisabledAndEnabled(t *testing.T) {
+	ctx := context.Background()
+	c := cache.NewSkipSlotCache()
+
+	r := [32]byte{'a'}
+	c.Disable()
+
+	require.NoError(t, c.MarkInProgress(r))
+
+	c.Enable()
+	wg := new(sync.WaitGroup)
+	wg.Add(1)
+	go func() {
+		// Get call will only terminate when
+		// it is not longer in progress.
+		obj, err := c.Get(ctx, r)
+		require.NoError(t, err)
+		require.IsNil(t, obj)
+		wg.Done()
+	}()
+
+	c.MarkNotInProgress(r)
+	wg.Wait()
+}

beacon-chain/cache/tracked_validators.go

@@ -41,3 +41,9 @@ func (t *TrackedValidatorsCache) Prune() {
 	defer t.Unlock()
 	t.trackedValidators = make(map[primitives.ValidatorIndex]TrackedValidator)
 }
+
+func (t *TrackedValidatorsCache) Validating() bool {
+	t.Lock()
+	defer t.Unlock()
+	return len(t.trackedValidators) > 0
+}

beacon-chain/core/blocks/signature.go

@@ -99,7 +99,7 @@ func VerifyBlockHeaderSignature(beaconState state.BeaconState, header *ethpb.Sig
 // VerifyBlockSignatureUsingCurrentFork verifies the proposer signature of a beacon block. This differs
 // from the above method by not using fork data from the state and instead retrieving it
 // via the respective epoch.
-func VerifyBlockSignatureUsingCurrentFork(beaconState state.ReadOnlyBeaconState, blk interfaces.ReadOnlySignedBeaconBlock) error {
+func VerifyBlockSignatureUsingCurrentFork(beaconState state.ReadOnlyBeaconState, blk interfaces.ReadOnlySignedBeaconBlock, blkRoot [32]byte) error {
 	currentEpoch := slots.ToEpoch(blk.Block().Slot())
 	fork, err := forks.Fork(currentEpoch)
 	if err != nil {
@@ -115,7 +115,9 @@ func VerifyBlockSignatureUsingCurrentFork(beaconState state.ReadOnlyBeaconState,
 	}
 	proposerPubKey := proposer.PublicKey
 	sig := blk.Signature()
-	return signing.VerifyBlockSigningRoot(proposerPubKey, sig[:], domain, blk.Block().HashTreeRoot)
+	return signing.VerifyBlockSigningRoot(proposerPubKey, sig[:], domain, func() ([32]byte, error) {
+		return blkRoot, nil
+	})
 }
 
 // BlockSignatureBatch retrieves the block signature batch from the provided block and its corresponding state.

beacon-chain/core/blocks/signature_test.go

@@ -79,11 +79,13 @@ func TestVerifyBlockSignatureUsingCurrentFork(t *testing.T) {
 	}
 	domain, err := signing.Domain(fData, 100, params.BeaconConfig().DomainBeaconProposer, bState.GenesisValidatorsRoot())
 	assert.NoError(t, err)
+	blkRoot, err := altairBlk.Block.HashTreeRoot()
+	assert.NoError(t, err)
 	rt, err := signing.ComputeSigningRoot(altairBlk.Block, domain)
 	assert.NoError(t, err)
 	sig := keys[0].Sign(rt[:]).Marshal()
 	altairBlk.Signature = sig
 	wsb, err := consensusblocks.NewSignedBeaconBlock(altairBlk)
 	require.NoError(t, err)
-	assert.NoError(t, blocks.VerifyBlockSignatureUsingCurrentFork(bState, wsb))
+	assert.NoError(t, blocks.VerifyBlockSignatureUsingCurrentFork(bState, wsb, blkRoot))
 }

beacon-chain/core/epoch/precompute/BUILD.bazel

@@ -14,6 +14,7 @@ go_library(
     visibility = [
         "//beacon-chain:__subpackages__",
         "//testing/spectest:__subpackages__",
+        "//tools:__subpackages__",
     ],
     deps = [
         "//beacon-chain/core/helpers:go_default_library",

beacon-chain/core/helpers/attestation.go

@@ -12,7 +12,6 @@ import (
 	ethpb "github.com/prysmaticlabs/prysm/v5/proto/prysm/v1alpha1"
 	prysmTime "github.com/prysmaticlabs/prysm/v5/time"
 	"github.com/prysmaticlabs/prysm/v5/time/slots"
-	log "github.com/sirupsen/logrus"
 )
 
 var (
@@ -133,9 +132,6 @@ func ComputeSubnetFromCommitteeAndSlot(activeValCount uint64, comIdx primitives.
 //
 // In the attestation must be within the range of 95 to 102 in the example above.
 func ValidateAttestationTime(attSlot primitives.Slot, genesisTime time.Time, clockDisparity time.Duration) error {
-	if err := slots.ValidateClock(attSlot, uint64(genesisTime.Unix())); err != nil {
-		return err
-	}
 	attTime, err := slots.ToTime(uint64(genesisTime.Unix()), attSlot)
 	if err != nil {
 		return err
@@ -182,24 +178,15 @@ func ValidateAttestationTime(attSlot primitives.Slot, genesisTime time.Time, clo
 	}
 
 	// EIP-7045: Starting in Deneb, allow any attestations from the current or previous epoch.
-
 	currentEpoch := slots.ToEpoch(currentSlot)
-	prevEpoch, err := currentEpoch.SafeSub(1)
-	if err != nil {
-		log.WithError(err).Debug("Ignoring underflow for a deneb attestation inclusion check in epoch 0")
-		prevEpoch = 0
-	}
-	attSlotEpoch := slots.ToEpoch(attSlot)
-	if attSlotEpoch != currentEpoch && attSlotEpoch != prevEpoch {
+	if attEpoch+1 < currentEpoch {
 		attError = fmt.Errorf(
-			"attestation epoch %d not within current epoch %d or previous epoch %d",
-			attSlot/params.BeaconConfig().SlotsPerEpoch,
+			"attestation epoch %d not within current epoch %d or previous epoch",
+			attEpoch,
 			currentEpoch,
-			prevEpoch,
 		)
 		return errors.Join(ErrTooLate, attError)
 	}
-
 	return nil
 }
 

beacon-chain/core/helpers/attestation_test.go

@@ -197,7 +197,7 @@ func Test_ValidateAttestationTime(t *testing.T) {
 					-500 * time.Duration(params.BeaconConfig().SecondsPerSlot) * time.Second,
 				).Add(200 * time.Millisecond),
 			},
-			wantedErr: "attestation epoch 8 not within current epoch 15 or previous epoch 14",
+			wantedErr: "attestation epoch 8 not within current epoch 15 or previous epoch",
 		},
 		{
 			name: "attestation.slot is well beyond current slot",
@@ -205,7 +205,7 @@ func Test_ValidateAttestationTime(t *testing.T) {
 				attSlot:     1 << 32,
 				genesisTime: prysmTime.Now().Add(-15 * time.Duration(params.BeaconConfig().SecondsPerSlot) * time.Second),
 			},
-			wantedErr: "which exceeds max allowed value relative to the local clock",
+			wantedErr: "attestation slot 4294967296 not within attestation propagation range of 0 to 15 (current slot)",
 		},
 	}
 	for _, tt := range tests {

beacon-chain/core/helpers/beacon_committee.go

@@ -391,6 +391,16 @@ func UpdateCachedCheckpointToStateRoot(state state.ReadOnlyBeaconState, cp *fork
 	return nil
 }
 
+// ExpandCommitteeCache resizes the cache to a higher limit.
+func ExpandCommitteeCache() {
+	committeeCache.ExpandCommitteeCache()
+}
+
+// CompressCommitteeCache resizes the cache to a lower limit.
+func CompressCommitteeCache() {
+	committeeCache.CompressCommitteeCache()
+}
+
 // ClearCache clears the beacon committee cache and sync committee cache.
 func ClearCache() {
 	committeeCache.Clear()

beacon-chain/core/helpers/rewards_penalties.go

@@ -22,7 +22,7 @@ var balanceCache = cache.NewEffectiveBalanceCache()
 //	 """
 //	 Return the combined effective balance of the ``indices``.
 //	 ``EFFECTIVE_BALANCE_INCREMENT`` Gwei minimum to avoid divisions by zero.
-//	 Math safe up to ~10B ETH, afterwhich this overflows uint64.
+//	 Math safe up to ~10B ETH, after which this overflows uint64.
 //	 """
 //	 return Gwei(max(EFFECTIVE_BALANCE_INCREMENT, sum([state.validators[index].effective_balance for index in indices])))
 func TotalBalance(state state.ReadOnlyValidators, indices []primitives.ValidatorIndex) uint64 {

beacon-chain/core/signing/signing_root.go

@@ -59,7 +59,7 @@ func ComputeDomainAndSign(st state.ReadOnlyBeaconState, epoch primitives.Epoch,
 	return ComputeDomainAndSignWithoutState(st.Fork(), epoch, domain, st.GenesisValidatorsRoot(), obj, key)
 }
 
-// ComputeDomainAndSignWithoutState offers the same functionalit as ComputeDomainAndSign without the need to provide a BeaconState.
+// ComputeDomainAndSignWithoutState offers the same functionality as ComputeDomainAndSign without the need to provide a BeaconState.
 // This is particularly helpful for signing values in tests.
 func ComputeDomainAndSignWithoutState(fork *ethpb.Fork, epoch primitives.Epoch, domain [4]byte, vr []byte, obj fssz.HashRoot, key bls.SecretKey) ([]byte, error) {
 	// EIP-7044: Beginning in Deneb, fix the fork version to Capella for signed exits.

beacon-chain/core/transition/BUILD.bazel

@@ -96,6 +96,7 @@ go_test(
         "//testing/benchmark:go_default_library",
         "//testing/require:go_default_library",
         "//testing/util:go_default_library",
+        "//time/slots:go_default_library",
         "@com_github_google_gofuzz//:go_default_library",
         "@com_github_prysmaticlabs_go_bitfield//:go_default_library",
         "@org_golang_google_protobuf//proto:go_default_library",

beacon-chain/core/transition/altair_transition_no_verify_sig_test.go

@@ -21,6 +21,7 @@ import (
 	"github.com/prysmaticlabs/prysm/v5/testing/assert"
 	"github.com/prysmaticlabs/prysm/v5/testing/require"
 	"github.com/prysmaticlabs/prysm/v5/testing/util"
+	"github.com/prysmaticlabs/prysm/v5/time/slots"
 )
 
 func TestExecuteAltairStateTransitionNoVerify_FullProcess(t *testing.T) {
@@ -48,7 +49,7 @@ func TestExecuteAltairStateTransitionNoVerify_FullProcess(t *testing.T) {
 	epoch := time.CurrentEpoch(beaconState)
 	randaoReveal, err := util.RandaoReveal(beaconState, epoch, privKeys)
 	require.NoError(t, err)
-	require.NoError(t, beaconState.SetSlot(beaconState.Slot()-1))
+	require.NoError(t, beaconState.SetSlot(slots.PrevSlot(beaconState.Slot())))
 
 	nextSlotState, err := transition.ProcessSlots(context.Background(), beaconState.Copy(), beaconState.Slot()+1)
 	require.NoError(t, err)
@@ -135,7 +136,7 @@ func TestExecuteAltairStateTransitionNoVerifySignature_CouldNotVerifyStateRoot(t
 	epoch := time.CurrentEpoch(beaconState)
 	randaoReveal, err := util.RandaoReveal(beaconState, epoch, privKeys)
 	require.NoError(t, err)
-	require.NoError(t, beaconState.SetSlot(beaconState.Slot()-1))
+	require.NoError(t, beaconState.SetSlot(slots.PrevSlot(beaconState.Slot())))
 
 	nextSlotState, err := transition.ProcessSlots(context.Background(), beaconState.Copy(), beaconState.Slot()+1)
 	require.NoError(t, err)

beacon-chain/core/transition/bellatrix_transition_no_verify_sig_test.go

@@ -23,6 +23,7 @@ import (
 	"github.com/prysmaticlabs/prysm/v5/testing/assert"
 	"github.com/prysmaticlabs/prysm/v5/testing/require"
 	"github.com/prysmaticlabs/prysm/v5/testing/util"
+	"github.com/prysmaticlabs/prysm/v5/time/slots"
 )
 
 func TestExecuteBellatrixStateTransitionNoVerify_FullProcess(t *testing.T) {
@@ -50,7 +51,7 @@ func TestExecuteBellatrixStateTransitionNoVerify_FullProcess(t *testing.T) {
 	epoch := time.CurrentEpoch(beaconState)
 	randaoReveal, err := util.RandaoReveal(beaconState, epoch, privKeys)
 	require.NoError(t, err)
-	require.NoError(t, beaconState.SetSlot(beaconState.Slot()-1))
+	require.NoError(t, beaconState.SetSlot(slots.PrevSlot(beaconState.Slot())))
 
 	nextSlotState, err := transition.ProcessSlots(context.Background(), beaconState.Copy(), beaconState.Slot()+1)
 	require.NoError(t, err)
@@ -124,7 +125,7 @@ func TestExecuteBellatrixStateTransitionNoVerifySignature_CouldNotVerifyStateRoo
 		DepositRoot:  bytesutil.PadTo([]byte{2}, 32),
 		BlockHash:    make([]byte, 32),
 	}
-	require.NoError(t, beaconState.SetSlot(params.BeaconConfig().SlotsPerEpoch-1))
+	require.NoError(t, beaconState.SetSlot(slots.PrevSlot(params.BeaconConfig().SlotsPerEpoch)))
 	e := beaconState.Eth1Data()
 	e.DepositCount = 100
 	require.NoError(t, beaconState.SetEth1Data(e))
@@ -137,7 +138,7 @@ func TestExecuteBellatrixStateTransitionNoVerifySignature_CouldNotVerifyStateRoo
 	epoch := time.CurrentEpoch(beaconState)
 	randaoReveal, err := util.RandaoReveal(beaconState, epoch, privKeys)
 	require.NoError(t, err)
-	require.NoError(t, beaconState.SetSlot(beaconState.Slot()-1))
+	require.NoError(t, beaconState.SetSlot(slots.PrevSlot(beaconState.Slot())))
 
 	nextSlotState, err := transition.ProcessSlots(context.Background(), beaconState.Copy(), beaconState.Slot()+1)
 	require.NoError(t, err)

beacon-chain/das/BUILD.bazel

@@ -40,7 +40,6 @@ go_test(
         "//consensus-types/blocks:go_default_library",
         "//consensus-types/primitives:go_default_library",
         "//encoding/bytesutil:go_default_library",
-        "//proto/prysm/v1alpha1:go_default_library",
         "//testing/require:go_default_library",
         "//testing/util:go_default_library",
         "//time/slots:go_default_library",

beacon-chain/das/availability.go

@@ -94,6 +94,8 @@ func (s *LazilyPersistentStore) IsDataAvailable(ctx context.Context, current pri
 	entry := s.cache.ensure(key)
 	defer s.cache.delete(key)
 	root := b.Root()
+	entry.setDiskSummary(s.store.Summary(root))
+
 	// Verify we have all the expected sidecars, and fail fast if any are missing or inconsistent.
 	// We don't try to salvage problematic batches because this indicates a misbehaving peer and we'd rather
 	// ignore their response and decrease their peer score.

beacon-chain/das/availability_test.go

@@ -13,7 +13,6 @@ import (
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/blocks"
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
 	"github.com/prysmaticlabs/prysm/v5/encoding/bytesutil"
-	ethpb "github.com/prysmaticlabs/prysm/v5/proto/prysm/v1alpha1"
 	"github.com/prysmaticlabs/prysm/v5/testing/require"
 	"github.com/prysmaticlabs/prysm/v5/testing/util"
 	"github.com/prysmaticlabs/prysm/v5/time/slots"
@@ -114,16 +113,6 @@ func Test_commitmentsToCheck(t *testing.T) {
 	}
 }
 
-func daAlwaysSucceeds(_ [][]byte, _ []*ethpb.BlobSidecar) error {
-	return nil
-}
-
-type mockDA struct {
-	t   *testing.T
-	scs []blocks.ROBlob
-	err error
-}
-
 func TestLazilyPersistent_Missing(t *testing.T) {
 	ctx := context.Background()
 	store := filesystem.NewEphemeralBlobStorage(t)

beacon-chain/das/cache.go

@@ -4,6 +4,7 @@ import (
 	"bytes"
 
 	"github.com/pkg/errors"
+	"github.com/prysmaticlabs/prysm/v5/beacon-chain/db/filesystem"
 	fieldparams "github.com/prysmaticlabs/prysm/v5/config/fieldparams"
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/blocks"
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
@@ -59,7 +60,12 @@ func (c *cache) delete(key cacheKey) {
 
 // cacheEntry holds a fixed-length cache of BlobSidecars.
 type cacheEntry struct {
-	scs [fieldparams.MaxBlobsPerBlock]*blocks.ROBlob
+	scs         [fieldparams.MaxBlobsPerBlock]*blocks.ROBlob
+	diskSummary filesystem.BlobStorageSummary
+}
+
+func (e *cacheEntry) setDiskSummary(sum filesystem.BlobStorageSummary) {
+	e.diskSummary = sum
 }
 
 // stash adds an item to the in-memory cache of BlobSidecars.
@@ -81,9 +87,17 @@ func (e *cacheEntry) stash(sc *blocks.ROBlob) error {
 // the cache do not match those found in the block. If err is nil, then all expected
 // commitments were found in the cache and the sidecar slice return value can be used
 // to perform a DA check against the cached sidecars.
+// filter only returns blobs that need to be checked. Blobs already available on disk will be excluded.
 func (e *cacheEntry) filter(root [32]byte, kc safeCommitmentArray) ([]blocks.ROBlob, error) {
-	scs := make([]blocks.ROBlob, kc.count())
+	if e.diskSummary.AllAvailable(kc.count()) {
+		return nil, nil
+	}
+	scs := make([]blocks.ROBlob, 0, kc.count())
 	for i := uint64(0); i < fieldparams.MaxBlobsPerBlock; i++ {
+		// We already have this blob, we don't need to write it or validate it.
+		if e.diskSummary.HasIndex(i) {
+			continue
+		}
 		if kc[i] == nil {
 			if e.scs[i] != nil {
 				return nil, errors.Wrapf(errCommitmentMismatch, "root=%#x, index=%#x, commitment=%#x, no block commitment", root, i, e.scs[i].KzgCommitment)
@@ -97,7 +111,7 @@ func (e *cacheEntry) filter(root [32]byte, kc safeCommitmentArray) ([]blocks.ROB
 		if !bytes.Equal(kc[i], e.scs[i].KzgCommitment) {
 			return nil, errors.Wrapf(errCommitmentMismatch, "root=%#x, index=%#x, commitment=%#x, block commitment=%#x", root, i, e.scs[i].KzgCommitment, kc[i])
 		}
-		scs[i] = *e.scs[i]
+		scs = append(scs, *e.scs[i])
 	}
 
 	return scs, nil

beacon-chain/das/cache_test.go

@@ -3,9 +3,14 @@ package das
 import (
 	"testing"
 
+	"github.com/prysmaticlabs/prysm/v5/beacon-chain/db/filesystem"
+	"github.com/prysmaticlabs/prysm/v5/config/params"
+	"github.com/prysmaticlabs/prysm/v5/consensus-types/blocks"
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
 	"github.com/prysmaticlabs/prysm/v5/encoding/bytesutil"
 	"github.com/prysmaticlabs/prysm/v5/testing/require"
+	"github.com/prysmaticlabs/prysm/v5/testing/util"
+	"github.com/prysmaticlabs/prysm/v5/time/slots"
 )
 
 func TestCacheEnsureDelete(t *testing.T) {
@@ -23,3 +28,145 @@ func TestCacheEnsureDelete(t *testing.T) {
 	var nilEntry *cacheEntry
 	require.Equal(t, nilEntry, c.entries[k])
 }
+
+type filterTestCaseSetupFunc func(t *testing.T) (*cacheEntry, safeCommitmentArray, []blocks.ROBlob)
+
+func filterTestCaseSetup(slot primitives.Slot, nBlobs int, onDisk []int, numExpected int) filterTestCaseSetupFunc {
+	return func(t *testing.T) (*cacheEntry, safeCommitmentArray, []blocks.ROBlob) {
+		blk, blobs := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, slot, nBlobs)
+		commits, err := commitmentsToCheck(blk, blk.Block().Slot())
+		require.NoError(t, err)
+		entry := &cacheEntry{}
+		if len(onDisk) > 0 {
+			od := map[[32]byte][]int{blk.Root(): onDisk}
+			sumz := filesystem.NewMockBlobStorageSummarizer(t, od)
+			sum := sumz.Summary(blk.Root())
+			entry.setDiskSummary(sum)
+		}
+		expected := make([]blocks.ROBlob, 0, nBlobs)
+		for i := 0; i < commits.count(); i++ {
+			if entry.diskSummary.HasIndex(uint64(i)) {
+				continue
+			}
+			// If we aren't telling the cache a blob is on disk, add it to the expected list and stash.
+			expected = append(expected, blobs[i])
+			require.NoError(t, entry.stash(&blobs[i]))
+		}
+		require.Equal(t, numExpected, len(expected))
+		return entry, commits, expected
+	}
+}
+
+func TestFilterDiskSummary(t *testing.T) {
+	denebSlot, err := slots.EpochStart(params.BeaconConfig().DenebForkEpoch)
+	require.NoError(t, err)
+	cases := []struct {
+		name  string
+		setup filterTestCaseSetupFunc
+	}{
+		{
+			name:  "full blobs, all on disk",
+			setup: filterTestCaseSetup(denebSlot, 6, []int{0, 1, 2, 3, 4, 5}, 0),
+		},
+		{
+			name:  "full blobs, first on disk",
+			setup: filterTestCaseSetup(denebSlot, 6, []int{0}, 5),
+		},
+		{
+			name:  "full blobs, middle on disk",
+			setup: filterTestCaseSetup(denebSlot, 6, []int{2}, 5),
+		},
+		{
+			name:  "full blobs, last on disk",
+			setup: filterTestCaseSetup(denebSlot, 6, []int{5}, 5),
+		},
+		{
+			name:  "full blobs, none on disk",
+			setup: filterTestCaseSetup(denebSlot, 6, []int{}, 6),
+		},
+		{
+			name:  "one commitment, on disk",
+			setup: filterTestCaseSetup(denebSlot, 1, []int{0}, 0),
+		},
+		{
+			name:  "one commitment, not on disk",
+			setup: filterTestCaseSetup(denebSlot, 1, []int{}, 1),
+		},
+		{
+			name:  "two commitments, first on disk",
+			setup: filterTestCaseSetup(denebSlot, 2, []int{0}, 1),
+		},
+		{
+			name:  "two commitments, last on disk",
+			setup: filterTestCaseSetup(denebSlot, 2, []int{1}, 1),
+		},
+		{
+			name:  "two commitments, none on disk",
+			setup: filterTestCaseSetup(denebSlot, 2, []int{}, 2),
+		},
+		{
+			name:  "two commitments, all on disk",
+			setup: filterTestCaseSetup(denebSlot, 2, []int{0, 1}, 0),
+		},
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			entry, commits, expected := c.setup(t)
+			// first (root) argument doesn't matter, it is just for logs
+			got, err := entry.filter([32]byte{}, commits)
+			require.NoError(t, err)
+			require.Equal(t, len(expected), len(got))
+		})
+	}
+}
+
+func TestFilter(t *testing.T) {
+	denebSlot, err := slots.EpochStart(params.BeaconConfig().DenebForkEpoch)
+	require.NoError(t, err)
+	cases := []struct {
+		name  string
+		setup func(t *testing.T) (*cacheEntry, safeCommitmentArray, []blocks.ROBlob)
+		err   error
+	}{
+		{
+			name: "commitments mismatch - extra sidecar",
+			setup: func(t *testing.T) (*cacheEntry, safeCommitmentArray, []blocks.ROBlob) {
+				entry, commits, expected := filterTestCaseSetup(denebSlot, 6, []int{0, 1}, 4)(t)
+				commits[5] = nil
+				return entry, commits, expected
+			},
+			err: errCommitmentMismatch,
+		},
+		{
+			name: "sidecar missing",
+			setup: func(t *testing.T) (*cacheEntry, safeCommitmentArray, []blocks.ROBlob) {
+				entry, commits, expected := filterTestCaseSetup(denebSlot, 6, []int{0, 1}, 4)(t)
+				entry.scs[5] = nil
+				return entry, commits, expected
+			},
+			err: errMissingSidecar,
+		},
+		{
+			name: "commitments mismatch - different bytes",
+			setup: func(t *testing.T) (*cacheEntry, safeCommitmentArray, []blocks.ROBlob) {
+				entry, commits, expected := filterTestCaseSetup(denebSlot, 6, []int{0, 1}, 4)(t)
+				entry.scs[5].KzgCommitment = []byte("nope")
+				return entry, commits, expected
+			},
+			err: errCommitmentMismatch,
+		},
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			entry, commits, expected := c.setup(t)
+			// first (root) argument doesn't matter, it is just for logs
+			got, err := entry.filter([32]byte{}, commits)
+			if c.err != nil {
+				require.ErrorIs(t, err, c.err)
+				return
+			}
+			require.NoError(t, err)
+			require.Equal(t, len(expected), len(got))
+		})
+	}
+}

beacon-chain/db/errors.go

@@ -19,9 +19,6 @@ var ErrNotFoundState = kv.ErrNotFoundState
 // ErrNotFoundOriginBlockRoot wraps ErrNotFound for an error specific to the origin block root.
 var ErrNotFoundOriginBlockRoot = kv.ErrNotFoundOriginBlockRoot
 
-// ErrNotFoundBackfillBlockRoot wraps ErrNotFound for an error specific to the backfill block root.
-var ErrNotFoundBackfillBlockRoot = kv.ErrNotFoundBackfillBlockRoot
-
 // IsNotFound allows callers to treat errors from a flat-file database, where the file record is missing,
 // as equivalent to db.ErrNotFound.
 func IsNotFound(err error) bool {

beacon-chain/db/filesystem/BUILD.bazel

@@ -4,22 +4,28 @@ go_library(
     name = "go_default_library",
     srcs = [
         "blob.go",
-        "ephemeral.go",
+        "cache.go",
+        "iteration.go",
+        "layout.go",
+        "log.go",
         "metrics.go",
+        "mock.go",
         "pruner.go",
     ],
     importpath = "github.com/prysmaticlabs/prysm/v5/beacon-chain/db/filesystem",
     visibility = ["//visibility:public"],
     deps = [
+        "//beacon-chain/db:go_default_library",
         "//beacon-chain/verification:go_default_library",
         "//config/fieldparams:go_default_library",
         "//config/params:go_default_library",
         "//consensus-types/blocks:go_default_library",
         "//consensus-types/primitives:go_default_library",
+        "//encoding/bytesutil:go_default_library",
         "//io/file:go_default_library",
-        "//proto/prysm/v1alpha1:go_default_library",
         "//runtime/logging:go_default_library",
         "//time/slots:go_default_library",
+        "@com_github_ethereum_go_ethereum//common/hexutil:go_default_library",
         "@com_github_pkg_errors//:go_default_library",
         "@com_github_prometheus_client_golang//prometheus:go_default_library",
         "@com_github_prometheus_client_golang//prometheus/promauto:go_default_library",
@@ -32,18 +38,25 @@ go_test(
     name = "go_default_test",
     srcs = [
         "blob_test.go",
+        "cache_test.go",
+        "iteration_test.go",
+        "layout_test.go",
+        "migration_test.go",
         "pruner_test.go",
     ],
     embed = [":go_default_library"],
     deps = [
+        "//beacon-chain/db:go_default_library",
         "//beacon-chain/verification:go_default_library",
         "//config/fieldparams:go_default_library",
+        "//config/params:go_default_library",
+        "//consensus-types/blocks:go_default_library",
         "//consensus-types/primitives:go_default_library",
         "//encoding/bytesutil:go_default_library",
         "//proto/prysm/v1alpha1:go_default_library",
         "//testing/require:go_default_library",
         "//testing/util:go_default_library",
-        "@com_github_pkg_errors//:go_default_library",
+        "//time/slots:go_default_library",
         "@com_github_prysmaticlabs_fastssz//:go_default_library",
         "@com_github_spf13_afero//:go_default_library",
     ],

beacon-chain/db/filesystem/blob.go

@@ -2,10 +2,8 @@ package filesystem
 
 import (
 	"fmt"
-	"os"
+	"math"
 	"path"
-	"strconv"
-	"strings"
 	"time"
 
 	"github.com/pkg/errors"
@@ -14,34 +12,50 @@ import (
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/blocks"
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
 	"github.com/prysmaticlabs/prysm/v5/io/file"
-	ethpb "github.com/prysmaticlabs/prysm/v5/proto/prysm/v1alpha1"
 	"github.com/prysmaticlabs/prysm/v5/runtime/logging"
-	log "github.com/sirupsen/logrus"
+	"github.com/sirupsen/logrus"
 	"github.com/spf13/afero"
 )
 
 var (
-	errIndexOutOfBounds = errors.New("blob index in file name >= MaxBlobsPerBlock")
+	errIndexOutOfBounds    = errors.New("blob index in file name >= MaxBlobsPerBlock")
+	errEmptyBlobWritten    = errors.New("zero bytes written to disk when saving blob sidecar")
+	errSidecarEmptySSZData = errors.New("sidecar marshalled to an empty ssz byte slice")
+	errNoBasePath          = errors.New("BlobStorage base path not specified in init")
 )
 
-const (
-	sszExt  = "ssz"
-	partExt = "part"
-
-	directoryPermissions = 0700
-)
+const directoryPermissions = 0700
 
 // BlobStorageOption is a functional option for configuring a BlobStorage.
 type BlobStorageOption func(*BlobStorage) error
 
+// WithBasePath is a required option that sets the base path of blob storage.
+func WithBasePath(base string) BlobStorageOption {
+	return func(b *BlobStorage) error {
+		b.base = base
+		return nil
+	}
+}
+
 // WithBlobRetentionEpochs is an option that changes the number of epochs blobs will be persisted.
 func WithBlobRetentionEpochs(e primitives.Epoch) BlobStorageOption {
 	return func(b *BlobStorage) error {
-		pruner, err := newBlobPruner(b.fs, e)
-		if err != nil {
-			return err
-		}
-		b.pruner = pruner
+		b.retentionEpochs = e
+		return nil
+	}
+}
+
+// WithSaveFsync is an option that causes Save to call fsync before renaming part files for improved durability.
+func WithSaveFsync(fsync bool) BlobStorageOption {
+	return func(b *BlobStorage) error {
+		b.fsync = fsync
+		return nil
+	}
+}
+
+func WithFs(fs afero.Fs) BlobStorageOption {
+	return func(b *BlobStorage) error {
+		b.fs = fs
 		return nil
 	}
 }
@@ -49,73 +63,89 @@ func WithBlobRetentionEpochs(e primitives.Epoch) BlobStorageOption {
 // NewBlobStorage creates a new instance of the BlobStorage object. Note that the implementation of BlobStorage may
 // attempt to hold a file lock to guarantee exclusive control of the blob storage directory, so this should only be
 // initialized once per beacon node.
-func NewBlobStorage(base string, opts ...BlobStorageOption) (*BlobStorage, error) {
-	base = path.Clean(base)
-	if err := file.MkdirAll(base); err != nil {
-		return nil, fmt.Errorf("failed to create blob storage at %s: %w", base, err)
-	}
-	fs := afero.NewBasePathFs(afero.NewOsFs(), base)
-	b := &BlobStorage{
-		fs: fs,
-	}
+func NewBlobStorage(opts ...BlobStorageOption) (*BlobStorage, error) {
+	b := &BlobStorage{}
 	for _, o := range opts {
 		if err := o(b); err != nil {
-			return nil, fmt.Errorf("failed to create blob storage at %s: %w", base, err)
+			return nil, errors.Wrap(err, "failed to create blob storage")
 		}
 	}
-	if b.pruner == nil {
-		log.Warn("Initializing blob filesystem storage with pruning disabled")
+	// Allow tests to set up a different fs using WithFs.
+	if b.fs == nil {
+		if b.base == "" {
+			return nil, errNoBasePath
+		}
+		b.base = path.Clean(b.base)
+		if err := file.MkdirAll(b.base); err != nil {
+			return nil, errors.Wrapf(err, "failed to create blob storage at %s", b.base)
+		}
+		b.fs = afero.NewBasePathFs(afero.NewOsFs(), b.base)
 	}
+	b.cache = newBlobStorageCache()
+	pruner := newBlobPruner(b.retentionEpochs)
+	layout, err := newPeriodicEpochLayout(b.fs, b.cache, pruner)
+	if err != nil {
+		return nil, err
+	}
+	b.layout = layout
 	return b, nil
 }
 
 // BlobStorage is the concrete implementation of the filesystem backend for saving and retrieving BlobSidecars.
 type BlobStorage struct {
-	fs     afero.Fs
-	pruner *blobPruner
+	base            string
+	retentionEpochs primitives.Epoch
+	fsync           bool
+	fs              afero.Fs
+	pruner          *blobPruner
+	layout          runtimeLayout
+	cache           *blobStorageCache
 }
 
 // WarmCache runs the prune routine with an expiration of slot of 0, so nothing will be pruned, but the pruner's cache
 // will be populated at node startup, avoiding a costly cold prune (~4s in syscalls) during syncing.
 func (bs *BlobStorage) WarmCache() {
-	if bs.pruner == nil {
-		return
+	start := time.Now()
+	if err := warmCache(bs.layout, bs.cache); err != nil {
+		log.WithError(err).Error("Error encountered while warming up blob filesystem cache.")
+	}
+	log.WithField("elapsed", time.Since(start)).Info("Blob filesystem cache warm-up complete.")
+	from := &flatRootLayout{fs: bs.fs}
+	if err := migrateLayout(bs.fs, from, bs.layout, bs.cache); err != nil {
+		log.WithError(err).Error("Error encountered while migrating legacy blob storage scheme.")
 	}
-	go func() {
-		if err := bs.pruner.prune(0); err != nil {
-			log.WithError(err).Error("Error encountered while warming up blob pruner cache.")
-		}
-	}()
 }
 
 // Save saves blobs given a list of sidecars.
 func (bs *BlobStorage) Save(sidecar blocks.VerifiedROBlob) error {
 	startTime := time.Now()
-	fname := namerForSidecar(sidecar)
-	sszPath := fname.path()
+	ident := identForSidecar(sidecar)
+	sszPath := bs.layout.sszPath(ident)
 	exists, err := afero.Exists(bs.fs, sszPath)
 	if err != nil {
 		return err
 	}
 	if exists {
-		log.WithFields(logging.BlobFields(sidecar.ROBlob)).Debug("ignoring a duplicate blob sidecar Save attempt")
+		log.WithFields(logging.BlobFields(sidecar.ROBlob)).Debug("Ignoring a duplicate blob sidecar save attempt")
 		return nil
 	}
-	if bs.pruner != nil {
-		if err := bs.pruner.notify(sidecar.BlockRoot(), sidecar.Slot(), sidecar.Index); err != nil {
-			return errors.Wrapf(err, "problem maintaining pruning cache/metrics for sidecar with root=%#x", sidecar.BlockRoot())
-		}
+
+	if err := bs.layout.notify(ident); err != nil {
+		return errors.Wrapf(err, "problem maintaining pruning cache/metrics for sidecar with root=%#x", sidecar.BlockRoot())
 	}
 
 	// Serialize the ethpb.BlobSidecar to binary data using SSZ.
 	sidecarData, err := sidecar.MarshalSSZ()
 	if err != nil {
 		return errors.Wrap(err, "failed to serialize sidecar data")
+	} else if len(sidecarData) == 0 {
+		return errSidecarEmptySSZData
 	}
-	if err := bs.fs.MkdirAll(fname.dir(), directoryPermissions); err != nil {
+
+	if err := bs.fs.MkdirAll(bs.layout.dir(ident), directoryPermissions); err != nil {
 		return err
 	}
-	partPath := fname.partPath()
+	partPath := bs.layout.partPath(ident, fmt.Sprintf("%p", sidecarData))
 
 	partialMoved := false
 	// Ensure the partial file is deleted.
@@ -126,9 +156,9 @@ func (bs *BlobStorage) Save(sidecar blocks.VerifiedROBlob) error {
 		// It's expected to error if the save is successful.
 		err = bs.fs.Remove(partPath)
 		if err == nil {
-			log.WithFields(log.Fields{
+			log.WithFields(logrus.Fields{
 				"partPath": partPath,
-			}).Debugf("removed partial file")
+			}).Debugf("Removed partial file")
 		}
 	}()
 
@@ -138,7 +168,7 @@ func (bs *BlobStorage) Save(sidecar blocks.VerifiedROBlob) error {
 		return errors.Wrap(err, "failed to create partial file")
 	}
 
-	_, err = partialFile.Write(sidecarData)
+	n, err := partialFile.Write(sidecarData)
 	if err != nil {
 		closeErr := partialFile.Close()
 		if closeErr != nil {
@@ -146,11 +176,24 @@ func (bs *BlobStorage) Save(sidecar blocks.VerifiedROBlob) error {
 		}
 		return errors.Wrap(err, "failed to write to partial file")
 	}
-	err = partialFile.Close()
-	if err != nil {
+	if bs.fsync {
+		if err := partialFile.Sync(); err != nil {
+			return err
+		}
+	}
+
+	if err := partialFile.Close(); err != nil {
 		return err
 	}
 
+	if n != len(sidecarData) {
+		return fmt.Errorf("failed to write the full bytes of sidecarData, wrote only %d of %d bytes", n, len(sidecarData))
+	}
+
+	if n == 0 {
+		return errEmptyBlobWritten
+	}
+
 	// Atomically rename the partial file to its final name.
 	err = bs.fs.Rename(partPath, sszPath)
 	if err != nil {
@@ -167,67 +210,37 @@ func (bs *BlobStorage) Save(sidecar blocks.VerifiedROBlob) error {
 // value is always a VerifiedROBlob.
 func (bs *BlobStorage) Get(root [32]byte, idx uint64) (blocks.VerifiedROBlob, error) {
 	startTime := time.Now()
-	expected := blobNamer{root: root, index: idx}
-	encoded, err := afero.ReadFile(bs.fs, expected.path())
-	var v blocks.VerifiedROBlob
+	ident, err := bs.layout.ident(root, idx)
 	if err != nil {
-		return v, err
-	}
-	s := &ethpb.BlobSidecar{}
-	if err := s.UnmarshalSSZ(encoded); err != nil {
-		return v, err
-	}
-	ro, err := blocks.NewROBlobWithRoot(s, root)
-	if err != nil {
-		return blocks.VerifiedROBlob{}, err
+		return verification.VerifiedROBlobError(err)
 	}
 	defer func() {
 		blobFetchLatency.Observe(float64(time.Since(startTime).Milliseconds()))
 	}()
-	return verification.BlobSidecarNoop(ro)
+	return verification.VerifiedROBlobFromDisk(bs.fs, root, bs.layout.sszPath(ident))
 }
 
 // Remove removes all blobs for a given root.
 func (bs *BlobStorage) Remove(root [32]byte) error {
-	rootDir := blobNamer{root: root}.dir()
-	return bs.fs.RemoveAll(rootDir)
+	dirIdent, err := bs.layout.dirIdent(root)
+	if err != nil {
+		return err
+	}
+	_, err = bs.layout.remove(dirIdent)
+	return err
 }
 
 // Indices generates a bitmap representing which BlobSidecar.Index values are present on disk for a given root.
 // This value can be compared to the commitments observed in a block to determine which indices need to be found
 // on the network to confirm data availability.
 func (bs *BlobStorage) Indices(root [32]byte) ([fieldparams.MaxBlobsPerBlock]bool, error) {
-	var mask [fieldparams.MaxBlobsPerBlock]bool
-	rootDir := blobNamer{root: root}.dir()
-	entries, err := afero.ReadDir(bs.fs, rootDir)
-	if err != nil {
-		if os.IsNotExist(err) {
-			return mask, nil
-		}
-		return mask, err
-	}
-	for i := range entries {
-		if entries[i].IsDir() {
-			continue
-		}
-		name := entries[i].Name()
-		if !strings.HasSuffix(name, sszExt) {
-			continue
-		}
-		parts := strings.Split(name, ".")
-		if len(parts) != 2 {
-			continue
-		}
-		u, err := strconv.ParseUint(parts[0], 10, 64)
-		if err != nil {
-			return mask, errors.Wrapf(err, "unexpected directory entry breaks listing, %s", parts[0])
-		}
-		if u >= fieldparams.MaxBlobsPerBlock {
-			return mask, errIndexOutOfBounds
-		}
-		mask[u] = true
-	}
-	return mask, nil
+	return bs.Summary(root).mask, nil
+}
+
+// Summary returns the BlobStorageSummary from the layout.
+// Internally, this is a cached representation of the directory listing for the given root.
+func (bs *BlobStorage) Summary(root [32]byte) BlobStorageSummary {
+	return bs.layout.summary(root)
 }
 
 // Clear deletes all files on the filesystem.
@@ -244,31 +257,11 @@ func (bs *BlobStorage) Clear() error {
 	return nil
 }
 
-type blobNamer struct {
-	root  [32]byte
-	index uint64
-}
-
-func namerForSidecar(sc blocks.VerifiedROBlob) blobNamer {
-	return blobNamer{root: sc.BlockRoot(), index: sc.Index}
-}
-
-func (p blobNamer) dir() string {
-	return rootString(p.root)
-}
-
-func (p blobNamer) fname(ext string) string {
-	return path.Join(p.dir(), fmt.Sprintf("%d.%s", p.index, ext))
-}
-
-func (p blobNamer) partPath() string {
-	return p.fname(partExt)
-}
-
-func (p blobNamer) path() string {
-	return p.fname(sszExt)
-}
-
-func rootString(root [32]byte) string {
-	return fmt.Sprintf("%#x", root)
+// WithinRetentionPeriod checks if the requested epoch is within the blob retention period.
+func (bs *BlobStorage) WithinRetentionPeriod(requested, current primitives.Epoch) bool {
+	if requested > math.MaxUint64-bs.retentionEpochs {
+		// If there is an overflow, then the retention period was set to an extremely large number.
+		return true
+	}
+	return requested+bs.retentionEpochs >= current
 }

beacon-chain/db/filesystem/blob_test.go

@@ -2,34 +2,33 @@ package filesystem
 
 import (
 	"bytes"
+	"math"
 	"os"
 	"path"
+	"sync"
 	"testing"
-	"time"
 
-	"github.com/pkg/errors"
 	ssz "github.com/prysmaticlabs/fastssz"
+	"github.com/prysmaticlabs/prysm/v5/beacon-chain/db"
 	"github.com/prysmaticlabs/prysm/v5/beacon-chain/verification"
 	fieldparams "github.com/prysmaticlabs/prysm/v5/config/fieldparams"
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
-	"github.com/prysmaticlabs/prysm/v5/encoding/bytesutil"
 	ethpb "github.com/prysmaticlabs/prysm/v5/proto/prysm/v1alpha1"
 	"github.com/prysmaticlabs/prysm/v5/testing/require"
 	"github.com/prysmaticlabs/prysm/v5/testing/util"
+	"github.com/prysmaticlabs/prysm/v5/time/slots"
 	"github.com/spf13/afero"
 )
 
 func TestBlobStorage_SaveBlobData(t *testing.T) {
 	_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, 1, fieldparams.MaxBlobsPerBlock)
-	testSidecars, err := verification.BlobSidecarSliceNoop(sidecars)
-	require.NoError(t, err)
+	testSidecars := verification.FakeVerifySliceForTest(t, sidecars)
 
 	t.Run("no error for duplicate", func(t *testing.T) {
-		fs, bs, err := NewEphemeralBlobStorageWithFs(t)
-		require.NoError(t, err)
+		fs, bs := NewEphemeralBlobStorageAndFs(t)
 		existingSidecar := testSidecars[0]
 
-		blobPath := namerForSidecar(existingSidecar).path()
+		blobPath := bs.layout.sszPath(identForSidecar(existingSidecar))
 		// Serialize the existing BlobSidecar to binary data.
 		existingSidecarData, err := ssz.MarshalSSZ(existingSidecar)
 		require.NoError(t, err)
@@ -86,7 +85,7 @@ func TestBlobStorage_SaveBlobData(t *testing.T) {
 
 		require.NoError(t, bs.Remove(expected.BlockRoot()))
 		_, err = bs.Get(expected.BlockRoot(), expected.Index)
-		require.ErrorContains(t, "file does not exist", err)
+		require.Equal(t, true, db.IsNotFound(err))
 	})
 
 	t.Run("clear", func(t *testing.T) {
@@ -101,40 +100,39 @@ func TestBlobStorage_SaveBlobData(t *testing.T) {
 		_, err = b.Get(blob.BlockRoot(), blob.Index)
 		require.ErrorIs(t, err, os.ErrNotExist)
 	})
-}
 
-// pollUntil polls a condition function until it returns true or a timeout is reached.
-func pollUntil(t *testing.T, fs afero.Fs, expected int) error {
-	var remainingFolders []os.FileInfo
-	var err error
-	// Define the condition function for polling
-	conditionFunc := func() bool {
-		remainingFolders, err = afero.ReadDir(fs, ".")
+	t.Run("race conditions", func(t *testing.T) {
+		// There was a bug where saving the same blob in multiple go routines would cause a partial blob
+		// to be empty. This test ensures that several routines can safely save the same blob at the
+		// same time. This isn't ideal behavior from the caller, but should be handled safely anyway.
+		// See https://github.com/prysmaticlabs/prysm/pull/13648
+		b, err := NewBlobStorage(WithBasePath(t.TempDir()))
 		require.NoError(t, err)
-		return len(remainingFolders) == expected
-	}
+		blob := testSidecars[0]
 
-	startTime := time.Now()
-	for {
-		if conditionFunc() {
-			break // Condition met, exit the loop
+		var wg sync.WaitGroup
+		for i := 0; i < 100; i++ {
+			wg.Add(1)
+			go func() {
+				defer wg.Done()
+				require.NoError(t, b.Save(blob))
+			}()
 		}
-		if time.Since(startTime) > 30*time.Second {
-			return errors.New("timeout")
-		}
-		time.Sleep(1 * time.Second) // Adjust the sleep interval as needed
-	}
-	require.Equal(t, expected, len(remainingFolders))
-	return nil
+
+		wg.Wait()
+		res, err := b.Get(blob.BlockRoot(), blob.Index)
+		require.NoError(t, err)
+		require.DeepSSZEqual(t, blob, res)
+	})
 }
 
 func TestBlobIndicesBounds(t *testing.T) {
-	fs, bs, err := NewEphemeralBlobStorageWithFs(t)
-	require.NoError(t, err)
+	fs := afero.NewMemMapFs()
 	root := [32]byte{}
 
 	okIdx := uint64(fieldparams.MaxBlobsPerBlock - 1)
-	writeFakeSSZ(t, fs, root, okIdx)
+	writeFakeSSZ(t, fs, root, 0, okIdx)
+	bs := NewEphemeralBlobStorageUsingFs(t, fs)
 	indices, err := bs.Indices(root)
 	require.NoError(t, err)
 	var expected [fieldparams.MaxBlobsPerBlock]bool
@@ -144,25 +142,27 @@ func TestBlobIndicesBounds(t *testing.T) {
 	}
 
 	oobIdx := uint64(fieldparams.MaxBlobsPerBlock)
-	writeFakeSSZ(t, fs, root, oobIdx)
-	_, err = bs.Indices(root)
-	require.ErrorIs(t, err, errIndexOutOfBounds)
+	writeFakeSSZ(t, fs, root, 0, oobIdx)
+	// This now fails at cache warmup time.
+	require.ErrorIs(t, err, warmCache(bs.layout, bs.cache))
 }
 
-func writeFakeSSZ(t *testing.T, fs afero.Fs, root [32]byte, idx uint64) {
-	namer := blobNamer{root: root, index: idx}
-	require.NoError(t, fs.MkdirAll(namer.dir(), 0700))
-	fh, err := fs.Create(namer.path())
+func writeFakeSSZ(t *testing.T, fs afero.Fs, root [32]byte, slot primitives.Slot, idx uint64) {
+	epoch := slots.ToEpoch(slot)
+	namer := newBlobIdent(root, epoch, idx)
+	layout := periodicEpochLayout{}
+	require.NoError(t, fs.MkdirAll(layout.dir(namer), 0700))
+	fh, err := fs.Create(layout.sszPath(namer))
 	require.NoError(t, err)
 	_, err = fh.Write([]byte("derp"))
 	require.NoError(t, err)
 	require.NoError(t, fh.Close())
 }
 
+/*
 func TestBlobStoragePrune(t *testing.T) {
 	currentSlot := primitives.Slot(200000)
-	fs, bs, err := NewEphemeralBlobStorageWithFs(t)
-	require.NoError(t, err)
+	fs, bs := NewEphemeralBlobStorageAndFs(t)
 
 	t.Run("PruneOne", func(t *testing.T) {
 		_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, 300, fieldparams.MaxBlobsPerBlock)
@@ -172,10 +172,15 @@ func TestBlobStoragePrune(t *testing.T) {
 		for _, sidecar := range testSidecars {
 			require.NoError(t, bs.Save(sidecar))
 		}
+		ident := identForSidecar(testSidecars[0])
+
+		beforeFolders, err := afero.ReadDir(fs, ident.groupDir())
+		require.NoError(t, err)
+		require.Equal(t, 1, len(beforeFolders))
 
 		require.NoError(t, bs.pruner.prune(currentSlot-bs.pruner.windowSize))
 
-		remainingFolders, err := afero.ReadDir(fs, ".")
+		remainingFolders, err := afero.ReadDir(fs, ident.groupDir())
 		require.NoError(t, err)
 		require.Equal(t, 0, len(remainingFolders))
 	})
@@ -183,6 +188,7 @@ func TestBlobStoragePrune(t *testing.T) {
 		_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, 299, fieldparams.MaxBlobsPerBlock)
 		testSidecars, err := verification.BlobSidecarSliceNoop(sidecars)
 		require.NoError(t, err)
+		ident := identForSidecar(testSidecars[0])
 
 		for _, sidecar := range testSidecars[4:] {
 			require.NoError(t, bs.Save(sidecar))
@@ -190,59 +196,97 @@ func TestBlobStoragePrune(t *testing.T) {
 
 		require.NoError(t, bs.pruner.prune(currentSlot-bs.pruner.windowSize))
 
-		remainingFolders, err := afero.ReadDir(fs, ".")
+		remainingFolders, err := afero.ReadDir(fs, ident.groupDir())
 		require.NoError(t, err)
 		require.Equal(t, 0, len(remainingFolders))
 	})
 	t.Run("PruneMany", func(t *testing.T) {
-		blockQty := 10
-		slot := primitives.Slot(1)
-
-		for j := 0; j <= blockQty; j++ {
-			root := bytesutil.ToBytes32(bytesutil.ToBytes(uint64(slot), 32))
-			_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, root, slot, fieldparams.MaxBlobsPerBlock)
+		pruneBefore := currentSlot - bs.pruner.windowSize
+		increment := primitives.Slot(10000)
+		slots := []primitives.Slot{
+			pruneBefore - increment,
+			pruneBefore - (2 * increment),
+			pruneBefore,
+			pruneBefore + increment,
+			pruneBefore + (2 * increment),
+		}
+		namers := make([]blobIdent, len(slots))
+		for i, s := range slots {
+			_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, s, 1)
 			testSidecars, err := verification.BlobSidecarSliceNoop(sidecars)
 			require.NoError(t, err)
 			require.NoError(t, bs.Save(testSidecars[0]))
-
-			slot += 10000
+			namers[i] = identForSidecar(testSidecars[0])
 		}
 
 		require.NoError(t, bs.pruner.prune(currentSlot-bs.pruner.windowSize))
 
-		remainingFolders, err := afero.ReadDir(fs, ".")
-		require.NoError(t, err)
-		require.Equal(t, 4, len(remainingFolders))
+		// first 2 subdirs should be removed
+		for _, nmr := range namers[0:2] {
+			entries, err := listDir(fs, nmr.dir())
+			require.Equal(t, 0, len(entries))
+			require.ErrorIs(t, err, os.ErrNotExist)
+		}
+		// the rest should still be there
+		for _, nmr := range namers[2:] {
+			entries, err := listDir(fs, nmr.dir())
+			require.NoError(t, err)
+			require.Equal(t, 1, len(entries))
+		}
 	})
 }
-
-func BenchmarkPruning(b *testing.B) {
-	var t *testing.T
-	_, bs, err := NewEphemeralBlobStorageWithFs(t)
-	require.NoError(t, err)
-
-	blockQty := 10000
-	currentSlot := primitives.Slot(150000)
-	slot := primitives.Slot(0)
-
-	for j := 0; j <= blockQty; j++ {
-		root := bytesutil.ToBytes32(bytesutil.ToBytes(uint64(slot), 32))
-		_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, root, slot, fieldparams.MaxBlobsPerBlock)
-		testSidecars, err := verification.BlobSidecarSliceNoop(sidecars)
-		require.NoError(t, err)
-		require.NoError(t, bs.Save(testSidecars[0]))
-
-		slot += 100
-	}
-	b.ReportAllocs()
-	b.ResetTimer()
-	for i := 0; i < b.N; i++ {
-		err := bs.pruner.prune(currentSlot)
-		require.NoError(b, err)
-	}
-}
+*/
 
 func TestNewBlobStorage(t *testing.T) {
-	_, err := NewBlobStorage(path.Join(t.TempDir(), "good"))
+	_, err := NewBlobStorage()
+	require.ErrorIs(t, err, errNoBasePath)
+	_, err = NewBlobStorage(WithBasePath(path.Join(t.TempDir(), "good")))
 	require.NoError(t, err)
 }
+
+func TestConfig_WithinRetentionPeriod(t *testing.T) {
+	retention := primitives.Epoch(16)
+	storage := &BlobStorage{retentionEpochs: retention}
+
+	cases := []struct {
+		name      string
+		requested primitives.Epoch
+		current   primitives.Epoch
+		within    bool
+	}{
+		{
+			name:      "before",
+			requested: 0,
+			current:   retention + 1,
+			within:    false,
+		},
+		{
+			name:      "same",
+			requested: 0,
+			current:   0,
+			within:    true,
+		},
+		{
+			name:      "boundary",
+			requested: 0,
+			current:   retention,
+			within:    true,
+		},
+		{
+			name:      "one less",
+			requested: retention - 1,
+			current:   retention,
+			within:    true,
+		},
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			require.Equal(t, c.within, storage.WithinRetentionPeriod(c.requested, c.current))
+		})
+	}
+
+	t.Run("overflow", func(t *testing.T) {
+		storage := &BlobStorage{retentionEpochs: math.MaxUint64}
+		require.Equal(t, true, storage.WithinRetentionPeriod(1, 1))
+	})
+}

beacon-chain/db/filesystem/cache.go

@@ -0,0 +1,151 @@
+package filesystem
+
+import (
+	"sync"
+
+	"github.com/prysmaticlabs/prysm/v5/beacon-chain/db"
+	fieldparams "github.com/prysmaticlabs/prysm/v5/config/fieldparams"
+	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
+)
+
+const bytesPerSidecar = 131928
+
+// blobIndexMask is a bitmask representing the set of blob indices that are currently set.
+type blobIndexMask [fieldparams.MaxBlobsPerBlock]bool
+
+// BlobStorageSummary represents cached information about the BlobSidecars on disk for each root the cache knows about.
+type BlobStorageSummary struct {
+	epoch primitives.Epoch
+	mask  blobIndexMask
+}
+
+// HasIndex returns true if the BlobSidecar at the given index is available in the filesystem.
+func (s BlobStorageSummary) HasIndex(idx uint64) bool {
+	// Protect from panic, but assume callers are sophisticated enough to not need an error telling them they have an invalid idx.
+	if idx >= fieldparams.MaxBlobsPerBlock {
+		return false
+	}
+	return s.mask[idx]
+}
+
+// AllAvailable returns true if we have all blobs for all indices from 0 to count-1.
+func (s BlobStorageSummary) AllAvailable(count int) bool {
+	if count > fieldparams.MaxBlobsPerBlock {
+		return false
+	}
+	for i := 0; i < count; i++ {
+		if !s.mask[i] {
+			return false
+		}
+	}
+	return true
+}
+
+// BlobStorageSummarizer can be used to receive a summary of metadata about blobs on disk for a given root.
+// The BlobStorageSummary can be used to check which indices (if any) are available for a given block by root.
+type BlobStorageSummarizer interface {
+	Summary(root [32]byte) BlobStorageSummary
+}
+
+type blobStorageCache struct {
+	mu     sync.RWMutex
+	nBlobs float64
+	cache  map[[32]byte]BlobStorageSummary
+}
+
+var _ BlobStorageSummarizer = &blobStorageCache{}
+
+func newBlobStorageCache() *blobStorageCache {
+	return &blobStorageCache{
+		cache: make(map[[32]byte]BlobStorageSummary),
+	}
+}
+
+// Summary returns the BlobStorageSummary for `root`. The BlobStorageSummary can be used to check for the presence of
+// BlobSidecars based on Index.
+func (s *blobStorageCache) Summary(root [32]byte) BlobStorageSummary {
+	s.mu.RLock()
+	defer s.mu.RUnlock()
+	return s.cache[root]
+}
+
+func (s *blobStorageCache) ensure(key [32]byte, epoch primitives.Epoch, idx uint64) error {
+	if idx >= fieldparams.MaxBlobsPerBlock {
+		return errIndexOutOfBounds
+	}
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	v := s.cache[key]
+	v.epoch = epoch
+	if !v.mask[idx] {
+		s.updateMetrics(1)
+	}
+	v.mask[idx] = true
+	s.cache[key] = v
+	return nil
+}
+
+func (s *blobStorageCache) epoch(key [32]byte) (primitives.Epoch, bool) {
+	s.mu.RLock()
+	defer s.mu.RUnlock()
+	v, ok := s.cache[key]
+	if !ok {
+		return 0, false
+	}
+	return v.epoch, ok
+}
+
+func (s *blobStorageCache) get(key [32]byte) (BlobStorageSummary, bool) {
+	s.mu.RLock()
+	defer s.mu.RUnlock()
+	v, ok := s.cache[key]
+	return v, ok
+}
+
+func (s *blobStorageCache) identForIdx(key [32]byte, idx uint64) (blobIdent, error) {
+	v, ok := s.get(key)
+	if !ok || !v.HasIndex(idx) {
+		return blobIdent{}, db.ErrNotFound
+	}
+	return blobIdent{
+		root:  key,
+		index: idx,
+		epoch: v.epoch,
+	}, nil
+}
+
+func (s *blobStorageCache) identForRoot(key [32]byte) (blobIdent, error) {
+	v, ok := s.get(key)
+	if !ok {
+		return blobIdent{}, db.ErrNotFound
+	}
+	return blobIdent{
+		root:  key,
+		epoch: v.epoch,
+	}, nil
+}
+
+func (s *blobStorageCache) evict(key [32]byte) int {
+	deleted := 0
+	s.mu.Lock()
+	v, ok := s.cache[key]
+	if ok {
+		for i := range v.mask {
+			if v.mask[i] {
+				deleted += 1
+			}
+		}
+	}
+	delete(s.cache, key)
+	s.mu.Unlock()
+	if deleted > 0 {
+		s.updateMetrics(-float64(deleted))
+	}
+	return deleted
+}
+
+func (s *blobStorageCache) updateMetrics(delta float64) {
+	s.nBlobs += delta
+	blobDiskCount.Set(s.nBlobs)
+	blobDiskSize.Set(s.nBlobs * bytesPerSidecar)
+}

beacon-chain/db/filesystem/cache_test.go

@@ -0,0 +1,150 @@
+package filesystem
+
+import (
+	"testing"
+
+	fieldparams "github.com/prysmaticlabs/prysm/v5/config/fieldparams"
+	"github.com/prysmaticlabs/prysm/v5/encoding/bytesutil"
+	"github.com/prysmaticlabs/prysm/v5/testing/require"
+)
+
+func TestSlotByRoot_Summary(t *testing.T) {
+	var noneSet, allSet, firstSet, lastSet, oneSet blobIndexMask
+	firstSet[0] = true
+	lastSet[len(lastSet)-1] = true
+	oneSet[1] = true
+	for i := range allSet {
+		allSet[i] = true
+	}
+	cases := []struct {
+		name     string
+		root     [32]byte
+		expected *blobIndexMask
+	}{
+		{
+			name: "not found",
+		},
+		{
+			name:     "none set",
+			expected: &noneSet,
+		},
+		{
+			name:     "index 1 set",
+			expected: &oneSet,
+		},
+		{
+			name:     "all set",
+			expected: &allSet,
+		},
+		{
+			name:     "first set",
+			expected: &firstSet,
+		},
+		{
+			name:     "last set",
+			expected: &lastSet,
+		},
+	}
+	sc := newBlobStorageCache()
+	for _, c := range cases {
+		if c.expected != nil {
+			key := bytesutil.ToBytes32([]byte(c.name))
+			sc.cache[key] = BlobStorageSummary{epoch: 0, mask: *c.expected}
+		}
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			key := bytesutil.ToBytes32([]byte(c.name))
+			sum := sc.Summary(key)
+			for i := range c.expected {
+				ui := uint64(i)
+				if c.expected == nil {
+					require.Equal(t, false, sum.HasIndex(ui))
+				} else {
+					require.Equal(t, c.expected[i], sum.HasIndex(ui))
+				}
+			}
+		})
+	}
+}
+
+func TestAllAvailable(t *testing.T) {
+	idxUpTo := func(u int) []int {
+		r := make([]int, u)
+		for i := range r {
+			r[i] = i
+		}
+		return r
+	}
+	require.DeepEqual(t, []int{}, idxUpTo(0))
+	require.DeepEqual(t, []int{0}, idxUpTo(1))
+	require.DeepEqual(t, []int{0, 1, 2, 3, 4, 5}, idxUpTo(6))
+	cases := []struct {
+		name   string
+		idxSet []int
+		count  int
+		aa     bool
+	}{
+		{
+			// If there are no blobs committed, then all the committed blobs are available.
+			name:  "none in idx, 0 arg",
+			count: 0,
+			aa:    true,
+		},
+		{
+			name:  "none in idx, 1 arg",
+			count: 1,
+			aa:    false,
+		},
+		{
+			name:   "first in idx, 1 arg",
+			idxSet: []int{0},
+			count:  1,
+			aa:     true,
+		},
+		{
+			name:   "second in idx, 1 arg",
+			idxSet: []int{1},
+			count:  1,
+			aa:     false,
+		},
+		{
+			name:   "first missing, 2 arg",
+			idxSet: []int{1},
+			count:  2,
+			aa:     false,
+		},
+		{
+			name:  "all missing, 1 arg",
+			count: 6,
+			aa:    false,
+		},
+		{
+			name:  "out of bound is safe",
+			count: fieldparams.MaxBlobsPerBlock + 1,
+			aa:    false,
+		},
+		{
+			name:   "max present",
+			count:  fieldparams.MaxBlobsPerBlock,
+			idxSet: idxUpTo(fieldparams.MaxBlobsPerBlock),
+			aa:     true,
+		},
+		{
+			name:   "one present",
+			count:  1,
+			idxSet: idxUpTo(1),
+			aa:     true,
+		},
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			var mask blobIndexMask
+			for _, idx := range c.idxSet {
+				mask[idx] = true
+			}
+			sum := BlobStorageSummary{mask: mask}
+			require.Equal(t, c.aa, sum.AllAvailable(c.count))
+		})
+	}
+}

beacon-chain/db/filesystem/ephemeral.go

@@ -1,63 +0,0 @@
-package filesystem
-
-import (
-	"testing"
-
-	"github.com/prysmaticlabs/prysm/v5/config/params"
-	"github.com/spf13/afero"
-)
-
-// NewEphemeralBlobStorage should only be used for tests.
-// The instance of BlobStorage returned is backed by an in-memory virtual filesystem,
-// improving test performance and simplifying cleanup.
-func NewEphemeralBlobStorage(t testing.TB) *BlobStorage {
-	fs := afero.NewMemMapFs()
-	pruner, err := newBlobPruner(fs, params.BeaconConfig().MinEpochsForBlobsSidecarsRequest)
-	if err != nil {
-		t.Fatal("test setup issue", err)
-	}
-	return &BlobStorage{fs: fs, pruner: pruner}
-}
-
-// NewEphemeralBlobStorageWithFs can be used by tests that want access to the virtual filesystem
-// in order to interact with it outside the parameters of the BlobStorage api.
-func NewEphemeralBlobStorageWithFs(t testing.TB) (afero.Fs, *BlobStorage, error) {
-	fs := afero.NewMemMapFs()
-	pruner, err := newBlobPruner(fs, params.BeaconConfig().MinEpochsForBlobsSidecarsRequest)
-	if err != nil {
-		t.Fatal("test setup issue", err)
-	}
-	return fs, &BlobStorage{fs: fs, pruner: pruner}, nil
-}
-
-type BlobMocker struct {
-	fs afero.Fs
-	bs *BlobStorage
-}
-
-// CreateFakeIndices creates empty blob sidecar files at the expected path for the given
-// root and indices to influence the result of Indices().
-func (bm *BlobMocker) CreateFakeIndices(root [32]byte, indices []uint64) error {
-	for i := range indices {
-		n := blobNamer{root: root, index: indices[i]}
-		if err := bm.fs.MkdirAll(n.dir(), directoryPermissions); err != nil {
-			return err
-		}
-		f, err := bm.fs.Create(n.path())
-		if err != nil {
-			return err
-		}
-		if err := f.Close(); err != nil {
-			return err
-		}
-	}
-	return nil
-}
-
-// NewEphemeralBlobStorageWithMocker returns a *BlobMocker value in addition to the BlobStorage value.
-// BlockMocker encapsulates things blob path construction to avoid leaking implementation details.
-func NewEphemeralBlobStorageWithMocker(_ testing.TB) (*BlobMocker, *BlobStorage) {
-	fs := afero.NewMemMapFs()
-	bs := &BlobStorage{fs: fs}
-	return &BlobMocker{fs: fs, bs: bs}, bs
-}

beacon-chain/db/filesystem/iteration.go

@@ -0,0 +1,328 @@
+package filesystem
+
+import (
+	"encoding/binary"
+	"fmt"
+	"io"
+	"path"
+	"path/filepath"
+	"strconv"
+	"strings"
+
+	"github.com/ethereum/go-ethereum/common/hexutil"
+	"github.com/pkg/errors"
+	fieldparams "github.com/prysmaticlabs/prysm/v5/config/fieldparams"
+	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
+	"github.com/prysmaticlabs/prysm/v5/encoding/bytesutil"
+	"github.com/prysmaticlabs/prysm/v5/time/slots"
+	"github.com/sirupsen/logrus"
+	"github.com/spf13/afero"
+)
+
+var errIdentFailure = errors.New("failed to determine blob metadata, ignoring all sub-path.")
+
+type identificationError struct {
+	err   error
+	path  string
+	ident blobIdent
+}
+
+func (ide *identificationError) Error() string {
+	return fmt.Sprintf("%s path=%s, err=%s", errIdentFailure.Error(), ide.path, ide.err.Error())
+}
+
+func (ide *identificationError) Unwrap() error {
+	return ide.err
+}
+
+func (ide *identificationError) Is(err error) bool {
+	return err == errIdentFailure
+}
+
+func (ide *identificationError) LogFields() logrus.Fields {
+	fields := ide.ident.logFields()
+	fields["path"] = ide.path
+	return fields
+}
+
+func newIdentificationError(path string, ident blobIdent, err error) *identificationError {
+	return &identificationError{path: path, ident: ident, err: err}
+}
+
+func listDir(fs afero.Fs, dir string) ([]string, error) {
+	top, err := fs.Open(dir)
+	if err != nil {
+		return nil, errors.Wrap(err, "failed to open directory descriptor")
+	}
+	defer func() {
+		if err := top.Close(); err != nil {
+			log.WithError(err).Errorf("Could not close file %s", dir)
+		}
+	}()
+	// re the -1 param: "If n <= 0, Readdirnames returns all the names from the directory in a single slice"
+	dirs, err := top.Readdirnames(-1)
+	if err != nil {
+		return nil, errors.Wrap(err, "failed to read directory listing")
+	}
+	return dirs, nil
+}
+
+type layoutLevel struct {
+	populateIdent identPopulator
+	filter        func(string) bool
+}
+
+type identPopulator func(blobIdent, string) (blobIdent, error)
+
+type identIterator struct {
+	fs      afero.Fs
+	path    string
+	child   *identIterator
+	ident   blobIdent
+	levels  []layoutLevel
+	entries []string
+	offset  int
+}
+
+func (iter *identIterator) next() (blobIdent, error) {
+	if iter.child != nil {
+		next, err := iter.child.next()
+		if err == nil {
+			return next, nil
+		}
+		if err != io.EOF {
+			return blobIdent{}, err
+		}
+	}
+	return iter.advanceChild()
+}
+
+func (iter *identIterator) advanceChild() (blobIdent, error) {
+	defer func() {
+		iter.offset += 1
+	}()
+	for i := iter.offset; i < len(iter.entries); i++ {
+		iter.offset = i
+		nextPath := filepath.Join(iter.path, iter.entries[iter.offset])
+		nextLevel := iter.levels[0]
+		if !nextLevel.filter(nextPath) {
+			continue
+		}
+		ident, err := nextLevel.populateIdent(iter.ident, nextPath)
+		if err != nil {
+			return ident, newIdentificationError(nextPath, ident, err)
+		}
+		// if we're at the leaf level, we can return the updated ident.
+		if len(iter.levels) == 1 {
+			return ident, nil
+		}
+
+		entries, err := listDir(iter.fs, nextPath)
+		if err != nil {
+			return blobIdent{}, err
+		}
+		if len(entries) == 0 {
+			continue
+		}
+		iter.child = &identIterator{
+			fs:      iter.fs,
+			path:    nextPath,
+			ident:   ident,
+			levels:  iter.levels[1:],
+			entries: entries,
+		}
+		return iter.child.next()
+	}
+
+	return blobIdent{}, io.EOF
+}
+
+func populateNoop(namer blobIdent, dir string) (blobIdent, error) {
+	return namer, nil
+}
+
+func populateEpoch(namer blobIdent, dir string) (blobIdent, error) {
+	epoch, err := epochFromPath(dir)
+	if err != nil {
+		return namer, err
+	}
+	namer.epoch = epoch
+	return namer, nil
+}
+
+func populateRoot(namer blobIdent, dir string) (blobIdent, error) {
+	root, err := rootFromPath(dir)
+	if err != nil {
+		return namer, err
+	}
+	namer.root = root
+	return namer, nil
+}
+
+func populateIndex(namer blobIdent, fname string) (blobIdent, error) {
+	idx, err := idxFromPath(fname)
+	if err != nil {
+		return namer, err
+	}
+	namer.index = idx
+	return namer, nil
+}
+
+type readSlotOncePerRoot struct {
+	fs       afero.Fs
+	lastRoot [32]byte
+	epoch    primitives.Epoch
+}
+
+func (l *readSlotOncePerRoot) populateIdent(ident blobIdent, fname string) (blobIdent, error) {
+	ident, err := populateIndex(ident, fname)
+	if err != nil {
+		return ident, err
+	}
+	if ident.root != l.lastRoot {
+		slot, err := slotFromFile(fname, l.fs)
+		if err != nil {
+			return ident, err
+		}
+		l.lastRoot = ident.root
+		l.epoch = slots.ToEpoch(slot)
+	}
+	ident.epoch = l.epoch
+	return ident, nil
+}
+
+func epochFromPath(p string) (primitives.Epoch, error) {
+	subdir := filepath.Base(p)
+	epoch, err := strconv.ParseUint(subdir, 10, 64)
+	if err != nil {
+		return 0, errors.Wrapf(errInvalidDirectoryLayout,
+			"failed to decode epoch as uint, err=%s, dir=%s", err.Error(), p)
+	}
+	return primitives.Epoch(epoch), nil
+}
+
+func periodFromPath(p string) (uint64, error) {
+	subdir := filepath.Base(p)
+	period, err := strconv.ParseUint(subdir, 10, 64)
+	if err != nil {
+		return 0, errors.Wrapf(errInvalidDirectoryLayout,
+			"failed to decode period from path as uint, err=%s, dir=%s", err.Error(), p)
+	}
+	return period, nil
+}
+
+func rootFromPath(p string) ([32]byte, error) {
+	subdir := filepath.Base(p)
+	root, err := stringToRoot(subdir)
+	if err != nil {
+		return root, errors.Wrapf(err, "invalid directory, could not parse subdir as root %s", p)
+	}
+	return root, nil
+}
+
+func idxFromPath(p string) (uint64, error) {
+	p = path.Base(p)
+
+	if !isSszFile(p) {
+		return 0, errors.Wrap(errNotBlobSSZ, "does not have .ssz extension")
+	}
+	parts := strings.Split(p, ".")
+	if len(parts) != 2 {
+		return 0, errors.Wrap(errNotBlobSSZ, "unexpected filename structure (want <index>.ssz)")
+	}
+	idx, err := strconv.ParseUint(parts[0], 10, 64)
+	if err != nil {
+		return 0, err
+	}
+	if idx >= fieldparams.MaxBlobsPerBlock {
+		return 0, errors.Wrapf(errIndexOutOfBounds, "index=%d", idx)
+	}
+	return idx, nil
+}
+
+// Read slot from marshaled BlobSidecar data in the given file. See slotFromBlob for details.
+func slotFromFile(name string, fs afero.Fs) (primitives.Slot, error) {
+	f, err := fs.Open(name)
+	if err != nil {
+		return 0, err
+	}
+	defer func() {
+		if err := f.Close(); err != nil {
+			log.WithError(err).Errorf("Could not close blob file")
+		}
+	}()
+	return slotFromBlob(f)
+}
+
+// slotFromBlob reads the ssz data of a file at the specified offset (8 + 131072 + 48 + 48 = 131176 bytes),
+// which is calculated based on the size of the BlobSidecar struct and is based on the size of the fields
+// preceding the slot information within SignedBeaconBlockHeader.
+func slotFromBlob(at io.ReaderAt) (primitives.Slot, error) {
+	b := make([]byte, 8)
+	_, err := at.ReadAt(b, 131176)
+	if err != nil {
+		return 0, err
+	}
+	rawSlot := binary.LittleEndian.Uint64(b)
+	return primitives.Slot(rawSlot), nil
+}
+
+func filterNoop(_ string) bool {
+	return true
+}
+
+func isRootDir(p string) bool {
+	dir := filepath.Base(p)
+	return len(dir) == rootStringLen && strings.HasPrefix(dir, "0x")
+}
+
+func isSszFile(s string) bool {
+	return filepath.Ext(s) == "."+sszExt
+}
+
+func isBeforeEpoch(before primitives.Epoch) func(string) bool {
+	if before == 0 {
+		return filterNoop
+	}
+	return func(p string) bool {
+		epoch, err := epochFromPath(p)
+		if err != nil {
+			return false
+		}
+		return epoch < before
+	}
+}
+
+func isBeforePeriod(before primitives.Epoch) func(string) bool {
+	if before == 0 {
+		return filterNoop
+	}
+	beforePeriod := periodForEpoch(before)
+	if before%4096 != 0 {
+		// Add one because we need to include the period the epoch is in, unless it is the first epoch in the period,
+		// in which case we can just look at any previous period.
+		beforePeriod += 1
+	}
+	return func(p string) bool {
+		period, err := periodFromPath(p)
+		if err != nil {
+			return false
+		}
+		return primitives.Epoch(period) < beforePeriod
+	}
+}
+
+func rootToString(root [32]byte) string {
+	return fmt.Sprintf("%#x", root)
+}
+
+func stringToRoot(str string) ([32]byte, error) {
+	if len(str) != rootStringLen {
+		return [32]byte{}, errors.Wrapf(errInvalidRootString, "incorrect len for input=%s", str)
+	}
+	slice, err := hexutil.Decode(str)
+	if err != nil {
+		return [32]byte{}, errors.Wrapf(errInvalidRootString, "input=%s", str)
+	}
+	return bytesutil.ToBytes32(slice), nil
+}

beacon-chain/db/filesystem/iteration_test.go

@@ -0,0 +1,242 @@
+package filesystem
+
+import (
+	"bytes"
+	"fmt"
+	"math"
+	"os"
+	"path"
+	"sort"
+	"testing"
+
+	"github.com/prysmaticlabs/prysm/v5/beacon-chain/verification"
+	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
+	"github.com/prysmaticlabs/prysm/v5/testing/require"
+	"github.com/prysmaticlabs/prysm/v5/testing/util"
+	"github.com/spf13/afero"
+)
+
+func TestRootFromDir(t *testing.T) {
+	cases := []struct {
+		name string
+		dir  string
+		err  error
+		root [32]byte
+	}{
+		{
+			name: "happy path",
+			dir:  "0xffff875e1d985c5ccb214894983f2428edb271f0f87b68ba7010e4a99df3b5cb",
+			root: [32]byte{255, 255, 135, 94, 29, 152, 92, 92, 203, 33, 72, 148, 152, 63, 36, 40,
+				237, 178, 113, 240, 248, 123, 104, 186, 112, 16, 228, 169, 157, 243, 181, 203},
+		},
+		{
+			name: "too short",
+			dir:  "0xffff875e1d985c5ccb214894983f2428edb271f0f87b68ba7010e4a99df3b5c",
+			err:  errInvalidRootString,
+		},
+		{
+			name: "too log",
+			dir:  "0xffff875e1d985c5ccb214894983f2428edb271f0f87b68ba7010e4a99df3b5cbb",
+			err:  errInvalidRootString,
+		},
+		{
+			name: "missing prefix",
+			dir:  "ffff875e1d985c5ccb214894983f2428edb271f0f87b68ba7010e4a99df3b5cb",
+			err:  errInvalidRootString,
+		},
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			root, err := stringToRoot(c.dir)
+			if c.err != nil {
+				require.ErrorIs(t, err, c.err)
+				return
+			}
+			require.NoError(t, err)
+			require.Equal(t, c.root, root)
+		})
+	}
+}
+
+func TestSlotFromFile(t *testing.T) {
+	cases := []struct {
+		slot primitives.Slot
+	}{
+		{slot: 0},
+		{slot: 2},
+		{slot: 1123581321},
+		{slot: math.MaxUint64},
+	}
+	for _, c := range cases {
+		t.Run(fmt.Sprintf("slot %d", c.slot), func(t *testing.T) {
+			fs, bs := NewEphemeralBlobStorageAndFs(t)
+			_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, c.slot, 1)
+			sc := verification.FakeVerifyForTest(t, sidecars[0])
+			require.NoError(t, bs.Save(sc))
+			namer := identForSidecar(sc)
+			sszPath := bs.layout.sszPath(namer)
+			slot, err := slotFromFile(sszPath, fs)
+			require.NoError(t, err)
+			require.Equal(t, c.slot, slot)
+		})
+	}
+}
+
+type dirFiles struct {
+	name     string
+	isDir    bool
+	children []dirFiles
+}
+
+func (df dirFiles) reify(t *testing.T, fs afero.Fs, base string) {
+	fullPath := path.Join(base, df.name)
+	if df.isDir {
+		if df.name != "" {
+			require.NoError(t, fs.Mkdir(fullPath, directoryPermissions))
+		}
+		for _, c := range df.children {
+			c.reify(t, fs, fullPath)
+		}
+	} else {
+		fp, err := fs.Create(fullPath)
+		require.NoError(t, err)
+		_, err = fp.WriteString("derp")
+		require.NoError(t, err)
+	}
+}
+
+func (df dirFiles) childNames() []string {
+	cn := make([]string, len(df.children))
+	for i := range df.children {
+		cn[i] = df.children[i].name
+	}
+	return cn
+}
+
+func TestListDir(t *testing.T) {
+	fs := afero.NewMemMapFs()
+	rootStrs := []string{
+		"0x0023dc5d063c7c1b37016bb54963c6ff4bfe5dfdf6dac29e7ceeb2b8fa81ed7a",
+		"0xff30526cd634a5af3a09cc9bff67f33a621fc5b975750bb4432f74df077554b4",
+		"0x23f5f795aaeb78c01fadaf3d06da2e99bd4b3622ae4dfea61b05b7d9adb119c2",
+	}
+
+	// parent directory
+	tree := dirFiles{isDir: true}
+	// break out each subdir for easier assertions
+	notABlob := dirFiles{name: "notABlob", isDir: true}
+	childlessBlob := dirFiles{name: rootStrs[0], isDir: true}
+	blobWithSsz := dirFiles{name: rootStrs[1], isDir: true,
+		children: []dirFiles{{name: "1.ssz"}, {name: "2.ssz"}},
+	}
+	blobWithSszAndTmp := dirFiles{name: rootStrs[2], isDir: true,
+		children: []dirFiles{{name: "5.ssz"}, {name: "0.part"}}}
+	tree.children = append(tree.children,
+		notABlob, childlessBlob, blobWithSsz, blobWithSszAndTmp)
+
+	topChildren := make([]string, len(tree.children))
+	for i := range tree.children {
+		topChildren[i] = tree.children[i].name
+	}
+
+	var filter = func(entries []string, filt func(string) bool) []string {
+		filtered := make([]string, 0, len(entries))
+		for i := range entries {
+			if filt(entries[i]) {
+				filtered = append(filtered, entries[i])
+			}
+		}
+		return filtered
+	}
+
+	tree.reify(t, fs, "")
+	cases := []struct {
+		name     string
+		dirPath  string
+		expected []string
+		filter   func(string) bool
+		err      error
+	}{
+		{
+			name:     "non-existent",
+			dirPath:  "derp",
+			expected: []string{},
+			err:      os.ErrNotExist,
+		},
+		{
+			name:     "empty",
+			dirPath:  childlessBlob.name,
+			expected: []string{},
+		},
+		{
+			name:     "top",
+			dirPath:  ".",
+			expected: topChildren,
+		},
+		{
+			name:     "custom filter: only notABlob",
+			dirPath:  ".",
+			expected: []string{notABlob.name},
+			filter: func(s string) bool {
+				return s == notABlob.name
+			},
+		},
+		{
+			name:     "root filter",
+			dirPath:  ".",
+			expected: []string{childlessBlob.name, blobWithSsz.name, blobWithSszAndTmp.name},
+			filter:   isRootDir,
+		},
+		{
+			name:     "ssz filter",
+			dirPath:  blobWithSsz.name,
+			expected: blobWithSsz.childNames(),
+			filter:   isSszFile,
+		},
+		{
+			name:     "ssz mixed filter",
+			dirPath:  blobWithSszAndTmp.name,
+			expected: []string{"5.ssz"},
+			filter:   isSszFile,
+		},
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			result, err := listDir(fs, c.dirPath)
+			if c.filter != nil {
+				result = filter(result, c.filter)
+			}
+			if c.err != nil {
+				require.ErrorIs(t, err, c.err)
+				require.Equal(t, 0, len(result))
+			} else {
+				require.NoError(t, err)
+				sort.Strings(c.expected)
+				sort.Strings(result)
+				require.DeepEqual(t, c.expected, result)
+			}
+		})
+	}
+}
+
+func TestSlotFromBlob(t *testing.T) {
+	cases := []struct {
+		slot primitives.Slot
+	}{
+		{slot: 0},
+		{slot: 2},
+		{slot: 1123581321},
+		{slot: math.MaxUint64},
+	}
+	for _, c := range cases {
+		t.Run(fmt.Sprintf("slot %d", c.slot), func(t *testing.T) {
+			_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, c.slot, 1)
+			sc := sidecars[0]
+			enc, err := sc.MarshalSSZ()
+			require.NoError(t, err)
+			slot, err := slotFromBlob(bytes.NewReader(enc))
+			require.NoError(t, err)
+			require.Equal(t, c.slot, slot)
+		})
+	}
+}

beacon-chain/db/filesystem/layout.go

@@ -0,0 +1,330 @@
+package filesystem
+
+import (
+	"fmt"
+	"io"
+	"path"
+	"path/filepath"
+	"time"
+
+	"github.com/pkg/errors"
+	"github.com/prysmaticlabs/prysm/v5/config/params"
+	"github.com/prysmaticlabs/prysm/v5/consensus-types/blocks"
+	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
+	"github.com/prysmaticlabs/prysm/v5/time/slots"
+	"github.com/sirupsen/logrus"
+	"github.com/spf13/afero"
+)
+
+const (
+	rootPrefixLen = 4
+	// Full root in directory will be 66 chars, eg:
+	// >>> len('0x0002fb4db510b8618b04dc82d023793739c26346a8b02eb73482e24b0fec0555') == 66
+	rootStringLen        = 66
+	sszExt               = "ssz"
+	partExt              = "part"
+	periodicEpochBaseDir = "by-epoch"
+	hexPrefixBaseDir     = "by-hex-prefix"
+)
+
+var (
+	errMigrationFailure       = errors.New("unable to migrate blob directory between old and new layout")
+	errCacheWarmFailed        = errors.New("failed to warm blob filesystem cache")
+	errPruneFailed            = errors.New("failed to prune root")
+	errInvalidRootString      = errors.New("Could not parse hex string as a [32]byte")
+	errInvalidDirectoryLayout = errors.New("Could not parse blob directory path")
+)
+
+type migratableLayout interface {
+	dir(n blobIdent) string
+	sszPath(n blobIdent) string
+	partPath(n blobIdent, entropy string) string
+	iterateIdents(before primitives.Epoch) (*identIterator, error)
+}
+
+type runtimeLayout interface {
+	migratableLayout
+	ident(root [32]byte, idx uint64) (blobIdent, error)
+	dirIdent(root [32]byte) (blobIdent, error)
+	summary(root [32]byte) BlobStorageSummary
+	notify(ident blobIdent) error
+	pruneBefore(before primitives.Epoch) (*pruneSummary, error)
+	remove(ident blobIdent) (int, error)
+}
+
+func warmCache(l runtimeLayout, cache *blobStorageCache) error {
+	iter, err := l.iterateIdents(0)
+	if err != nil {
+		return errors.Wrap(errCacheWarmFailed, err.Error())
+	}
+	for ident, err := iter.next(); err != io.EOF; ident, err = iter.next() {
+		if errors.Is(err, errIdentFailure) {
+			idf := &identificationError{}
+			if errors.As(err, &idf) {
+				log.WithFields(idf.LogFields()).WithError(err).Error("Failed to cache blob data for path")
+			}
+			continue
+		}
+		if err != nil {
+			return errors.Wrapf(errCacheWarmFailed, "failed to populate blob data cache err=%s", err.Error())
+		}
+		if err := cache.ensure(ident.root, ident.epoch, ident.index); err != nil {
+			return errors.Wrapf(errCacheWarmFailed, "failed to write cache entry for %s, err=%s", l.sszPath(ident), err.Error())
+		}
+	}
+	return nil
+}
+
+func migrateLayout(fs afero.Fs, from, to migratableLayout, cache *blobStorageCache) error {
+	start := time.Now()
+	iter, err := from.iterateIdents(0)
+	if err != nil {
+		return errors.Wrapf(errMigrationFailure, "failed to iterate legacy structure while migrating blobs, err=%s", err.Error())
+	}
+	lastMoved := ""
+	parentDirs := make(map[string]bool) // this map should have < 65k keys by design
+	moved := 0
+	for ident, err := iter.next(); err != io.EOF; ident, err = iter.next() {
+		if err != nil {
+			if errors.Is(err, errIdentFailure) {
+				idf := &identificationError{}
+				if errors.As(err, &idf) {
+					log.WithFields(idf.LogFields()).WithError(err).Error("Failed to migrate blob path")
+				}
+				continue
+			}
+			return errors.Wrapf(errMigrationFailure, "failed to iterate legacy structure while migrating blobs, err=%s", err.Error())
+		}
+		src := from.dir(ident)
+		target := to.dir(ident)
+		if src != lastMoved {
+			targetParent := filepath.Dir(target)
+			if targetParent != "" && targetParent != "." && !parentDirs[targetParent] {
+				if err := fs.MkdirAll(targetParent, directoryPermissions); err != nil {
+					return errors.Wrapf(errMigrationFailure, "failed to make enclosing path before moving %s to %s", src, target)
+				}
+				parentDirs[targetParent] = true
+			}
+			if err := fs.Rename(src, target); err != nil {
+				return errors.Wrapf(errMigrationFailure, "could not rename %s to %s", src, target)
+			}
+			moved += 1
+			lastMoved = src
+		}
+		if err := cache.ensure(ident.root, ident.epoch, ident.index); err != nil {
+			return errors.Wrapf(errMigrationFailure, "could not cache path %s, err=%s", to.sszPath(ident), err.Error())
+		}
+	}
+	if moved > 0 {
+		log.WithField("dirsMoved", moved).WithField("elapsed", time.Since(start)).
+			Info("Blob filesystem migration complete.")
+	}
+	return nil
+}
+
+type blobIdent struct {
+	root  [32]byte
+	epoch primitives.Epoch
+	index uint64
+}
+
+func newBlobIdent(root [32]byte, epoch primitives.Epoch, index uint64) blobIdent {
+	return blobIdent{root: root, epoch: epoch, index: index}
+}
+
+func identForSidecar(sc blocks.VerifiedROBlob) blobIdent {
+	return newBlobIdent(sc.BlockRoot(), slots.ToEpoch(sc.Slot()), sc.Index)
+}
+
+func (n blobIdent) sszFname() string {
+	return fmt.Sprintf("%d.%s", n.index, sszExt)
+}
+
+func (n blobIdent) partFname(entropy string) string {
+	return fmt.Sprintf("%s-%d.%s", entropy, n.index, partExt)
+}
+
+func (n blobIdent) logFields() logrus.Fields {
+	return logrus.Fields{
+		"root":  fmt.Sprintf("%#x", n.root),
+		"epoch": n.epoch,
+		"index": n.index,
+	}
+}
+
+type pruneSummary struct {
+	blobsPruned    int
+	failedRemovals []string
+}
+
+func (s pruneSummary) LogFields() logrus.Fields {
+	return logrus.Fields{}
+}
+
+func newPeriodicEpochLayout(fs afero.Fs, cache *blobStorageCache, pruner *blobPruner) (*periodicEpochLayout, error) {
+	l := &periodicEpochLayout{fs: fs, cache: cache, pruner: pruner}
+	if err := l.initialize(); err != nil {
+		return nil, err
+	}
+	return l, nil
+}
+
+var _ migratableLayout = &flatRootLayout{}
+var _ runtimeLayout = &periodicEpochLayout{}
+
+type periodicEpochLayout struct {
+	fs     afero.Fs
+	cache  *blobStorageCache
+	pruner *blobPruner
+}
+
+func (l *periodicEpochLayout) notify(ident blobIdent) error {
+	if err := l.cache.ensure(ident.root, ident.epoch, ident.index); err != nil {
+		return err
+	}
+	l.pruner.notify(ident.epoch, l)
+	return nil
+}
+
+func (l *periodicEpochLayout) initialize() error {
+	return l.fs.MkdirAll(periodicEpochBaseDir, directoryPermissions)
+}
+
+// If before == 0, it won't be used as a filter and all idents will be returned.
+func (l *periodicEpochLayout) iterateIdents(before primitives.Epoch) (*identIterator, error) {
+	// iterate root, which should have directories named by "period"
+	entries, err := listDir(l.fs, periodicEpochBaseDir)
+	if err != nil {
+		return nil, errors.Wrapf(err, "failed to list %s", periodicEpochBaseDir)
+	}
+
+	return &identIterator{
+		fs:   l.fs,
+		path: periodicEpochBaseDir,
+		levels: []layoutLevel{
+			{populateIdent: populateNoop, filter: isBeforePeriod(before)},
+			{populateIdent: populateEpoch, filter: isBeforeEpoch(before)},
+			{populateIdent: populateRoot, filter: isRootDir},  // extract root from path
+			{populateIdent: populateIndex, filter: isSszFile}, // extract index from filename
+		},
+		entries: entries,
+	}, nil
+}
+
+func (l *periodicEpochLayout) ident(root [32]byte, idx uint64) (blobIdent, error) {
+	return l.cache.identForIdx(root, idx)
+}
+
+func (l *periodicEpochLayout) dirIdent(root [32]byte) (blobIdent, error) {
+	return l.cache.identForRoot(root)
+}
+
+func (l *periodicEpochLayout) summary(root [32]byte) BlobStorageSummary {
+	return l.cache.Summary(root)
+}
+
+func (l *periodicEpochLayout) dir(n blobIdent) string {
+	return filepath.Join(l.epochDir(n.epoch), rootToString(n.root))
+}
+
+func (l *periodicEpochLayout) epochDir(epoch primitives.Epoch) string {
+	return filepath.Join(periodicEpochBaseDir, fmt.Sprintf("%d", periodForEpoch(epoch)), fmt.Sprintf("%d", epoch))
+}
+
+func periodForEpoch(epoch primitives.Epoch) primitives.Epoch {
+	return epoch / params.BeaconConfig().MinEpochsForBlobsSidecarsRequest
+}
+
+func (l *periodicEpochLayout) sszPath(n blobIdent) string {
+	return filepath.Join(l.dir(n), n.sszFname())
+}
+
+func (l *periodicEpochLayout) partPath(n blobIdent, entropy string) string {
+	return path.Join(l.dir(n), n.partFname(entropy))
+}
+
+func (l *periodicEpochLayout) pruneBefore(before primitives.Epoch) (*pruneSummary, error) {
+	sums := make(map[primitives.Epoch]*pruneSummary)
+	iter, err := l.iterateIdents(before)
+
+	rollup := &pruneSummary{}
+	for ident, err := iter.next(); err != io.EOF; ident, err = iter.next() {
+		if err != nil {
+			if errors.Is(err, errIdentFailure) {
+				idf := &identificationError{}
+				if errors.As(err, &idf) {
+					log.WithFields(idf.LogFields()).WithError(err).Error("Failed to prune blob path due to identification errors")
+				}
+				continue
+			}
+			log.WithError(err).Error("encountered unhandled error during pruning")
+			return nil, errors.Wrap(errPruneFailed, err.Error())
+		}
+		_, ok := sums[ident.epoch]
+		if !ok {
+			sums[ident.epoch] = &pruneSummary{}
+		}
+		s := sums[ident.epoch]
+		removed, err := l.remove(ident)
+		if err != nil {
+			s.failedRemovals = append(s.failedRemovals, l.dir(ident))
+			log.WithField("root", fmt.Sprintf("%#x", ident.root)).Error("Failed to delete blob directory for root")
+		}
+		s.blobsPruned += removed
+	}
+
+	// Roll up summaries and clean up per-epoch directories.
+	for epoch, sum := range sums {
+		rollup.blobsPruned += sum.blobsPruned
+		rollup.failedRemovals = append(rollup.failedRemovals, sum.failedRemovals...)
+		rmdir := l.epochDir(epoch)
+		if len(sum.failedRemovals) == 0 {
+			if err := l.fs.Remove(rmdir); err != nil {
+				log.WithField("dir", rmdir).WithError(err).Error("Failed to remove epoch directory while pruning")
+			}
+		} else {
+			log.WithField("dir", rmdir).WithField("numFailed", len(sum.failedRemovals)).WithError(err).Error("Unable to remove epoch directory due to pruning failures")
+		}
+	}
+
+	return rollup, nil
+}
+
+func (l *periodicEpochLayout) remove(ident blobIdent) (int, error) {
+	removed := l.cache.evict(ident.root)
+	if err := l.fs.RemoveAll(l.dir(ident)); err != nil {
+		return removed, err
+	}
+	return removed, nil
+}
+
+type flatRootLayout struct {
+	fs afero.Fs
+}
+
+func (l *flatRootLayout) iterateIdents(_ primitives.Epoch) (*identIterator, error) {
+	entries, err := listDir(l.fs, ".")
+	if err != nil {
+		return nil, errors.Wrapf(err, "could not list root directory")
+	}
+	slotAndIndex := &readSlotOncePerRoot{fs: l.fs}
+	return &identIterator{
+		fs: l.fs,
+		levels: []layoutLevel{
+			{populateIdent: populateRoot, filter: isRootDir},
+			{populateIdent: slotAndIndex.populateIdent, filter: isSszFile}},
+		entries: entries,
+	}, nil
+}
+
+func (l *flatRootLayout) dir(n blobIdent) string {
+	return rootToString(n.root)
+}
+
+func (l *flatRootLayout) sszPath(n blobIdent) string {
+	return path.Join(l.dir(n), n.sszFname())
+}
+
+func (l *flatRootLayout) partPath(n blobIdent, entropy string) string {
+	return path.Join(l.dir(n), n.partFname(entropy))
+}

beacon-chain/db/filesystem/layout_test.go

@@ -0,0 +1,52 @@
+package filesystem
+
+import (
+	"github.com/prysmaticlabs/prysm/v5/consensus-types/blocks"
+	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
+)
+
+type mockLayout struct {
+	pruneBeforeFunc func(primitives.Epoch) (*pruneSummary, error)
+}
+
+func (m *mockLayout) dir(n blobIdent) string {
+	return ""
+}
+
+func (m *mockLayout) sszPath(n blobIdent) string {
+	return ""
+}
+
+func (m *mockLayout) partPath(n blobIdent, entropy string) string {
+	return ""
+}
+
+func (m *mockLayout) iterateIdents(before primitives.Epoch) (*identIterator, error) {
+	return nil, nil
+}
+
+func (m *mockLayout) ident(root [32]byte, idx uint64) (blobIdent, error) {
+	return blobIdent{}, nil
+}
+
+func (m *mockLayout) dirIdent(root [32]byte) (blobIdent, error) {
+	return blobIdent{}, nil
+}
+
+func (m *mockLayout) summary(root [32]byte) BlobStorageSummary {
+	return BlobStorageSummary{}
+}
+
+func (m *mockLayout) notify(sidecar blocks.VerifiedROBlob) error {
+	return nil
+}
+
+func (m *mockLayout) pruneBefore(before primitives.Epoch) (*pruneSummary, error) {
+	return m.pruneBeforeFunc(before)
+}
+
+func (m *mockLayout) remove(ident blobIdent) (int, error) {
+	return 0, nil
+}
+
+var _ runtimeLayout = &mockLayout{}

beacon-chain/db/filesystem/log.go

@@ -0,0 +1,5 @@
+package filesystem
+
+import "github.com/sirupsen/logrus"
+
+var log = logrus.WithField("prefix", "filesystem")

beacon-chain/db/filesystem/migration_test.go

@@ -0,0 +1,194 @@
+package filesystem
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
+	"github.com/prysmaticlabs/prysm/v5/testing/require"
+	"github.com/prysmaticlabs/prysm/v5/testing/util"
+	"github.com/prysmaticlabs/prysm/v5/time/slots"
+	"github.com/spf13/afero"
+)
+
+func testSetupPaths(t *testing.T, fs afero.Fs, paths []migrateBeforeAfter) {
+	for _, ba := range paths {
+		slot, err := slots.EpochStart(ba.epoch)
+		require.NoError(t, err)
+		slot += ba.slotOffset
+		_, sc := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, slot, 1)
+		scb, err := sc[0].MarshalSSZ()
+		require.NoError(t, err)
+		p := ba.before
+		dir := filepath.Dir(p)
+		require.NoError(t, fs.MkdirAll(dir, directoryPermissions))
+		require.NoError(t, afero.WriteFile(fs, p, scb, 0666))
+		_, err = fs.Stat(ba.before)
+		require.NoError(t, err)
+	}
+}
+
+func testAssertNewPaths(t *testing.T, fs afero.Fs, bs *BlobStorage, paths []migrateBeforeAfter) {
+	for _, ba := range paths {
+		if ba.before != ba.after {
+			_, err := fs.Stat(ba.before)
+			require.ErrorIs(t, err, os.ErrNotExist)
+			dir := filepath.Dir(ba.before)
+			_, err = listDir(fs, dir)
+			require.ErrorIs(t, err, os.ErrNotExist)
+		}
+		_, err := fs.Stat(ba.after)
+		require.NoError(t, err)
+		root, err := stringToRoot(ba.root)
+		require.NoError(t, err)
+		namer, err := bs.layout.ident(root, ba.index)
+		require.NoError(t, err)
+		path := bs.layout.sszPath(namer)
+		require.Equal(t, ba.after, path)
+	}
+}
+
+type migrateBeforeAfter struct {
+	before     string
+	after      string
+	epoch      primitives.Epoch
+	slotOffset primitives.Slot
+	index      uint64
+	root       string
+}
+
+func TestPeriodicEpochMigrator(t *testing.T) {
+	cases := []struct {
+		name string
+		plan []migrateBeforeAfter
+		err  error
+	}{
+		{
+			name: "happy path",
+			plan: []migrateBeforeAfter{
+				{
+					before:     "0x0125e54c64c925018c9296965a5b622d9f5ab626c10917860dcfb6aa09a0a00b/0.ssz",
+					epoch:      1234,
+					slotOffset: 0,
+					root:       "0x0125e54c64c925018c9296965a5b622d9f5ab626c10917860dcfb6aa09a0a00b",
+					index:      0,
+					after:      periodicEpochBaseDir + "/0/1234/0x0125e54c64c925018c9296965a5b622d9f5ab626c10917860dcfb6aa09a0a00b/0.ssz",
+				},
+				{
+					before:     "0x0127dba6fd30fdbb47e73e861d5c6e602b38ac3ddc945bb6a2fc4e10761e9a86/0.ssz",
+					root:       "0x0127dba6fd30fdbb47e73e861d5c6e602b38ac3ddc945bb6a2fc4e10761e9a86",
+					index:      0,
+					epoch:      5330,
+					slotOffset: 0,
+					after:      periodicEpochBaseDir + "/1/5330/0x0127dba6fd30fdbb47e73e861d5c6e602b38ac3ddc945bb6a2fc4e10761e9a86/0.ssz",
+				},
+				{
+					before:     "0x0127dba6fd30fdbb47e73e861d5c6e602b38ac3ddc945bb6a2fc4e10761e9a86/1.ssz",
+					root:       "0x0127dba6fd30fdbb47e73e861d5c6e602b38ac3ddc945bb6a2fc4e10761e9a86",
+					index:      1,
+					epoch:      5330,
+					slotOffset: 31,
+					after:      periodicEpochBaseDir + "/1/5330/0x0127dba6fd30fdbb47e73e861d5c6e602b38ac3ddc945bb6a2fc4e10761e9a86/1.ssz",
+				},
+				{
+					before:     "0x0232521756a0b965eab2c2245d7ad85feaeaf5f427cd14d1a7531f9d555b415c/0.ssz",
+					root:       "0x0232521756a0b965eab2c2245d7ad85feaeaf5f427cd14d1a7531f9d555b415c",
+					index:      0,
+					epoch:      16777216,
+					slotOffset: 16,
+					after:      periodicEpochBaseDir + "/4096/16777216/0x0232521756a0b965eab2c2245d7ad85feaeaf5f427cd14d1a7531f9d555b415c/0.ssz",
+				},
+			},
+		},
+		{
+			name: "mix old and new",
+			plan: []migrateBeforeAfter{
+				{
+					before:     "0x0125e54c64c925018c9296965a5b622d9f5ab626c10917860dcfb6aa09a0a00b/0.ssz",
+					root:       "0x0125e54c64c925018c9296965a5b622d9f5ab626c10917860dcfb6aa09a0a00b",
+					index:      0,
+					epoch:      1234,
+					slotOffset: 0,
+					after:      periodicEpochBaseDir + "/0/1234/0x0125e54c64c925018c9296965a5b622d9f5ab626c10917860dcfb6aa09a0a00b/0.ssz",
+				},
+				{
+					before:     "0x0127dba6fd30fdbb47e73e861d5c6e602b38ac3ddc945bb6a2fc4e10761e9a86/0.ssz",
+					root:       "0x0127dba6fd30fdbb47e73e861d5c6e602b38ac3ddc945bb6a2fc4e10761e9a86",
+					index:      0,
+					epoch:      5330,
+					slotOffset: 0,
+					after:      periodicEpochBaseDir + "/1/5330/0x0127dba6fd30fdbb47e73e861d5c6e602b38ac3ddc945bb6a2fc4e10761e9a86/0.ssz",
+				},
+				{
+					before:     "0x0127dba6fd30fdbb47e73e861d5c6e602b38ac3ddc945bb6a2fc4e10761e9a86/1.ssz",
+					root:       "0x0127dba6fd30fdbb47e73e861d5c6e602b38ac3ddc945bb6a2fc4e10761e9a86",
+					index:      1,
+					epoch:      5330,
+					slotOffset: 31,
+					after:      periodicEpochBaseDir + "/1/5330/0x0127dba6fd30fdbb47e73e861d5c6e602b38ac3ddc945bb6a2fc4e10761e9a86/1.ssz",
+				},
+				{
+					before:     "0x0232521756a0b965eab2c2245d7ad85feaeaf5f427cd14d1a7531f9d555b415c/0.ssz",
+					root:       "0x0232521756a0b965eab2c2245d7ad85feaeaf5f427cd14d1a7531f9d555b415c",
+					index:      0,
+					epoch:      16777216,
+					slotOffset: 16,
+					after:      periodicEpochBaseDir + "/4096/16777216/0x0232521756a0b965eab2c2245d7ad85feaeaf5f427cd14d1a7531f9d555b415c/0.ssz",
+				},
+				{
+					before:     periodicEpochBaseDir + "/4096/16777217/0x42eabe3d2c125410cd226de6f2825fb7575ab896c3f52e43de1fa29e4c809aba/0.ssz",
+					root:       "0x42eabe3d2c125410cd226de6f2825fb7575ab896c3f52e43de1fa29e4c809aba",
+					index:      0,
+					epoch:      16777217,
+					slotOffset: 16,
+					after:      periodicEpochBaseDir + "/4096/16777217/0x42eabe3d2c125410cd226de6f2825fb7575ab896c3f52e43de1fa29e4c809aba/0.ssz",
+				},
+				{
+					before:     "0x0232521756a0b965eab2c2245d7ad85feaeaf5f427cd14d1a7531f9d555b415c/0.ssz",
+					root:       "0x0232521756a0b965eab2c2245d7ad85feaeaf5f427cd14d1a7531f9d555b415c",
+					index:      0,
+					epoch:      16777216,
+					slotOffset: 16,
+					after:      periodicEpochBaseDir + "/4096/16777216/0x0232521756a0b965eab2c2245d7ad85feaeaf5f427cd14d1a7531f9d555b415c/0.ssz",
+				},
+				{
+					before:     periodicEpochBaseDir + "/4096/16777216/0x2326de064f828c564740da17fc247b30d7e7300da24b0aae39a0c91791acc19f/0.ssz",
+					root:       "0x2326de064f828c564740da17fc247b30d7e7300da24b0aae39a0c91791acc19f",
+					index:      0,
+					epoch:      16777216,
+					slotOffset: 31,
+					after:      periodicEpochBaseDir + "/4096/16777216/0x2326de064f828c564740da17fc247b30d7e7300da24b0aae39a0c91791acc19f/0.ssz",
+				},
+				{
+					before:     periodicEpochBaseDir + "/2/11235/0x666cea5034e22bd3b849cb33914cad59afd88ee08e4d5bc0e997411c945fbc1d/1.ssz",
+					root:       "0x666cea5034e22bd3b849cb33914cad59afd88ee08e4d5bc0e997411c945fbc1d",
+					index:      1,
+					epoch:      11235,
+					slotOffset: 0,
+					after:      periodicEpochBaseDir + "/2/11235/0x666cea5034e22bd3b849cb33914cad59afd88ee08e4d5bc0e997411c945fbc1d/1.ssz",
+				},
+			},
+		},
+	}
+
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			fs, bs := NewEphemeralBlobStorageAndFs(t)
+			from := &flatRootLayout{fs: fs}
+			cache := newBlobStorageCache()
+			pruner := newBlobPruner(bs.retentionEpochs)
+			to, err := newPeriodicEpochLayout(fs, cache, pruner)
+			require.NoError(t, err)
+			testSetupPaths(t, fs, c.plan)
+			err = migrateLayout(fs, from, to, cache)
+			if c.err != nil {
+				require.ErrorIs(t, err, c.err)
+				return
+			}
+			require.NoError(t, err)
+			require.NoError(t, warmCache(bs.layout, bs.cache))
+			testAssertNewPaths(t, fs, bs, c.plan)
+		})
+	}
+}

beacon-chain/db/filesystem/mock.go

@@ -0,0 +1,73 @@
+package filesystem
+
+import (
+	"testing"
+
+	"github.com/prysmaticlabs/prysm/v5/config/params"
+	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
+	"github.com/prysmaticlabs/prysm/v5/time/slots"
+	"github.com/spf13/afero"
+)
+
+// NewEphemeralBlobStorage should only be used for tests.
+// The instance of BlobStorage returned is backed by an in-memory virtual filesystem,
+// improving test performance and simplifying cleanup.
+func NewEphemeralBlobStorage(t testing.TB) *BlobStorage {
+	return NewEphemeralBlobStorageUsingFs(t, afero.NewMemMapFs())
+}
+
+// NewEphemeralBlobStorageAndFs can be used by tests that want access to the virtual filesystem
+// in order to interact with it outside the parameters of the BlobStorage api.
+func NewEphemeralBlobStorageAndFs(t testing.TB) (afero.Fs, *BlobStorage) {
+	fs := afero.NewMemMapFs()
+	bs := NewEphemeralBlobStorageUsingFs(t, fs)
+	return fs, bs
+}
+
+func NewEphemeralBlobStorageUsingFs(t testing.TB, fs afero.Fs) *BlobStorage {
+	opts := []BlobStorageOption{
+		WithBlobRetentionEpochs(params.BeaconConfig().MinEpochsForBlobsSidecarsRequest),
+		WithFs(fs),
+	}
+	bs, err := NewBlobStorage(opts...)
+	if err != nil {
+		t.Fatalf("error initializing test BlobStorage, err=%s", err.Error())
+	}
+	bs.WarmCache()
+	return bs
+}
+
+type BlobMocker struct {
+	fs afero.Fs
+	bs *BlobStorage
+}
+
+// CreateFakeIndices creates empty blob sidecar files at the expected path for the given
+// root and indices to influence the result of Indices().
+func (bm *BlobMocker) CreateFakeIndices(root [32]byte, slot primitives.Slot, indices ...uint64) error {
+	for i := range indices {
+		if err := bm.bs.layout.notify(newBlobIdent(root, slots.ToEpoch(slot), indices[i])); err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// NewEphemeralBlobStorageWithMocker returns a *BlobMocker value in addition to the BlobStorage value.
+// BlockMocker encapsulates things blob path construction to avoid leaking implementation details.
+func NewEphemeralBlobStorageWithMocker(t testing.TB) (*BlobMocker, *BlobStorage) {
+	fs, bs := NewEphemeralBlobStorageAndFs(t)
+	return &BlobMocker{fs: fs, bs: bs}, bs
+}
+
+func NewMockBlobStorageSummarizer(t *testing.T, set map[[32]byte][]int) BlobStorageSummarizer {
+	c := newBlobStorageCache()
+	for k, v := range set {
+		for i := range v {
+			if err := c.ensure(k, 0, uint64(v[i])); err != nil {
+				t.Fatal(err)
+			}
+		}
+	}
+	return c
+}

beacon-chain/db/filesystem/pruner.go

@@ -1,27 +1,16 @@
 package filesystem
 
 import (
-	"encoding/binary"
-	"io"
-	"path"
-	"path/filepath"
-	"strconv"
-	"strings"
 	"sync"
 	"sync/atomic"
 	"time"
 
 	"github.com/pkg/errors"
-	fieldparams "github.com/prysmaticlabs/prysm/v5/config/fieldparams"
-	"github.com/prysmaticlabs/prysm/v5/config/params"
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
-	"github.com/prysmaticlabs/prysm/v5/time/slots"
-	log "github.com/sirupsen/logrus"
-	"github.com/spf13/afero"
+	"github.com/sirupsen/logrus"
 )
 
 const retentionBuffer primitives.Epoch = 2
-const bytesPerSidecar = 131928
 
 var (
 	errPruningFailures = errors.New("blobs could not be pruned for some roots")
@@ -29,311 +18,46 @@ var (
 )
 
 type blobPruner struct {
-	sync.Mutex
-	prunedBefore atomic.Uint64
-	windowSize   primitives.Slot
-	slotMap      *slotForRoot
-	fs           afero.Fs
+	mu              sync.Mutex
+	prunedBefore    atomic.Uint64
+	retentionPeriod primitives.Epoch
 }
 
-func newBlobPruner(fs afero.Fs, retain primitives.Epoch) (*blobPruner, error) {
-	r, err := slots.EpochStart(retain + retentionBuffer)
-	if err != nil {
-		return nil, errors.Wrap(err, "could not set retentionSlots")
-	}
-	return &blobPruner{fs: fs, windowSize: r, slotMap: newSlotForRoot()}, nil
+func newBlobPruner(retain primitives.Epoch) *blobPruner {
+	p := &blobPruner{retentionPeriod: retain + retentionBuffer}
+	return p
 }
 
-// notify updates the pruner's view of root->blob mappings. This allows the pruner to build a cache
-// of root->slot mappings and decide when to evict old blobs based on the age of present blobs.
-func (p *blobPruner) notify(root [32]byte, latest primitives.Slot, idx uint64) error {
-	if err := p.slotMap.ensure(rootString(root), latest, idx); err != nil {
-		return err
-	}
-	pruned := uint64(windowMin(latest, p.windowSize))
-	if p.prunedBefore.Swap(pruned) == pruned {
-		return nil
+func (p *blobPruner) notify(latest primitives.Epoch, layout runtimeLayout) chan struct{} {
+	done := make(chan struct{})
+	floor := periodFloor(latest, p.retentionPeriod)
+	if primitives.Epoch(p.prunedBefore.Swap(uint64(floor))) >= floor {
+		// Only trigger pruning if the atomic swap changed the previous value of prunedBefore.
+		close(done)
+		return done
 	}
 	go func() {
-		if err := p.prune(primitives.Slot(pruned)); err != nil {
-			log.WithError(err).Errorf("Failed to prune blobs from slot %d", latest)
+		p.mu.Lock()
+		start := time.Now()
+		defer p.mu.Unlock()
+		sum, err := layout.pruneBefore(floor)
+		if err != nil {
+			log.WithError(err).WithFields(sum.LogFields()).Warn("Encountered errors during blob pruning.")
 		}
+		log.WithFields(logrus.Fields{
+			"upToEpoch":    floor,
+			"duration":     time.Since(start).String(),
+			"filesRemoved": sum.blobsPruned,
+		}).Debug("Pruned old blobs")
+		blobsPrunedCounter.Add(float64(sum.blobsPruned))
+		close(done)
 	}()
-	return nil
+	return done
 }
 
-func windowMin(latest primitives.Slot, offset primitives.Slot) primitives.Slot {
-	// Safely compute the first slot in the epoch for the latest slot
-	latest = latest - latest%params.BeaconConfig().SlotsPerEpoch
-	if latest < offset {
+func periodFloor(latest, period primitives.Epoch) primitives.Epoch {
+	if latest < period {
 		return 0
 	}
-	return latest - offset
-}
-
-// Prune prunes blobs in the base directory based on the retention epoch.
-// It deletes blobs older than currentEpoch - (retentionEpochs+bufferEpochs).
-// This is so that we keep a slight buffer and blobs are deleted after n+2 epochs.
-func (p *blobPruner) prune(pruneBefore primitives.Slot) error {
-	p.Lock()
-	defer p.Unlock()
-	start := time.Now()
-	totalPruned, totalErr := 0, 0
-	// Customize logging/metrics behavior for the initial cache warmup when slot=0.
-	// We'll never see a prune request for slot 0, unless this is the initial call to warm up the cache.
-	if pruneBefore == 0 {
-		defer func() {
-			log.WithField("duration", time.Since(start).String()).Debug("Warmed up pruner cache")
-		}()
-	} else {
-		defer func() {
-			log.WithFields(log.Fields{
-				"upToEpoch":    slots.ToEpoch(pruneBefore),
-				"duration":     time.Since(start).String(),
-				"filesRemoved": totalPruned,
-			}).Debug("Pruned old blobs")
-			blobsPrunedCounter.Add(float64(totalPruned))
-		}()
-	}
-
-	entries, err := listDir(p.fs, ".")
-	if err != nil {
-		return errors.Wrap(err, "unable to list root blobs directory")
-	}
-	dirs := filter(entries, filterRoot)
-	for _, dir := range dirs {
-		pruned, err := p.tryPruneDir(dir, pruneBefore)
-		if err != nil {
-			totalErr += 1
-			log.WithError(err).WithField("directory", dir).Error("Unable to prune directory")
-		}
-		totalPruned += pruned
-	}
-
-	if totalErr > 0 {
-		return errors.Wrapf(errPruningFailures, "pruning failed for %d root directories", totalErr)
-	}
-	return nil
-}
-
-func shouldRetain(slot, pruneBefore primitives.Slot) bool {
-	return slot >= pruneBefore
-}
-
-func (p *blobPruner) tryPruneDir(dir string, pruneBefore primitives.Slot) (int, error) {
-	root := rootFromDir(dir)
-	slot, slotCached := p.slotMap.slot(root)
-	// Return early if the slot is cached and doesn't need pruning.
-	if slotCached && shouldRetain(slot, pruneBefore) {
-		return 0, nil
-	}
-
-	// entries will include things that aren't ssz files, like dangling .part files. We need these to
-	// completely clean up the directory.
-	entries, err := listDir(p.fs, dir)
-	if err != nil {
-		return 0, errors.Wrapf(err, "failed to list blobs in directory %s", dir)
-	}
-	// scFiles filters the dir listing down to the ssz encoded BlobSidecar files. This allows us to peek
-	// at the first one in the list to figure out the slot.
-	scFiles := filter(entries, filterSsz)
-	if len(scFiles) == 0 {
-		log.WithField("dir", dir).Warn("Pruner ignoring directory with no blob files")
-		return 0, nil
-	}
-	if !slotCached {
-		slot, err = slotFromFile(path.Join(dir, scFiles[0]), p.fs)
-		if err != nil {
-			return 0, errors.Wrapf(err, "slot could not be read from blob file %s", scFiles[0])
-		}
-		for i := range scFiles {
-			idx, err := idxFromPath(scFiles[i])
-			if err != nil {
-				return 0, errors.Wrapf(err, "index could not be determined for blob file %s", scFiles[i])
-			}
-			if err := p.slotMap.ensure(root, slot, idx); err != nil {
-				return 0, errors.Wrapf(err, "could not update prune cache for blob file %s", scFiles[i])
-			}
-		}
-		if shouldRetain(slot, pruneBefore) {
-			return 0, nil
-		}
-	}
-
-	removed := 0
-	for _, fname := range entries {
-		fullName := path.Join(dir, fname)
-		if err := p.fs.Remove(fullName); err != nil {
-			return removed, errors.Wrapf(err, "unable to remove %s", fullName)
-		}
-		// Don't count other files that happen to be in the dir, like dangling .part files.
-		if filterSsz(fname) {
-			removed += 1
-		}
-		// Log a warning whenever we clean up a .part file
-		if filterPart(fullName) {
-			log.WithField("file", fullName).Warn("Deleting abandoned blob .part file")
-		}
-	}
-	if err := p.fs.Remove(dir); err != nil {
-		return removed, errors.Wrapf(err, "unable to remove blob directory %s", dir)
-	}
-
-	p.slotMap.evict(rootFromDir(dir))
-	return len(scFiles), nil
-}
-
-func idxFromPath(fname string) (uint64, error) {
-	fname = path.Base(fname)
-
-	if filepath.Ext(fname) != dotSszExt {
-		return 0, errors.Wrap(errNotBlobSSZ, "does not have .ssz extension")
-	}
-	parts := strings.Split(fname, ".")
-	if len(parts) != 2 {
-		return 0, errors.Wrap(errNotBlobSSZ, "unexpected filename structure (want <index>.ssz)")
-	}
-	return strconv.ParseUint(parts[0], 10, 64)
-}
-
-func rootFromDir(dir string) string {
-	return filepath.Base(dir) // end of the path should be the blob directory, named by hex encoding of root
-}
-
-// Read slot from marshaled BlobSidecar data in the given file. See slotFromBlob for details.
-func slotFromFile(file string, fs afero.Fs) (primitives.Slot, error) {
-	f, err := fs.Open(file)
-	if err != nil {
-		return 0, err
-	}
-	defer func() {
-		if err := f.Close(); err != nil {
-			log.WithError(err).Errorf("Could not close blob file")
-		}
-	}()
-	return slotFromBlob(f)
-}
-
-// slotFromBlob reads the ssz data of a file at the specified offset (8 + 131072 + 48 + 48 = 131176 bytes),
-// which is calculated based on the size of the BlobSidecar struct and is based on the size of the fields
-// preceding the slot information within SignedBeaconBlockHeader.
-func slotFromBlob(at io.ReaderAt) (primitives.Slot, error) {
-	b := make([]byte, 8)
-	_, err := at.ReadAt(b, 131176)
-	if err != nil {
-		return 0, err
-	}
-	rawSlot := binary.LittleEndian.Uint64(b)
-	return primitives.Slot(rawSlot), nil
-}
-
-func listDir(fs afero.Fs, dir string) ([]string, error) {
-	top, err := fs.Open(dir)
-	if err != nil {
-		return nil, errors.Wrap(err, "failed to open directory descriptor")
-	}
-	defer func() {
-		if err := top.Close(); err != nil {
-			log.WithError(err).Errorf("Could not close file %s", dir)
-		}
-	}()
-	// re the -1 param: "If n <= 0, Readdirnames returns all the names from the directory in a single slice"
-	dirs, err := top.Readdirnames(-1)
-	if err != nil {
-		return nil, errors.Wrap(err, "failed to read directory listing")
-	}
-	return dirs, nil
-}
-
-func filter(entries []string, filt func(string) bool) []string {
-	filtered := make([]string, 0, len(entries))
-	for i := range entries {
-		if filt(entries[i]) {
-			filtered = append(filtered, entries[i])
-		}
-	}
-	return filtered
-}
-
-func filterRoot(s string) bool {
-	return strings.HasPrefix(s, "0x")
-}
-
-var dotSszExt = "." + sszExt
-var dotPartExt = "." + partExt
-
-func filterSsz(s string) bool {
-	return filepath.Ext(s) == dotSszExt
-}
-
-func filterPart(s string) bool {
-	return filepath.Ext(s) == dotPartExt
-}
-
-func newSlotForRoot() *slotForRoot {
-	return &slotForRoot{
-		cache: make(map[string]*slotCacheEntry, params.BeaconConfig().MinEpochsForBlobsSidecarsRequest*fieldparams.SlotsPerEpoch),
-	}
-}
-
-type slotCacheEntry struct {
-	slot primitives.Slot
-	mask [fieldparams.MaxBlobsPerBlock]bool
-}
-
-type slotForRoot struct {
-	sync.RWMutex
-	nBlobs float64
-	cache  map[string]*slotCacheEntry
-}
-
-func (s *slotForRoot) updateMetrics(delta float64) {
-	s.nBlobs += delta
-	blobDiskCount.Set(s.nBlobs)
-	blobDiskSize.Set(s.nBlobs * bytesPerSidecar)
-}
-
-func (s *slotForRoot) ensure(key string, slot primitives.Slot, idx uint64) error {
-	if idx >= fieldparams.MaxBlobsPerBlock {
-		return errIndexOutOfBounds
-	}
-	s.Lock()
-	defer s.Unlock()
-	v, ok := s.cache[key]
-	if !ok {
-		v = &slotCacheEntry{}
-	}
-	v.slot = slot
-	if !v.mask[idx] {
-		s.updateMetrics(1)
-	}
-	v.mask[idx] = true
-	s.cache[key] = v
-	return nil
-}
-
-func (s *slotForRoot) slot(key string) (primitives.Slot, bool) {
-	s.RLock()
-	defer s.RUnlock()
-	v, ok := s.cache[key]
-	if !ok {
-		return 0, false
-	}
-	return v.slot, ok
-}
-
-func (s *slotForRoot) evict(key string) {
-	s.Lock()
-	defer s.Unlock()
-	v, ok := s.cache[key]
-	var deleted float64
-	if ok {
-		for i := range v.mask {
-			if v.mask[i] {
-				deleted += 1
-			}
-		}
-		s.updateMetrics(-deleted)
-	}
-	delete(s.cache, key)
+	return latest - period
 }

beacon-chain/db/filesystem/pruner_test.go

@@ -1,327 +1,196 @@
 package filesystem
 
 import (
-	"bytes"
-	"fmt"
-	"math"
+	"encoding/binary"
 	"os"
-	"path"
-	"sort"
 	"testing"
 
 	"github.com/prysmaticlabs/prysm/v5/beacon-chain/verification"
-	fieldparams "github.com/prysmaticlabs/prysm/v5/config/fieldparams"
+	"github.com/prysmaticlabs/prysm/v5/config/params"
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
 	"github.com/prysmaticlabs/prysm/v5/testing/require"
 	"github.com/prysmaticlabs/prysm/v5/testing/util"
+	"github.com/prysmaticlabs/prysm/v5/time/slots"
 	"github.com/spf13/afero"
 )
 
-func TestTryPruneDir_CachedNotExpired(t *testing.T) {
-	fs := afero.NewMemMapFs()
-	pr, err := newBlobPruner(fs, 0)
-	require.NoError(t, err)
-	slot := pr.windowSize
-	_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, slot, fieldparams.MaxBlobsPerBlock)
-	sc, err := verification.BlobSidecarNoop(sidecars[0])
-	require.NoError(t, err)
-	root := fmt.Sprintf("%#x", sc.BlockRoot())
-	// This slot is right on the edge of what would need to be pruned, so by adding it to the cache and
-	// skipping any other test setup, we can be certain the hot cache path never touches the filesystem.
-	require.NoError(t, pr.slotMap.ensure(root, sc.Slot(), 0))
-	pruned, err := pr.tryPruneDir(root, pr.windowSize)
-	require.NoError(t, err)
-	require.Equal(t, 0, pruned)
+type prunerScenario struct {
+	name            string
+	prunedBefore    primitives.Epoch
+	retentionPeriod primitives.Epoch
+	latest          primitives.Epoch
+	expected        pruneExpectation
 }
 
-func TestTryPruneDir_CachedExpired(t *testing.T) {
-	t.Run("empty directory", func(t *testing.T) {
-		fs := afero.NewMemMapFs()
-		pr, err := newBlobPruner(fs, 0)
-		require.NoError(t, err)
-		var slot primitives.Slot = 0
-		_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, slot, 1)
-		sc, err := verification.BlobSidecarNoop(sidecars[0])
-		require.NoError(t, err)
-		root := fmt.Sprintf("%#x", sc.BlockRoot())
-		require.NoError(t, fs.Mkdir(root, directoryPermissions)) // make empty directory
-		require.NoError(t, pr.slotMap.ensure(root, sc.Slot(), 0))
-		pruned, err := pr.tryPruneDir(root, slot+1)
-		require.NoError(t, err)
-		require.Equal(t, 0, pruned)
-	})
-	t.Run("blobs to delete", func(t *testing.T) {
-		fs, bs, err := NewEphemeralBlobStorageWithFs(t)
-		require.NoError(t, err)
-		var slot primitives.Slot = 0
-		_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, slot, 2)
-		scs, err := verification.BlobSidecarSliceNoop(sidecars)
-		require.NoError(t, err)
-
-		require.NoError(t, bs.Save(scs[0]))
-		require.NoError(t, bs.Save(scs[1]))
-
-		// check that the root->slot is cached
-		root := fmt.Sprintf("%#x", scs[0].BlockRoot())
-		cs, cok := bs.pruner.slotMap.slot(root)
-		require.Equal(t, true, cok)
-		require.Equal(t, slot, cs)
-
-		// ensure that we see the saved files in the filesystem
-		files, err := listDir(fs, root)
-		require.NoError(t, err)
-		require.Equal(t, 2, len(files))
-
-		pruned, err := bs.pruner.tryPruneDir(root, slot+1)
-		require.NoError(t, err)
-		require.Equal(t, 2, pruned)
-		files, err = listDir(fs, root)
-		require.ErrorIs(t, err, os.ErrNotExist)
-		require.Equal(t, 0, len(files))
-	})
+type pruneExpectation struct {
+	called  bool
+	arg     primitives.Epoch
+	summary *pruneSummary
+	err     error
 }
 
-func TestTryPruneDir_SlotFromFile(t *testing.T) {
-	t.Run("expired blobs deleted", func(t *testing.T) {
-		fs, bs, err := NewEphemeralBlobStorageWithFs(t)
-		require.NoError(t, err)
-		var slot primitives.Slot = 0
-		_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, slot, 2)
-		scs, err := verification.BlobSidecarSliceNoop(sidecars)
-		require.NoError(t, err)
-
-		require.NoError(t, bs.Save(scs[0]))
-		require.NoError(t, bs.Save(scs[1]))
-
-		// check that the root->slot is cached
-		root := fmt.Sprintf("%#x", scs[0].BlockRoot())
-		cs, ok := bs.pruner.slotMap.slot(root)
-		require.Equal(t, true, ok)
-		require.Equal(t, slot, cs)
-		// evict it from the cache so that we trigger the file read path
-		bs.pruner.slotMap.evict(root)
-		_, ok = bs.pruner.slotMap.slot(root)
-		require.Equal(t, false, ok)
-
-		// ensure that we see the saved files in the filesystem
-		files, err := listDir(fs, root)
-		require.NoError(t, err)
-		require.Equal(t, 2, len(files))
-
-		pruned, err := bs.pruner.tryPruneDir(root, slot+1)
-		require.NoError(t, err)
-		require.Equal(t, 2, pruned)
-		files, err = listDir(fs, root)
-		require.ErrorIs(t, err, os.ErrNotExist)
-		require.Equal(t, 0, len(files))
-	})
-	t.Run("not expired, intact", func(t *testing.T) {
-		fs, bs, err := NewEphemeralBlobStorageWithFs(t)
-		require.NoError(t, err)
-		// Set slot equal to the window size, so it should be retained.
-		var slot primitives.Slot = bs.pruner.windowSize
-		_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, slot, 2)
-		scs, err := verification.BlobSidecarSliceNoop(sidecars)
-		require.NoError(t, err)
-
-		require.NoError(t, bs.Save(scs[0]))
-		require.NoError(t, bs.Save(scs[1]))
-
-		// Evict slot mapping from the cache so that we trigger the file read path.
-		root := fmt.Sprintf("%#x", scs[0].BlockRoot())
-		bs.pruner.slotMap.evict(root)
-		_, ok := bs.pruner.slotMap.slot(root)
-		require.Equal(t, false, ok)
-
-		// Ensure that we see the saved files in the filesystem.
-		files, err := listDir(fs, root)
-		require.NoError(t, err)
-		require.Equal(t, 2, len(files))
-
-		// This should use the slotFromFile code (simulating restart).
-		// Setting pruneBefore == slot, so that the slot will be outside the window (at the boundary).
-		pruned, err := bs.pruner.tryPruneDir(root, slot)
-		require.NoError(t, err)
-		require.Equal(t, 0, pruned)
-
-		// Ensure files are still present.
-		files, err = listDir(fs, root)
-		require.NoError(t, err)
-		require.Equal(t, 2, len(files))
-	})
+func (e *pruneExpectation) record(before primitives.Epoch) (*pruneSummary, error) {
+	e.called = true
+	e.arg = before
+	if e.summary == nil {
+		e.summary = &pruneSummary{}
+	}
+	return e.summary, e.err
 }
 
-func TestSlotFromBlob(t *testing.T) {
-	cases := []struct {
-		slot primitives.Slot
-	}{
-		{slot: 0},
-		{slot: 2},
-		{slot: 1123581321},
-		{slot: math.MaxUint64},
-	}
-	for _, c := range cases {
-		t.Run(fmt.Sprintf("slot %d", c.slot), func(t *testing.T) {
-			_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, c.slot, 1)
-			sc := sidecars[0]
-			enc, err := sc.MarshalSSZ()
-			require.NoError(t, err)
-			slot, err := slotFromBlob(bytes.NewReader(enc))
-			require.NoError(t, err)
-			require.Equal(t, c.slot, slot)
-		})
-	}
-}
-
-func TestSlotFromFile(t *testing.T) {
-	cases := []struct {
-		slot primitives.Slot
-	}{
-		{slot: 0},
-		{slot: 2},
-		{slot: 1123581321},
-		{slot: math.MaxUint64},
-	}
-	for _, c := range cases {
-		t.Run(fmt.Sprintf("slot %d", c.slot), func(t *testing.T) {
-			fs, bs, err := NewEphemeralBlobStorageWithFs(t)
-			require.NoError(t, err)
-			_, sidecars := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, c.slot, 1)
-			sc, err := verification.BlobSidecarNoop(sidecars[0])
-			require.NoError(t, err)
-			require.NoError(t, bs.Save(sc))
-			fname := namerForSidecar(sc)
-			sszPath := fname.path()
-			slot, err := slotFromFile(sszPath, fs)
-			require.NoError(t, err)
-			require.Equal(t, c.slot, slot)
-		})
-	}
-}
-
-type dirFiles struct {
-	name     string
-	isDir    bool
-	children []dirFiles
-}
-
-func (df dirFiles) reify(t *testing.T, fs afero.Fs, base string) {
-	fullPath := path.Join(base, df.name)
-	if df.isDir {
-		if df.name != "" {
-			require.NoError(t, fs.Mkdir(fullPath, directoryPermissions))
-		}
-		for _, c := range df.children {
-			c.reify(t, fs, fullPath)
-		}
-	} else {
-		fp, err := fs.Create(fullPath)
-		require.NoError(t, err)
-		_, err = fp.WriteString("derp")
-		require.NoError(t, err)
-	}
-}
-
-func (df dirFiles) childNames() []string {
-	cn := make([]string, len(df.children))
-	for i := range df.children {
-		cn[i] = df.children[i].name
-	}
-	return cn
-}
-
-func TestListDir(t *testing.T) {
-	fs := afero.NewMemMapFs()
-
-	// parent directory
-	fsLayout := dirFiles{isDir: true}
-	// break out each subdir for easier assertions
-	notABlob := dirFiles{name: "notABlob", isDir: true}
-	childlessBlob := dirFiles{name: "0x0987654321", isDir: true}
-	blobWithSsz := dirFiles{name: "0x1123581321", isDir: true,
-		children: []dirFiles{{name: "1.ssz"}, {name: "2.ssz"}},
-	}
-	blobWithSszAndTmp := dirFiles{name: "0x1234567890", isDir: true,
-		children: []dirFiles{{name: "5.ssz"}, {name: "0.part"}}}
-	fsLayout.children = append(fsLayout.children, notABlob)
-	fsLayout.children = append(fsLayout.children, childlessBlob)
-	fsLayout.children = append(fsLayout.children, blobWithSsz)
-	fsLayout.children = append(fsLayout.children, blobWithSszAndTmp)
-
-	topChildren := make([]string, len(fsLayout.children))
-	for i := range fsLayout.children {
-		topChildren[i] = fsLayout.children[i].name
-	}
-
-	fsLayout.reify(t, fs, "")
-	cases := []struct {
-		name     string
-		dirPath  string
-		expected []string
-		filter   func(string) bool
-		err      error
-	}{
+func TestPrunerNotify(t *testing.T) {
+	defaultRetention := params.BeaconConfig().MinEpochsForBlobsSidecarsRequest
+	cases := []prunerScenario{
 		{
-			name:     "non-existent",
-			dirPath:  "derp",
-			expected: []string{},
-			err:      os.ErrNotExist,
+			name:            "last epoch of period",
+			retentionPeriod: defaultRetention,
+			prunedBefore:    11235,
+			latest:          defaultRetention + 11235,
+			expected:        pruneExpectation{called: false},
 		},
 		{
-			name:     "empty",
-			dirPath:  childlessBlob.name,
-			expected: []string{},
+			name:            "within period",
+			retentionPeriod: defaultRetention,
+			prunedBefore:    11235,
+			latest:          11235 + defaultRetention - 1,
+			expected:        pruneExpectation{called: false},
 		},
 		{
-			name:     "top",
-			dirPath:  ".",
-			expected: topChildren,
+			name:            "triggers",
+			retentionPeriod: defaultRetention,
+			prunedBefore:    11235,
+			latest:          11235 + 1 + defaultRetention,
+			expected:        pruneExpectation{called: true, arg: 11235 + 1},
 		},
 		{
-			name:     "custom filter: only notABlob",
-			dirPath:  ".",
-			expected: []string{notABlob.name},
-			filter: func(s string) bool {
-				if s == notABlob.name {
-					return true
-				}
-				return false
-			},
+			name:            "from zero - before first period",
+			retentionPeriod: defaultRetention,
+			prunedBefore:    0,
+			latest:          defaultRetention - 1,
+			expected:        pruneExpectation{called: false},
 		},
 		{
-			name:     "root filter",
-			dirPath:  ".",
-			expected: []string{childlessBlob.name, blobWithSsz.name, blobWithSszAndTmp.name},
-			filter:   filterRoot,
+			name:            "from zero - at boundary",
+			retentionPeriod: defaultRetention,
+			prunedBefore:    0,
+			latest:          defaultRetention,
+			expected:        pruneExpectation{called: false},
 		},
 		{
-			name:     "ssz filter",
-			dirPath:  blobWithSsz.name,
-			expected: blobWithSsz.childNames(),
-			filter:   filterSsz,
-		},
-		{
-			name:     "ssz mixed filter",
-			dirPath:  blobWithSszAndTmp.name,
-			expected: []string{"5.ssz"},
-			filter:   filterSsz,
+			name:            "from zero - triggers",
+			retentionPeriod: defaultRetention,
+			prunedBefore:    0,
+			latest:          defaultRetention + 1,
+			expected:        pruneExpectation{called: true, arg: 1},
 		},
 	}
 	for _, c := range cases {
 		t.Run(c.name, func(t *testing.T) {
-			result, err := listDir(fs, c.dirPath)
-			if c.filter != nil {
-				result = filter(result, c.filter)
-			}
-			if c.err != nil {
-				require.ErrorIs(t, err, c.err)
-				require.Equal(t, 0, len(result))
-			} else {
-				require.NoError(t, err)
-				sort.Strings(c.expected)
-				sort.Strings(result)
-				require.DeepEqual(t, c.expected, result)
-			}
+			actual := &pruneExpectation{}
+			l := &mockLayout{pruneBeforeFunc: actual.record}
+			pruner := &blobPruner{retentionPeriod: c.retentionPeriod}
+			pruner.prunedBefore.Store(uint64(c.prunedBefore))
+			done := pruner.notify(c.latest, l)
+			<-done
+			require.Equal(t, c.expected.called, actual.called)
+			require.Equal(t, c.expected.arg, actual.arg)
+		})
+	}
+}
+
+func testSetupBlobIdentPaths(t *testing.T, fs afero.Fs, bs *BlobStorage, idents []testIdent) []blobIdent {
+	created := make([]blobIdent, len(idents))
+	for i, id := range idents {
+		slot, err := slots.EpochStart(id.epoch)
+		require.NoError(t, err)
+		slot += id.offset
+		_, scs := util.GenerateTestDenebBlockWithSidecar(t, [32]byte{}, slot, 1)
+		sc := verification.FakeVerifyForTest(t, scs[0])
+		require.NoError(t, bs.Save(sc))
+		ident := identForSidecar(sc)
+		_, err = fs.Stat(bs.layout.sszPath(ident))
+		require.NoError(t, err)
+		created[i] = ident
+	}
+	return created
+}
+
+func testAssertBlobsPruned(t *testing.T, fs afero.Fs, bs *BlobStorage, pruned, remain []blobIdent) {
+	for _, id := range pruned {
+		_, err := fs.Stat(bs.layout.sszPath(id))
+		require.Equal(t, true, os.IsNotExist(err))
+	}
+	for _, id := range remain {
+		_, err := fs.Stat(bs.layout.sszPath(id))
+		require.NoError(t, err)
+	}
+}
+
+type testIdent struct {
+	blobIdent
+	offset primitives.Slot
+}
+
+func testRoots(n int) [][32]byte {
+	roots := make([][32]byte, n)
+	for i := range roots {
+		binary.LittleEndian.PutUint32(roots[i][:], uint32(1+i))
+	}
+	return roots
+}
+
+func TestLayoutPruneBefore(t *testing.T) {
+	roots := testRoots(10)
+	cases := []struct {
+		name        string
+		pruned      []testIdent
+		remain      []testIdent
+		pruneBefore primitives.Epoch
+		err         error
+		sum         pruneSummary
+	}{
+		{
+			name:        "none pruned",
+			pruneBefore: 1,
+			pruned:      []testIdent{},
+			remain: []testIdent{
+				{offset: 1, blobIdent: blobIdent{root: roots[0], epoch: 1, index: 0}},
+				{offset: 1, blobIdent: blobIdent{root: roots[1], epoch: 1, index: 0}},
+			},
+		},
+		{
+			name:        "expected pruned before epoch",
+			pruneBefore: 3,
+			pruned: []testIdent{
+				{offset: 0, blobIdent: blobIdent{root: roots[0], epoch: 1, index: 0}},
+				{offset: 31, blobIdent: blobIdent{root: roots[1], epoch: 1, index: 5}},
+				{offset: 0, blobIdent: blobIdent{root: roots[2], epoch: 2, index: 0}},
+				{offset: 31, blobIdent: blobIdent{root: roots[3], epoch: 2, index: 3}},
+			},
+			remain: []testIdent{
+				{offset: 0, blobIdent: blobIdent{root: roots[4], epoch: 3, index: 2}},  // boundary
+				{offset: 31, blobIdent: blobIdent{root: roots[5], epoch: 3, index: 0}}, // boundary
+				{offset: 0, blobIdent: blobIdent{root: roots[6], epoch: 4, index: 1}},
+				{offset: 31, blobIdent: blobIdent{root: roots[7], epoch: 4, index: 5}},
+			},
+			sum: pruneSummary{blobsPruned: 4},
+		},
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			fs, bs := NewEphemeralBlobStorageAndFs(t)
+			pruned := testSetupBlobIdentPaths(t, fs, bs, c.pruned)
+			remain := testSetupBlobIdentPaths(t, fs, bs, c.remain)
+			sum, err := bs.layout.pruneBefore(c.pruneBefore)
+			if c.err != nil {
+				require.ErrorIs(t, err, c.err)
+				return
+			}
+			require.NoError(t, err)
+			testAssertBlobsPruned(t, fs, bs, pruned, remain)
+			require.Equal(t, c.sum.blobsPruned, sum.blobsPruned)
+			require.Equal(t, len(c.pruned), sum.blobsPruned)
+			require.Equal(t, len(c.sum.failedRemovals), len(sum.failedRemovals))
 		})
 	}
 }

beacon-chain/db/iface/interface.go

@@ -118,7 +118,7 @@ type HeadAccessDatabase interface {
 // SlasherDatabase interface for persisting data related to detecting slashable offenses on Ethereum.
 type SlasherDatabase interface {
 	io.Closer
-	SaveLastEpochsWrittenForValidators(
+	SaveLastEpochWrittenForValidators(
 		ctx context.Context, epochByValidator map[primitives.ValidatorIndex]primitives.Epoch,
 	) error
 	SaveAttestationRecordsForValidators(

beacon-chain/db/kv/BUILD.bazel

@@ -20,6 +20,7 @@ go_library(
         "migration.go",
         "migration_archived_index.go",
         "migration_block_slot_index.go",
+        "migration_finalized_parent.go",
         "migration_state_validators.go",
         "schema.go",
         "state.go",

beacon-chain/db/kv/finalized_block_roots.go

@@ -201,21 +201,20 @@ func (s *Store) BackfillFinalizedIndex(ctx context.Context, blocks []blocks.ROBl
 			return err
 		}
 		encs[i-1] = penc
-
-		// The final element is the parent of finalizedChildRoot. This is checked inside the db transaction using
-		// the parent_root value stored in the index data for finalizedChildRoot.
-		if i == len(blocks)-1 {
-			fbrs[i].ChildRoot = finalizedChildRoot[:]
-			// Final element is complete, so it is pre-encoded like the others.
-			enc, err := encode(ctx, fbrs[i])
-			if err != nil {
-				tracing.AnnotateError(span, err)
-				return err
-			}
-			encs[i] = enc
-		}
 	}
 
+	// The final element is the parent of finalizedChildRoot. This is checked inside the db transaction using
+	// the parent_root value stored in the index data for finalizedChildRoot.
+	lastIdx := len(blocks) - 1
+	fbrs[lastIdx].ChildRoot = finalizedChildRoot[:]
+	// Final element is complete, so it is pre-encoded like the others.
+	enc, err := encode(ctx, fbrs[lastIdx])
+	if err != nil {
+		tracing.AnnotateError(span, err)
+		return err
+	}
+	encs[lastIdx] = enc
+
 	return s.db.Update(func(tx *bolt.Tx) error {
 		bkt := tx.Bucket(finalizedBlockRootsIndexBucket)
 		child := bkt.Get(finalizedChildRoot[:])

beacon-chain/db/kv/finalized_block_roots_test.go

@@ -237,6 +237,50 @@ func makeBlocksAltair(t *testing.T, startIdx, num uint64, previousRoot [32]byte)
 	return ifaceBlocks
 }
 
+func TestStore_BackfillFinalizedIndexSingle(t *testing.T) {
+	db := setupDB(t)
+	ctx := context.Background()
+	// we're making 4 blocks so we can test an element without a valid child at the end
+	blks, err := consensusblocks.NewROBlockSlice(makeBlocks(t, 0, 4, [32]byte{}))
+	require.NoError(t, err)
+
+	// existing is the child that we'll set up in the index by hand to seed the index.
+	existing := blks[3]
+
+	// toUpdate is a single item update, emulating a backfill batch size of 1. it is the parent of `existing`.
+	toUpdate := blks[2]
+
+	// set up existing finalized block
+	ebpr := existing.Block().ParentRoot()
+	ebr := existing.Root()
+	ebf := &ethpb.FinalizedBlockRootContainer{
+		ParentRoot: ebpr[:],
+		ChildRoot:  make([]byte, 32), // we're bypassing validation to seed the db, so we don't need a valid child.
+	}
+	enc, err := encode(ctx, ebf)
+	require.NoError(t, err)
+	// writing this to the index outside of the validating function to seed the test.
+	err = db.db.Update(func(tx *bolt.Tx) error {
+		bkt := tx.Bucket(finalizedBlockRootsIndexBucket)
+		return bkt.Put(ebr[:], enc)
+	})
+	require.NoError(t, err)
+
+	require.NoError(t, db.BackfillFinalizedIndex(ctx, []consensusblocks.ROBlock{toUpdate}, ebr))
+
+	// make sure that we still correctly validate descendents in the single item case.
+	noChild := blks[0] // will fail to update because we don't have blks[1] in the db.
+	// test wrong child param
+	require.ErrorIs(t, db.BackfillFinalizedIndex(ctx, []consensusblocks.ROBlock{noChild}, ebr), errNotConnectedToFinalized)
+	// test parent of child that isn't finalized
+	require.ErrorIs(t, db.BackfillFinalizedIndex(ctx, []consensusblocks.ROBlock{noChild}, blks[1].Root()), errFinalizedChildNotFound)
+
+	// now make it work by writing the missing block
+	require.NoError(t, db.BackfillFinalizedIndex(ctx, []consensusblocks.ROBlock{blks[1]}, blks[2].Root()))
+	// since blks[1] is now in the index, we should be able to update blks[0]
+	require.NoError(t, db.BackfillFinalizedIndex(ctx, []consensusblocks.ROBlock{blks[0]}, blks[1].Root()))
+}
+
 func TestStore_BackfillFinalizedIndex(t *testing.T) {
 	db := setupDB(t)
 	ctx := context.Background()
@@ -252,23 +296,23 @@ func TestStore_BackfillFinalizedIndex(t *testing.T) {
 		ParentRoot: ebpr[:],
 		ChildRoot:  chldr[:],
 	}
-	disjoint := []consensusblocks.ROBlock{
-		blks[0],
-		blks[2],
-	}
 	enc, err := encode(ctx, ebf)
 	require.NoError(t, err)
 	err = db.db.Update(func(tx *bolt.Tx) error {
 		bkt := tx.Bucket(finalizedBlockRootsIndexBucket)
 		return bkt.Put(ebr[:], enc)
 	})
+	require.NoError(t, err)
 
 	// reslice to remove the existing blocks
 	blks = blks[0:64]
 	// check the other error conditions with a descendent root that really doesn't exist
-	require.NoError(t, err)
+
+	disjoint := []consensusblocks.ROBlock{
+		blks[0],
+		blks[2],
+	}
 	require.ErrorIs(t, db.BackfillFinalizedIndex(ctx, disjoint, [32]byte{}), errIncorrectBlockParent)
-	require.NoError(t, err)
 	require.ErrorIs(t, errFinalizedChildNotFound, db.BackfillFinalizedIndex(ctx, blks, [32]byte{}))
 
 	// use the real root so that it succeeds

beacon-chain/db/kv/kv.go

@@ -100,37 +100,24 @@ func StoreDatafilePath(dirPath string) string {
 }
 
 var Buckets = [][]byte{
-	attestationsBucket,
 	blocksBucket,
 	stateBucket,
-	proposerSlashingsBucket,
-	attesterSlashingsBucket,
-	voluntaryExitsBucket,
 	chainMetadataBucket,
 	checkpointBucket,
 	powchainBucket,
 	stateSummaryBucket,
 	stateValidatorsBucket,
 	// Indices buckets.
-	attestationHeadBlockRootBucket,
-	attestationSourceRootIndicesBucket,
-	attestationSourceEpochIndicesBucket,
-	attestationTargetRootIndicesBucket,
-	attestationTargetEpochIndicesBucket,
 	blockSlotIndicesBucket,
 	stateSlotIndicesBucket,
 	blockParentRootIndicesBucket,
 	finalizedBlockRootsIndexBucket,
 	blockRootValidatorHashesBucket,
-	// State management service bucket.
-	newStateServiceCompatibleBucket,
 	// Migrations
 	migrationsBucket,
 
 	feeRecipientBucket,
 	registrationBucket,
-
-	blobsBucket,
 }
 
 // KVStoreOption is a functional option that modifies a kv.Store.
@@ -150,7 +137,7 @@ func NewKVStore(ctx context.Context, dirPath string, opts ...KVStoreOption) (*St
 		}
 	}
 	datafile := StoreDatafilePath(dirPath)
-	log.Infof("Opening Bolt DB at %s", datafile)
+	log.WithField("path", datafile).Info("Opening Bolt DB")
 	boltDB, err := bolt.Open(
 		datafile,
 		params.BeaconIoConfig().ReadWritePermissions,

beacon-chain/db/kv/migration.go

@@ -14,6 +14,7 @@ var migrations = []migration{
 	migrateArchivedIndex,
 	migrateBlockSlotIndex,
 	migrateStateValidators,
+	migrateFinalizedParent,
 }
 
 // RunMigrations defined in the migrations array.

beacon-chain/db/kv/migration_finalized_parent.go

@@ -0,0 +1,87 @@
+package kv
+
+import (
+	"bytes"
+	"context"
+	"fmt"
+
+	"github.com/pkg/errors"
+	"github.com/prysmaticlabs/prysm/v5/config/params"
+	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
+	ethpb "github.com/prysmaticlabs/prysm/v5/proto/prysm/v1alpha1"
+	bolt "go.etcd.io/bbolt"
+)
+
+var migrationFinalizedParent = []byte("parent_bug_32fb183")
+
+func migrateFinalizedParent(ctx context.Context, db *bolt.DB) error {
+	if updateErr := db.Update(func(tx *bolt.Tx) error {
+		mb := tx.Bucket(migrationsBucket)
+		if b := mb.Get(migrationFinalizedParent); bytes.Equal(b, migrationCompleted) {
+			return nil // Migration already completed.
+		}
+
+		bkt := tx.Bucket(finalizedBlockRootsIndexBucket)
+		if bkt == nil {
+			return fmt.Errorf("unable to read %s bucket for migration", finalizedBlockRootsIndexBucket)
+		}
+		bb := tx.Bucket(blocksBucket)
+		if bb == nil {
+			return fmt.Errorf("unable to read %s bucket for migration", blocksBucket)
+		}
+
+		c := bkt.Cursor()
+		var slotsWithoutBug primitives.Slot
+		maxBugSearch := params.BeaconConfig().SlotsPerEpoch * 10
+		for k, v := c.Last(); k != nil; k, v = c.Prev() {
+			// check if context is cancelled in between
+			if ctx.Err() != nil {
+				return ctx.Err()
+			}
+
+			idxEntry := &ethpb.FinalizedBlockRootContainer{}
+			if err := decode(ctx, v, idxEntry); err != nil {
+				return errors.Wrapf(err, "unable to decode finalized block root container for root=%#x", k)
+			}
+			// Not one of the corrupt values
+			if !bytes.Equal(idxEntry.ParentRoot, k) {
+				slotsWithoutBug += 1
+				if slotsWithoutBug > maxBugSearch {
+					break
+				}
+				continue
+			}
+			slotsWithoutBug = 0
+			log.WithField("root", fmt.Sprintf("%#x", k)).Debug("found index entry with incorrect parent root")
+
+			// Look up full block to get the correct parent root.
+			encBlk := bb.Get(k)
+			if encBlk == nil {
+				return errors.Wrapf(ErrNotFound, "could not find block for corrupt finalized index entry %#x", k)
+			}
+			blk, err := unmarshalBlock(ctx, encBlk)
+			if err != nil {
+				return errors.Wrapf(err, "unable to decode block for root=%#x", k)
+			}
+			// Replace parent root in the index with the correct value and write it back.
+			pr := blk.Block().ParentRoot()
+			idxEntry.ParentRoot = pr[:]
+			idxEnc, err := encode(ctx, idxEntry)
+			if err != nil {
+				return errors.Wrapf(err, "failed to encode finalized index entry for root=%#x", k)
+			}
+			if err := bkt.Put(k, idxEnc); err != nil {
+				return errors.Wrapf(err, "failed to update finalized index entry for root=%#x", k)
+			}
+			log.WithField("root", fmt.Sprintf("%#x", k)).
+				WithField("parentRoot", fmt.Sprintf("%#x", idxEntry.ParentRoot)).
+				Debug("updated corrupt index entry with correct parent")
+		}
+		// Mark migration complete.
+		return mb.Put(migrationFinalizedParent, migrationCompleted)
+	}); updateErr != nil {
+		log.WithError(updateErr).Errorf("could not run finalized parent root index repair migration")
+		return updateErr
+	}
+	return nil
+}

beacon-chain/db/kv/schema.go

@@ -7,20 +7,15 @@ package kv
 // it easy to scan for keys that have a certain shard number as a prefix and return those
 // corresponding attestations.
 var (
-	attestationsBucket      = []byte("attestations")
-	blobsBucket             = []byte("blobs")
-	blocksBucket            = []byte("blocks")
-	stateBucket             = []byte("state")
-	stateSummaryBucket      = []byte("state-summary")
-	proposerSlashingsBucket = []byte("proposer-slashings")
-	attesterSlashingsBucket = []byte("attester-slashings")
-	voluntaryExitsBucket    = []byte("voluntary-exits")
-	chainMetadataBucket     = []byte("chain-metadata")
-	checkpointBucket        = []byte("check-point")
-	powchainBucket          = []byte("powchain")
-	stateValidatorsBucket   = []byte("state-validators")
-	feeRecipientBucket      = []byte("fee-recipient")
-	registrationBucket      = []byte("registration")
+	blocksBucket          = []byte("blocks")
+	stateBucket           = []byte("state")
+	stateSummaryBucket    = []byte("state-summary")
+	chainMetadataBucket   = []byte("chain-metadata")
+	checkpointBucket      = []byte("check-point")
+	powchainBucket        = []byte("powchain")
+	stateValidatorsBucket = []byte("state-validators")
+	feeRecipientBucket    = []byte("fee-recipient")
+	registrationBucket    = []byte("registration")
 
 	// Deprecated: This bucket was migrated in PR 6461. Do not use, except for migrations.
 	slotsHasObjectBucket = []byte("slots-has-objects")
@@ -28,16 +23,11 @@ var (
 	archivedRootBucket = []byte("archived-index-root")
 
 	// Key indices buckets.
-	blockParentRootIndicesBucket        = []byte("block-parent-root-indices")
-	blockSlotIndicesBucket              = []byte("block-slot-indices")
-	stateSlotIndicesBucket              = []byte("state-slot-indices")
-	attestationHeadBlockRootBucket      = []byte("attestation-head-block-root-indices")
-	attestationSourceRootIndicesBucket  = []byte("attestation-source-root-indices")
-	attestationSourceEpochIndicesBucket = []byte("attestation-source-epoch-indices")
-	attestationTargetRootIndicesBucket  = []byte("attestation-target-root-indices")
-	attestationTargetEpochIndicesBucket = []byte("attestation-target-epoch-indices")
-	finalizedBlockRootsIndexBucket      = []byte("finalized-block-roots-index")
-	blockRootValidatorHashesBucket      = []byte("block-root-validator-hashes")
+	blockParentRootIndicesBucket   = []byte("block-parent-root-indices")
+	blockSlotIndicesBucket         = []byte("block-slot-indices")
+	stateSlotIndicesBucket         = []byte("state-slot-indices")
+	finalizedBlockRootsIndexBucket = []byte("finalized-block-roots-index")
+	blockRootValidatorHashesBucket = []byte("block-root-validator-hashes")
 
 	// Specific item keys.
 	headBlockRootKey           = []byte("head-root")
@@ -69,9 +59,6 @@ var (
 	// Deprecated: This index key was migrated in PR 6461. Do not use, except for migrations.
 	savedStateSlotsKey = []byte("saved-state-slots")
 
-	// New state management service compatibility bucket.
-	newStateServiceCompatibleBucket = []byte("new-state-compatible")
-
 	// Migrations
 	migrationsBucket = []byte("migrations")
 )

beacon-chain/db/kv/wss.go

@@ -11,6 +11,7 @@ import (
 	"github.com/prysmaticlabs/prysm/v5/proto/dbval"
 	ethpb "github.com/prysmaticlabs/prysm/v5/proto/prysm/v1alpha1"
 	"github.com/prysmaticlabs/prysm/v5/runtime/version"
+	"github.com/sirupsen/logrus"
 )
 
 // SaveOrigin loads an ssz serialized Block & BeaconState from an io.Reader
@@ -27,7 +28,11 @@ func (s *Store) SaveOrigin(ctx context.Context, serState, serBlock []byte) error
 		return fmt.Errorf("config mismatch, beacon node configured to connect to %s, detected state is for %s", params.BeaconConfig().ConfigName, cf.Config.ConfigName)
 	}
 
-	log.Infof("detected supported config for state & block version, config name=%s, fork name=%s", cf.Config.ConfigName, version.String(cf.Fork))
+	log.WithFields(logrus.Fields{
+		"configName": cf.Config.ConfigName,
+		"forkName":   version.String(cf.Fork),
+	}).Info("Detected supported config for state & block version")
+
 	state, err := cf.UnmarshalBeaconState(serState)
 	if err != nil {
 		return errors.Wrap(err, "failed to initialize origin state w/ bytes + config+fork")
@@ -57,13 +62,13 @@ func (s *Store) SaveOrigin(ctx context.Context, serState, serBlock []byte) error
 		return errors.Wrap(err, "unable to save backfill status data to db for checkpoint sync")
 	}
 
-	log.Infof("saving checkpoint block to db, w/ root=%#x", blockRoot)
+	log.WithField("root", fmt.Sprintf("%#x", blockRoot)).Info("Saving checkpoint block to db")
 	if err := s.SaveBlock(ctx, wblk); err != nil {
 		return errors.Wrap(err, "could not save checkpoint block")
 	}
 
 	// save state
-	log.Infof("calling SaveState w/ blockRoot=%x", blockRoot)
+	log.WithField("blockRoot", fmt.Sprintf("%#x", blockRoot)).Info("Calling SaveState")
 	if err = s.SaveState(ctx, state, blockRoot); err != nil {
 		return errors.Wrap(err, "could not save state")
 	}

beacon-chain/db/restore.go

@@ -22,7 +22,14 @@ func Restore(cliCtx *cli.Context) error {
 	targetDir := cliCtx.String(cmd.RestoreTargetDirFlag.Name)
 
 	restoreDir := path.Join(targetDir, kv.BeaconNodeDbDirName)
-	if file.Exists(path.Join(restoreDir, kv.DatabaseFileName)) {
+	restoreFile := path.Join(restoreDir, kv.DatabaseFileName)
+
+	dbExists, err := file.Exists(restoreFile, file.Regular)
+	if err != nil {
+		return errors.Wrapf(err, "could not check if database exists in %s", restoreFile)
+	}
+
+	if dbExists {
 		resp, err := prompt.ValidatePrompt(
 			os.Stdin, dbExistsYesNoPrompt, prompt.ValidateYesOrNo,
 		)

beacon-chain/db/slasherkv/BUILD.bazel

@@ -48,7 +48,6 @@ go_test(
         "//consensus-types/primitives:go_default_library",
         "//encoding/bytesutil:go_default_library",
         "//proto/prysm/v1alpha1:go_default_library",
-        "//testing/assert:go_default_library",
         "//testing/require:go_default_library",
         "//time/slots:go_default_library",
         "@com_github_prysmaticlabs_fastssz//:go_default_library",

beacon-chain/db/slasherkv/slasher.go

@@ -23,6 +23,10 @@ import (
 const (
 	attestationRecordKeySize = 32 // Bytes.
 	rootSize                 = 32 // Bytes.
+
+	// For database performance reasons, database read/write operations
+	// are chunked into batches of maximum `batchSize` elements.
+	batchSize = 10_000
 )
 
 // LastEpochWrittenForValidators given a list of validator indices returns the latest
@@ -66,12 +70,12 @@ func (s *Store) LastEpochWrittenForValidators(
 	return attestedEpochs, err
 }
 
-// SaveLastEpochsWrittenForValidators updates the latest epoch a slice
-// of validator indices has attested to.
-func (s *Store) SaveLastEpochsWrittenForValidators(
+// SaveLastEpochWrittenForValidators saves the latest epoch
+// that each validator has attested to in the provided map.
+func (s *Store) SaveLastEpochWrittenForValidators(
 	ctx context.Context, epochByValIndex map[primitives.ValidatorIndex]primitives.Epoch,
 ) error {
-	ctx, span := trace.StartSpan(ctx, "BeaconDB.SaveLastEpochsWrittenForValidators")
+	ctx, span := trace.StartSpan(ctx, "BeaconDB.SaveLastEpochWrittenForValidators")
 	defer span.End()
 
 	const batchSize = 10000
@@ -153,7 +157,7 @@ func (s *Store) CheckAttesterDoubleVotes(
 				attRecordsBkt := tx.Bucket(attestationRecordsBucket)
 
 				encEpoch := encodeTargetEpoch(attToProcess.IndexedAttestation.Data.Target.Epoch)
-				localDoubleVotes := []*slashertypes.AttesterDoubleVote{}
+				localDoubleVotes := make([]*slashertypes.AttesterDoubleVote, 0)
 
 				for _, valIdx := range attToProcess.IndexedAttestation.AttestingIndices {
 					// Check if there is signing root in the database for this combination
@@ -162,7 +166,7 @@ func (s *Store) CheckAttesterDoubleVotes(
 					validatorEpochKey := append(encEpoch, encIdx...)
 					attRecordsKey := signingRootsBkt.Get(validatorEpochKey)
 
-					// An attestation record key is comprised of a signing root (32 bytes).
+					// An attestation record key consists of a signing root (32 bytes).
 					if len(attRecordsKey) < attestationRecordKeySize {
 						// If there is no signing root for this combination,
 						// then there is no double vote. We can continue to the next validator.
@@ -259,14 +263,23 @@ func (s *Store) AttestationRecordForValidator(
 // then only the first one is (arbitrarily) saved in the `attestationDataRootsBucket` bucket.
 func (s *Store) SaveAttestationRecordsForValidators(
 	ctx context.Context,
-	attestations []*slashertypes.IndexedAttestationWrapper,
+	attWrappers []*slashertypes.IndexedAttestationWrapper,
 ) error {
 	_, span := trace.StartSpan(ctx, "BeaconDB.SaveAttestationRecordsForValidators")
 	defer span.End()
-	encodedTargetEpoch := make([][]byte, len(attestations))
-	encodedRecords := make([][]byte, len(attestations))
 
-	for i, attestation := range attestations {
+	attWrappersCount := len(attWrappers)
+
+	// If no attestations are provided, skip.
+	if attWrappersCount == 0 {
+		return nil
+	}
+
+	// Build encoded target epochs and encoded records
+	encodedTargetEpoch := make([][]byte, attWrappersCount)
+	encodedRecords := make([][]byte, attWrappersCount)
+
+	for i, attestation := range attWrappers {
 		encEpoch := encodeTargetEpoch(attestation.IndexedAttestation.Data.Target.Epoch)
 
 		value, err := encodeAttestationRecord(attestation)
@@ -278,60 +291,115 @@ func (s *Store) SaveAttestationRecordsForValidators(
 		encodedRecords[i] = value
 	}
 
-	return s.db.Update(func(tx *bolt.Tx) error {
-		attRecordsBkt := tx.Bucket(attestationRecordsBucket)
-		dataRootsBkt := tx.Bucket(attestationDataRootsBucket)
+	// Save attestation records in the database by batch.
+	for stop := attWrappersCount; stop >= 0; stop -= batchSize {
+		start := max(0, stop-batchSize)
 
-		for i := len(attestations) - 1; i >= 0; i-- {
-			attestation := attestations[i]
+		attWrappersBatch := attWrappers[start:stop]
+		encodedTargetEpochBatch := encodedTargetEpoch[start:stop]
+		encodedRecordsBatch := encodedRecords[start:stop]
 
-			if err := attRecordsBkt.Put(attestation.DataRoot[:], encodedRecords[i]); err != nil {
-				return err
-			}
-
-			for _, valIdx := range attestation.IndexedAttestation.AttestingIndices {
-				encIdx := encodeValidatorIndex(primitives.ValidatorIndex(valIdx))
-
-				key := append(encodedTargetEpoch[i], encIdx...)
-				if err := dataRootsBkt.Put(key, attestation.DataRoot[:]); err != nil {
-					return err
-				}
-			}
+		// Perform basic check.
+		if len(encodedTargetEpochBatch) != len(encodedRecordsBatch) {
+			return fmt.Errorf(
+				"cannot save attestation records, got %d target epochs and %d records",
+				len(encodedTargetEpochBatch), len(encodedRecordsBatch),
+			)
 		}
 
-		return nil
-	})
+		currentBatchSize := len(encodedTargetEpochBatch)
+
+		// Save attestation records in the database.
+		if err := s.db.Update(func(tx *bolt.Tx) error {
+			attRecordsBkt := tx.Bucket(attestationRecordsBucket)
+			dataRootsBkt := tx.Bucket(attestationDataRootsBucket)
+
+			for i := currentBatchSize - 1; i >= 0; i-- {
+				attWrapper := attWrappersBatch[i]
+				dataRoot := attWrapper.DataRoot
+
+				encodedTargetEpoch := encodedTargetEpochBatch[i]
+				encodedRecord := encodedRecordsBatch[i]
+
+				if err := attRecordsBkt.Put(dataRoot[:], encodedRecord); err != nil {
+					return err
+				}
+
+				for _, validatorIndex := range attWrapper.IndexedAttestation.AttestingIndices {
+					encodedIndex := encodeValidatorIndex(primitives.ValidatorIndex(validatorIndex))
+
+					key := append(encodedTargetEpoch, encodedIndex...)
+					if err := dataRootsBkt.Put(key, dataRoot[:]); err != nil {
+						return err
+					}
+				}
+			}
+
+			return nil
+		}); err != nil {
+			return errors.Wrap(err, "failed to save attestation records")
+		}
+	}
+
+	return nil
 }
 
 // LoadSlasherChunks given a chunk kind and a disk keys, retrieves chunks for a validator
 // min or max span used by slasher from our database.
 func (s *Store) LoadSlasherChunks(
-	ctx context.Context, kind slashertypes.ChunkKind, diskKeys [][]byte,
+	ctx context.Context, kind slashertypes.ChunkKind, chunkKeys [][]byte,
 ) ([][]uint16, []bool, error) {
 	_, span := trace.StartSpan(ctx, "BeaconDB.LoadSlasherChunk")
 	defer span.End()
-	chunks := make([][]uint16, 0)
-	var exists []bool
-	err := s.db.View(func(tx *bolt.Tx) error {
-		bkt := tx.Bucket(slasherChunksBucket)
-		for _, diskKey := range diskKeys {
-			key := append(ssz.MarshalUint8(make([]byte, 0), uint8(kind)), diskKey...)
-			chunkBytes := bkt.Get(key)
-			if chunkBytes == nil {
-				chunks = append(chunks, []uint16{})
-				exists = append(exists, false)
-				continue
+
+	keysCount := len(chunkKeys)
+
+	chunks := make([][]uint16, 0, keysCount)
+	exists := make([]bool, 0, keysCount)
+	encodedKeys := make([][]byte, 0, keysCount)
+
+	// Encode kind.
+	encodedKind := ssz.MarshalUint8(make([]byte, 0), uint8(kind))
+
+	// Encode keys.
+	for _, chunkKey := range chunkKeys {
+		encodedKey := append(encodedKind, chunkKey...)
+		encodedKeys = append(encodedKeys, encodedKey)
+	}
+
+	// Read chunks from the database by batch.
+	for start := 0; start < keysCount; start += batchSize {
+		stop := min(start+batchSize, len(encodedKeys))
+		encodedKeysBatch := encodedKeys[start:stop]
+
+		if err := s.db.View(func(tx *bolt.Tx) error {
+			bkt := tx.Bucket(slasherChunksBucket)
+
+			for _, encodedKey := range encodedKeysBatch {
+				chunkBytes := bkt.Get(encodedKey)
+
+				if chunkBytes == nil {
+					chunks = append(chunks, []uint16{})
+					exists = append(exists, false)
+					continue
+				}
+
+				chunk, err := decodeSlasherChunk(chunkBytes)
+				if err != nil {
+					return err
+				}
+
+				chunks = append(chunks, chunk)
+				exists = append(exists, true)
 			}
-			chunk, err := decodeSlasherChunk(chunkBytes)
-			if err != nil {
-				return err
-			}
-			chunks = append(chunks, chunk)
-			exists = append(exists, true)
+
+			return nil
+		}); err != nil {
+			return nil, nil, err
 		}
-		return nil
-	})
-	return chunks, exists, err
+	}
+
+	return chunks, exists, nil
 }
 
 // SaveSlasherChunks given a chunk kind, list of disk keys, and list of chunks,
@@ -341,25 +409,60 @@ func (s *Store) SaveSlasherChunks(
 ) error {
 	_, span := trace.StartSpan(ctx, "BeaconDB.SaveSlasherChunks")
 	defer span.End()
-	encodedKeys := make([][]byte, len(chunkKeys))
-	encodedChunks := make([][]byte, len(chunkKeys))
-	for i := 0; i < len(chunkKeys); i++ {
-		encodedKeys[i] = append(ssz.MarshalUint8(make([]byte, 0), uint8(kind)), chunkKeys[i]...)
-		encodedChunk, err := encodeSlasherChunk(chunks[i])
+
+	// Ensure we have the same number of keys and chunks.
+	if len(chunkKeys) != len(chunks) {
+		return fmt.Errorf(
+			"cannot save slasher chunks, got %d keys and %d chunks",
+			len(chunkKeys), len(chunks),
+		)
+	}
+
+	chunksCount := len(chunks)
+
+	// Encode kind.
+	encodedKind := ssz.MarshalUint8(make([]byte, 0), uint8(kind))
+
+	// Encode keys and chunks.
+	encodedKeys := make([][]byte, chunksCount)
+	encodedChunks := make([][]byte, chunksCount)
+
+	for i := 0; i < chunksCount; i++ {
+		chunkKey, chunk := chunkKeys[i], chunks[i]
+		encodedKey := append(encodedKind, chunkKey...)
+
+		encodedChunk, err := encodeSlasherChunk(chunk)
 		if err != nil {
-			return err
+			return errors.Wrapf(err, "failed to encode slasher chunk for key %v", chunkKey)
 		}
+
+		encodedKeys[i] = encodedKey
 		encodedChunks[i] = encodedChunk
 	}
-	return s.db.Update(func(tx *bolt.Tx) error {
-		bkt := tx.Bucket(slasherChunksBucket)
-		for i := 0; i < len(chunkKeys); i++ {
-			if err := bkt.Put(encodedKeys[i], encodedChunks[i]); err != nil {
-				return err
+
+	// Save chunks in the database by batch.
+	for start := 0; start < chunksCount; start += batchSize {
+		stop := min(start+batchSize, len(encodedKeys))
+		encodedKeysBatch := encodedKeys[start:stop]
+		encodedChunksBatch := encodedChunks[start:stop]
+		batchSize := len(encodedKeysBatch)
+
+		if err := s.db.Update(func(tx *bolt.Tx) error {
+			bkt := tx.Bucket(slasherChunksBucket)
+
+			for i := 0; i < batchSize; i++ {
+				if err := bkt.Put(encodedKeysBatch[i], encodedChunksBatch[i]); err != nil {
+					return err
+				}
 			}
+
+			return nil
+		}); err != nil {
+			return errors.Wrap(err, "failed to save slasher chunks")
 		}
-		return nil
-	})
+	}
+
+	return nil
 }
 
 // CheckDoubleBlockProposals takes in a list of proposals and for each,
@@ -594,7 +697,7 @@ func decodeSlasherChunk(enc []byte) ([]uint16, error) {
 }
 
 // Encode attestation record to bytes.
-// The output encoded attestation record consists in the signing root concatened with the compressed attestation record.
+// The output encoded attestation record consists in the signing root concatenated with the compressed attestation record.
 func encodeAttestationRecord(att *slashertypes.IndexedAttestationWrapper) ([]byte, error) {
 	if att == nil || att.IndexedAttestation == nil {
 		return []byte{}, errors.New("nil proposal record")
@@ -613,7 +716,7 @@ func encodeAttestationRecord(att *slashertypes.IndexedAttestationWrapper) ([]byt
 }
 
 // Decode attestation record from bytes.
-// The input encoded attestation record consists in the signing root concatened with the compressed attestation record.
+// The input encoded attestation record consists in the signing root concatenated with the compressed attestation record.
 func decodeAttestationRecord(encoded []byte) (*slashertypes.IndexedAttestationWrapper, error) {
 	if len(encoded) < rootSize {
 		return nil, fmt.Errorf("wrong length for encoded attestation record, want minimum %d, got %d", rootSize, len(encoded))

beacon-chain/db/slasherkv/slasher_test.go

@@ -14,33 +14,56 @@ import (
 	"github.com/prysmaticlabs/prysm/v5/consensus-types/primitives"
 	"github.com/prysmaticlabs/prysm/v5/encoding/bytesutil"
 	ethpb "github.com/prysmaticlabs/prysm/v5/proto/prysm/v1alpha1"
-	"github.com/prysmaticlabs/prysm/v5/testing/assert"
 	"github.com/prysmaticlabs/prysm/v5/testing/require"
 )
 
 func TestStore_AttestationRecordForValidator_SaveRetrieve(t *testing.T) {
-	ctx := context.Background()
-	beaconDB := setupDB(t)
-	valIdx := primitives.ValidatorIndex(1)
-	target := primitives.Epoch(5)
-	source := primitives.Epoch(4)
-	attRecord, err := beaconDB.AttestationRecordForValidator(ctx, valIdx, target)
-	require.NoError(t, err)
-	require.Equal(t, true, attRecord == nil)
+	const attestationsCount = 11_000
 
-	sr := [32]byte{1}
-	err = beaconDB.SaveAttestationRecordsForValidators(
-		ctx,
-		[]*slashertypes.IndexedAttestationWrapper{
-			createAttestationWrapper(source, target, []uint64{uint64(valIdx)}, sr[:]),
-		},
-	)
+	// Create context.
+	ctx := context.Background()
+
+	// Create database.
+	beaconDB := setupDB(t)
+
+	// Define the validator index.
+	validatorIndex := primitives.ValidatorIndex(1)
+
+	// Defines attestations to save and retrieve.
+	attWrappers := make([]*slashertypes.IndexedAttestationWrapper, attestationsCount)
+	for i := 0; i < attestationsCount; i++ {
+		var dataRoot [32]byte
+		binary.LittleEndian.PutUint64(dataRoot[:], uint64(i))
+
+		attWrapper := createAttestationWrapper(
+			primitives.Epoch(i),
+			primitives.Epoch(i+1),
+			[]uint64{uint64(validatorIndex)},
+			dataRoot[:],
+		)
+
+		attWrappers[i] = attWrapper
+	}
+
+	// Check on a sample of validators that no attestation records are available.
+	for i := 0; i < attestationsCount; i += 100 {
+		attRecord, err := beaconDB.AttestationRecordForValidator(ctx, validatorIndex, primitives.Epoch(i+1))
+		require.NoError(t, err)
+		require.Equal(t, true, attRecord == nil)
+	}
+
+	// Save the attestation records to the database.
+	err := beaconDB.SaveAttestationRecordsForValidators(ctx, attWrappers)
 	require.NoError(t, err)
-	attRecord, err = beaconDB.AttestationRecordForValidator(ctx, valIdx, target)
-	require.NoError(t, err)
-	assert.DeepEqual(t, target, attRecord.IndexedAttestation.Data.Target.Epoch)
-	assert.DeepEqual(t, source, attRecord.IndexedAttestation.Data.Source.Epoch)
-	assert.DeepEqual(t, sr, attRecord.DataRoot)
+
+	// Check on a sample of validators that attestation records are available.
+	for i := 0; i < attestationsCount; i += 100 {
+		expected := attWrappers[i]
+		actual, err := beaconDB.AttestationRecordForValidator(ctx, validatorIndex, primitives.Epoch(i+1))
+		require.NoError(t, err)
+
+		require.DeepEqual(t, expected.IndexedAttestation.Data.Source.Epoch, actual.IndexedAttestation.Data.Source.Epoch)
+	}
 }
 
 func TestStore_LastEpochWrittenForValidators(t *testing.T) {
@@ -66,7 +89,7 @@ func TestStore_LastEpochWrittenForValidators(t *testing.T) {
 	require.NoError(t, err)
 	require.Equal(t, 0, len(attestedEpochs))
 
-	err = beaconDB.SaveLastEpochsWrittenForValidators(ctx, epochsByValidator)
+	err = beaconDB.SaveLastEpochWrittenForValidators(ctx, epochsByValidator)
 	require.NoError(t, err)
 
 	retrievedEpochs, err := beaconDB.LastEpochWrittenForValidators(ctx, indices)
@@ -138,61 +161,113 @@ func TestStore_CheckAttesterDoubleVotes(t *testing.T) {
 }
 
 func TestStore_SlasherChunk_SaveRetrieve(t *testing.T) {
+	// Define test parameters.
+	const (
+		elemsPerChunk = 16
+		totalChunks   = 11_000
+	)
+
+	// Create context.
 	ctx := context.Background()
+
+	// Create database.
 	beaconDB := setupDB(t)
-	elemsPerChunk := 16
-	totalChunks := 64
-	chunkKeys := make([][]byte, totalChunks)
-	chunks := make([][]uint16, totalChunks)
+
+	// Create min chunk keys and chunks.
+	minChunkKeys := make([][]byte, totalChunks)
+	minChunks := make([][]uint16, totalChunks)
+
 	for i := 0; i < totalChunks; i++ {
+		// Create chunk key.
+		chunkKey := ssz.MarshalUint64(make([]byte, 0), uint64(i))
+		minChunkKeys[i] = chunkKey
+
+		// Create chunk.
 		chunk := make([]uint16, elemsPerChunk)
+
 		for j := 0; j < len(chunk); j++ {
-			chunk[j] = uint16(0)
+			chunk[j] = uint16(i + j)
 		}
-		chunks[i] = chunk
-		chunkKeys[i] = ssz.MarshalUint64(make([]byte, 0), uint64(i))
+
+		minChunks[i] = chunk
 	}
 
-	// We save chunks for min spans.
-	err := beaconDB.SaveSlasherChunks(ctx, slashertypes.MinSpan, chunkKeys, chunks)
+	// Create max chunk keys and chunks.
+	maxChunkKeys := make([][]byte, totalChunks)
+	maxChunks := make([][]uint16, totalChunks)
+
+	for i := 0; i < totalChunks; i++ {
+		// Create chunk key.
+		chunkKey := ssz.MarshalUint64(make([]byte, 0), uint64(i+1))
+		maxChunkKeys[i] = chunkKey
+
+		// Create chunk.
+		chunk := make([]uint16, elemsPerChunk)
+
+		for j := 0; j < len(chunk); j++ {
+			chunk[j] = uint16(i + j + 1)
+		}
+
+		maxChunks[i] = chunk
+	}
+
+	// Save chunks for min spans.
+	err := beaconDB.SaveSlasherChunks(ctx, slashertypes.MinSpan, minChunkKeys, minChunks)
 	require.NoError(t, err)
 
-	// We expect no chunks to be stored for max spans.
+	// Expect no chunks to be stored for max spans.
 	_, chunksExist, err := beaconDB.LoadSlasherChunks(
-		ctx, slashertypes.MaxSpan, chunkKeys,
+		ctx, slashertypes.MaxSpan, minChunkKeys,
 	)
 	require.NoError(t, err)
-	require.Equal(t, len(chunks), len(chunksExist))
+	require.Equal(t, len(minChunks), len(chunksExist))
+
 	for _, exists := range chunksExist {
 		require.Equal(t, false, exists)
 	}
 
-	// We check we saved the right chunks.
+	// Check the right chunks are saved.
 	retrievedChunks, chunksExist, err := beaconDB.LoadSlasherChunks(
-		ctx, slashertypes.MinSpan, chunkKeys,
+		ctx, slashertypes.MinSpan, minChunkKeys,
 	)
 	require.NoError(t, err)
-	require.Equal(t, len(chunks), len(retrievedChunks))
-	require.Equal(t, len(chunks), len(chunksExist))
+	require.Equal(t, len(minChunks), len(retrievedChunks))
+	require.Equal(t, len(minChunks), len(chunksExist))
+
 	for i, exists := range chunksExist {
 		require.Equal(t, true, exists)
-		require.DeepEqual(t, chunks[i], retrievedChunks[i])
+		require.DeepEqual(t, minChunks[i], retrievedChunks[i])
 	}
 
-	// We save chunks for max spans.
-	err = beaconDB.SaveSlasherChunks(ctx, slashertypes.MaxSpan, chunkKeys, chunks)
+	// Save chunks for max spans.
+	err = beaconDB.SaveSlasherChunks(ctx, slashertypes.MaxSpan, maxChunkKeys, maxChunks)
 	require.NoError(t, err)
 
-	// We check we saved the right chunks.
+	// Check right chunks are saved.
 	retrievedChunks, chunksExist, err = beaconDB.LoadSlasherChunks(
-		ctx, slashertypes.MaxSpan, chunkKeys,
+		ctx, slashertypes.MaxSpan, maxChunkKeys,
 	)
 	require.NoError(t, err)
-	require.Equal(t, len(chunks), len(retrievedChunks))
-	require.Equal(t, len(chunks), len(chunksExist))
+
+	require.Equal(t, len(maxChunks), len(retrievedChunks))
+	require.Equal(t, len(maxChunks), len(chunksExist))
+
 	for i, exists := range chunksExist {
 		require.Equal(t, true, exists)
-		require.DeepEqual(t, chunks[i], retrievedChunks[i])
+		require.DeepEqual(t, maxChunks[i], retrievedChunks[i])
+	}
+
+	// Check the right chunks are still saved for min span.
+	retrievedChunks, chunksExist, err = beaconDB.LoadSlasherChunks(
+		ctx, slashertypes.MinSpan, minChunkKeys,
+	)
+	require.NoError(t, err)
+	require.Equal(t, len(minChunks), len(retrievedChunks))
+	require.Equal(t, len(minChunks), len(chunksExist))
+
+	for i, exists := range chunksExist {
+		require.Equal(t, true, exists)
+		require.DeepEqual(t, minChunks[i], retrievedChunks[i])
 	}
 }
 

beacon-chain/execution/block_reader.go

@@ -110,7 +110,7 @@ func (s *Service) BlockByTimestamp(ctx context.Context, time uint64) (*types.Hea
 		return nil, errors.Wrap(errBlockTimeTooLate, fmt.Sprintf("(%d > %d)", time, latestBlkTime))
 	}
 	// Initialize a pointer to eth1 chain's history to start our search from.
-	cursorNum := big.NewInt(0).SetUint64(latestBlkHeight)
+	cursorNum := new(big.Int).SetUint64(latestBlkHeight)
 	cursorTime := latestBlkTime
 
 	var numOfBlocks uint64
@@ -156,15 +156,15 @@ func (s *Service) BlockByTimestamp(ctx context.Context, time uint64) (*types.Hea
 		return s.retrieveHeaderInfo(ctx, cursorNum.Uint64())
 	}
 	if cursorTime > time {
-		return s.findMaxTargetEth1Block(ctx, big.NewInt(0).SetUint64(estimatedBlk), time)
+		return s.findMaxTargetEth1Block(ctx, new(big.Int).SetUint64(estimatedBlk), time)
 	}
-	return s.findMinTargetEth1Block(ctx, big.NewInt(0).SetUint64(estimatedBlk), time)
+	return s.findMinTargetEth1Block(ctx, new(big.Int).SetUint64(estimatedBlk), time)
 }
 
 // Performs a search to find a target eth1 block which is earlier than or equal to the
 // target time. This method is used when head.time > targetTime
 func (s *Service) findMaxTargetEth1Block(ctx context.Context, upperBoundBlk *big.Int, targetTime uint64) (*types.HeaderInfo, error) {
-	for bn := upperBoundBlk; ; bn = big.NewInt(0).Sub(bn, big.NewInt(1)) {
+	for bn := upperBoundBlk; ; bn = new(big.Int).Sub(bn, big.NewInt(1)) {
 		if ctx.Err() != nil {
 			return nil, ctx.Err()
 		}
@@ -181,7 +181,7 @@ func (s *Service) findMaxTargetEth1Block(ctx context.Context, upperBoundBlk *big
 // Performs a search to find a target eth1 block which is just earlier than or equal to the
 // target time. This method is used when head.time < targetTime
 func (s *Service) findMinTargetEth1Block(ctx context.Context, lowerBoundBlk *big.Int, targetTime uint64) (*types.HeaderInfo, error) {
-	for bn := lowerBoundBlk; ; bn = big.NewInt(0).Add(bn, big.NewInt(1)) {
+	for bn := lowerBoundBlk; ; bn = new(big.Int).Add(bn, big.NewInt(1)) {
 		if ctx.Err() != nil {
 			return nil, ctx.Err()
 		}
@@ -201,7 +201,7 @@ func (s *Service) findMinTargetEth1Block(ctx context.Context, lowerBoundBlk *big
 }
 
 func (s *Service) retrieveHeaderInfo(ctx context.Context, bNum uint64) (*types.HeaderInfo, error) {
-	bn := big.NewInt(0).SetUint64(bNum)
+	bn := new(big.Int).SetUint64(bNum)
 	exists, info, err := s.headerCache.HeaderInfoByHeight(bn)
 	if err != nil {
 		return nil, err

beacon-chain/execution/errors.go

@@ -23,9 +23,6 @@ var (
 	ErrInvalidPayloadAttributes = errors.New("payload attributes are invalid / inconsistent")
 	// ErrUnknownPayloadStatus when the payload status is unknown.
 	ErrUnknownPayloadStatus = errors.New("unknown payload status")
-	// ErrConfigMismatch when the execution node's terminal total difficulty or
-	// terminal block hash received via the API mismatches Prysm's configuration value.
-	ErrConfigMismatch = errors.New("execution client configuration mismatch")
 	// ErrAcceptedSyncingPayloadStatus when the status of the payload is syncing or accepted.
 	ErrAcceptedSyncingPayloadStatus = errors.New("payload status is SYNCING or ACCEPTED")
 	// ErrInvalidPayloadStatus when the status of the payload is invalid.

beacon-chain/execution/log_processing.go

@@ -269,7 +269,7 @@ func (s *Service) ProcessChainStart(genesisTime uint64, eth1BlockHash [32]byte,
 	}
 
 	log.WithFields(logrus.Fields{
-		"ChainStartTime": chainStartTime,
+		"chainStartTime": chainStartTime,
 	}).Info("Minimum number of validators reached for beacon-chain to start")
 	s.cfg.stateNotifier.StateFeed().Send(&feed.Event{
 		Type: statefeed.ChainStarted,
@@ -298,9 +298,7 @@ func (s *Service) processPastLogs(ctx context.Context) error {
 	// Start from the deployment block if our last requested block
 	// is behind it. This is as the deposit logs can only start from the
 	// block of the deployment of the deposit contract.
-	if deploymentBlock > currentBlockNum {
-		currentBlockNum = deploymentBlock
-	}
+	currentBlockNum = max(currentBlockNum, deploymentBlock)
 	// To store all blocks.
 	headersMap := make(map[uint64]*types.HeaderInfo)
 	rawLogCount, err := s.depositContractCaller.GetDepositCount(&bind.CallOpts{})
@@ -384,15 +382,13 @@ func (s *Service) processBlockInBatch(ctx context.Context, currentBlockNum uint6
 	end := currentBlockNum + batchSize
 	// Appropriately bound the request, as we do not
 	// want request blocks beyond the current follow distance.
-	if end > latestFollowHeight {
-		end = latestFollowHeight
-	}
+	end = min(end, latestFollowHeight)
 	query := ethereum.FilterQuery{
 		Addresses: []common.Address{
 			s.cfg.depositContractAddr,
 		},
-		FromBlock: big.NewInt(0).SetUint64(start),
-		ToBlock:   big.NewInt(0).SetUint64(end),
+		FromBlock: new(big.Int).SetUint64(start),
+		ToBlock:   new(big.Int).SetUint64(end),
 	}
 	remainingLogs := logCount - uint64(s.lastReceivedMerkleIndex+1)
 	// only change the end block if the remaining logs are below the required log limit.
@@ -400,7 +396,7 @@ func (s *Service) processBlockInBatch(ctx context.Context, currentBlockNum uint6
 	withinLimit := remainingLogs < depositLogRequestLimit
 	aboveFollowHeight := end >= latestFollowHeight
 	if withinLimit && aboveFollowHeight {
-		query.ToBlock = big.NewInt(0).SetUint64(latestFollowHeight)
+		query.ToBlock = new(big.Int).SetUint64(latestFollowHeight)
 		end = latestFollowHeight
 	}
 	logs, err := s.httpLogger.FilterLogs(ctx, query)
@@ -482,11 +478,11 @@ func (s *Service) requestBatchedHeadersAndLogs(ctx context.Context) error {
 	}
 	for i := s.latestEth1Data.LastRequestedBlock + 1; i <= requestedBlock; i++ {
 		// Cache eth1 block header here.
-		_, err := s.BlockHashByHeight(ctx, big.NewInt(0).SetUint64(i))
+		_, err := s.BlockHashByHeight(ctx, new(big.Int).SetUint64(i))
 		if err != nil {
 			return err
 		}
-		err = s.ProcessETH1Block(ctx, big.NewInt(0).SetUint64(i))
+		err = s.ProcessETH1Block(ctx, new(big.Int).SetUint64(i))
 		if err != nil {
 			return err
 		}

beacon-chain/execution/prometheus_test.go

@@ -10,7 +10,7 @@ import (
 )
 
 // TestCleanup ensures that the cleanup function unregisters the prometheus.Collection
-// also tests the interchangability of the explicit prometheus Register/Unregister
+// also tests the interchangeability of the explicit prometheus Register/Unregister
 // and the implicit methods within the collector implementation
 func TestCleanup(t *testing.T) {
 	ctx := context.Background()
@@ -32,11 +32,11 @@ func TestCleanup(t *testing.T) {
 	assert.Equal(t, true, unregistered, "prometheus.Unregister failed to unregister PowchainCollector on final cleanup")
 }
 
-// TestCancelation tests that canceling the context passed into
+// TestCancellation tests that canceling the context passed into
 // NewPowchainCollector cleans everything up as expected. This
 // does come at the cost of an extra channel cluttering up
 // PowchainCollector, just for this test.
-func TestCancelation(t *testing.T) {
+func TestCancellation(t *testing.T) {
 	ctx, cancel := context.WithCancel(context.Background())
 	pc, err := NewPowchainCollector(ctx)
 	assert.NoError(t, err, "Unexpected error calling NewPowchainCollector")