@lodestar/beacon-node 1.47.0-dev.0bcaaaebb5 → 1.47.0-dev.101e2c290d

package/src/chain/validation/proposerPreferences.ts

@@ -1,9 +1,9 @@
 import {IForkChoice, ProtoBlock} from "@lodestar/fork-choice";
-import {GENESIS_EPOCH, GENESIS_SLOT, MIN_SEED_LOOKAHEAD, SLOTS_PER_EPOCH} from "@lodestar/params";
+import {GENESIS_EPOCH, GENESIS_SLOT, MIN_SEED_LOOKAHEAD, SLOTS_PER_EPOCH, isForkPostGloas} from "@lodestar/params";
 import {
   computeEpochAtSlot,
   computeStartSlotAtEpoch,
-  createSingleSignatureSetFromComponents,
+  createIndexedSignatureSetFromComponents,
   getProposerPreferencesSigningRoot,
 } from "@lodestar/state-transition";
 import {Epoch, Slot, ValidatorIndex, gloas} from "@lodestar/types";
@@ -13,7 +13,7 @@ import {IBeaconChain} from "../index.js";
 
 /**
  * Validates a gossiped `SignedProposerPreferences` per
- * https://github.com/ethereum/consensus-specs/blob/v1.7.0-alpha.13/specs/gloas/p2p-interface.md#new-proposer_preferences
+ * https://github.com/ethereum/consensus-specs/blob/v1.7.0-alpha.14/specs/gloas/p2p-interface.md#new-proposer_preferences
  */
 export async function validateGossipProposerPreferences(
   chain: IBeaconChain,
@@ -24,6 +24,16 @@ export async function validateGossipProposerPreferences(
   const dependentRootHex = toRootHex(dependentRoot);
   const proposalEpoch = computeEpochAtSlot(proposalSlot);
 
+  // [IGNORE] The proposal epoch is after the Gloas upgrade
+  const proposalFork = chain.config.getForkName(proposalSlot);
+  if (!isForkPostGloas(proposalFork)) {
+    throw new ProposerPreferencesError(GossipAction.IGNORE, {
+      code: ProposerPreferencesErrorCode.PRE_GLOAS_PROPOSAL_SLOT,
+      proposalSlot,
+      proposalFork,
+    });
+  }
+
   // [IGNORE] `preferences.proposal_slot` is within the proposer lookahead,
   // allowing for `MAXIMUM_GOSSIP_CLOCK_DISPARITY`.
   const currentEpoch = computeEpochAtSlot(chain.clock.currentSlotWithGossipDisparity);
@@ -129,8 +139,8 @@ export async function validateGossipProposerPreferences(
   }
 
   // [REJECT] `signed_proposer_preferences.signature` is valid with respect to the validator's public key.
-  const signatureSet = createSingleSignatureSetFromComponents(
-    chain.pubkeyCache.getOrThrow(validatorIndex),
+  const signatureSet = createIndexedSignatureSetFromComponents(
+    validatorIndex,
     getProposerPreferencesSigningRoot(chain.config, preferences),
     signedProposerPreferences.signature
   );

package/src/chain/validation/proposerSlashing.ts

@@ -37,7 +37,7 @@ async function validateProposerSlashing(
   try {
     const proposer = state.getValidator(proposerSlashing.signedHeader1.message.proposerIndex);
     // verifySignature = false, verified in batch below
-    assertValidProposerSlashing(chain.config, chain.pubkeyCache, state.slot, proposerSlashing, proposer, false);
+    assertValidProposerSlashing(chain.config, state.slot, proposerSlashing, proposer, false);
   } catch (e) {
     throw new ProposerSlashingError(GossipAction.REJECT, {
       code: ProposerSlashingErrorCode.INVALID,

package/src/chain/validation/voluntaryExit.ts

@@ -9,12 +9,69 @@ import {
 import {IBeaconChain} from "../index.js";
 import {RegenCaller} from "../regen/index.js";
 
+export type ApiVoluntaryExitResult = {status: "published"} | {status: "deferred"; validity: VoluntaryExitValidity};
+
+// Variants that can become valid in a future epoch without user action.
+// - earlyEpoch: exit.message.epoch is in the future; valid once current epoch catches up
+// - shortTimeActive: validator active < SHARD_COMMITTEE_PERIOD; valid once enough time passes
+// - pendingWithdrawals: Electra; valid once pending partial withdrawals drain
+// Note: VoluntaryExitValidity.inactive is intentionally excluded. It conflates
+// "validator does not exist" (permanent) with "validator not yet activated"
+// (transient), and cleanly classifying it requires splitting the enum variant
+// upstream. Left for a future follow-up.
+
+export function isTransientExitValidity(v: VoluntaryExitValidity): boolean {
+  switch (v) {
+    case VoluntaryExitValidity.earlyEpoch:
+    case VoluntaryExitValidity.shortTimeActive:
+    case VoluntaryExitValidity.pendingWithdrawals:
+      return true;
+    case VoluntaryExitValidity.valid:
+    case VoluntaryExitValidity.inactive:
+    case VoluntaryExitValidity.alreadyExited:
+    case VoluntaryExitValidity.invalidSignature:
+      return false;
+  }
+}
+
+// Comments for each call are present inside `validateVoluntaryExit`.
 export async function validateApiVoluntaryExit(
   chain: IBeaconChain,
   voluntaryExit: phase0.SignedVoluntaryExit
-): Promise<void> {
+): Promise<ApiVoluntaryExitResult> {
   const prioritizeBls = true;
-  return validateVoluntaryExit(chain, voluntaryExit, prioritizeBls);
+
+  if (
+    chain.opPool.hasSeenVoluntaryExit(voluntaryExit.message.validatorIndex) ||
+    chain.deferredVoluntaryExitPool.has(voluntaryExit.message.validatorIndex)
+  ) {
+    throw new VoluntaryExitError(GossipAction.IGNORE, {
+      code: VoluntaryExitErrorCode.ALREADY_EXISTS,
+    });
+  }
+
+  const state = await chain.getHeadStateAtCurrentEpoch(RegenCaller.validateApiVoluntaryExit);
+  const validity = state.getVoluntaryExitValidity(voluntaryExit, false);
+
+  if (validity !== VoluntaryExitValidity.valid && !isTransientExitValidity(validity)) {
+    throw new VoluntaryExitError(GossipAction.REJECT, {
+      code: voluntaryExitValidityToErrorCode(validity),
+    });
+  }
+
+  const signatureSet = getVoluntaryExitSignatureSet(chain.config, state, voluntaryExit);
+  if (!(await chain.bls.verifySignatureSets([signatureSet], {batchable: true, priority: prioritizeBls}))) {
+    throw new VoluntaryExitError(GossipAction.REJECT, {
+      code: VoluntaryExitErrorCode.INVALID_SIGNATURE,
+    });
+  }
+
+  if (validity !== VoluntaryExitValidity.valid) {
+    // Transient failure — signature is good, defer
+    return {status: "deferred", validity};
+  }
+
+  return {status: "published"};
 }
 
 export async function validateGossipVoluntaryExit(

package/src/execution/builder/http.ts

@@ -71,15 +71,18 @@ export class NoBidReceived extends Error {
 }
 
 /**
- * Beacon clients select randomized values from the following ranges when initializing
- * the circuit breaker (so at boot time and once for each unique boot).
+ * Default circuit breaker parameters:
  *
- * ALLOWED_FAULTS: between 1 and SLOTS_PER_EPOCH // 4
- * FAULT_INSPECTION_WINDOW: between SLOTS_PER_EPOCH and 2 * SLOTS_PER_EPOCH
+ * SLOTS_PER_EPOCH <= FAULT_INSPECTION_WINDOW < 2 * SLOTS_PER_EPOCH (randomized at initialization)
+ * ALLOWED_FAULTS: FAULT_INSPECTION_WINDOW // 4
  *
- * The values are randomized per node so builders cannot predict when a given proposer will
- * fall back to local blocks. With fixed thresholds a builder could withhold payloads right up
- * to the limit without ever tripping the breaker.
+ * e.g. on mainnet SLOTS_PER_EPOCH is 32, so FAULT_INSPECTION_WINDOW is between 32 and 63,
+ * and a window of 40 results in ALLOWED_FAULTS = 10.
+ *
+ * The randomized default window keeps the exact slots under inspection unpredictable per node,
+ * so a builder cannot tell when past faults age out of a given proposer's window and time
+ * withholding around trip or recovery points. Explicitly configured windows are clamped to at
+ * least SLOTS_PER_EPOCH, and configured allowedFaults is capped at the default ~25% budget.
  */
 export function getFaultInspectionParams(opts: {faultInspectionWindow?: number; allowedFaults?: number}): {
   faultInspectionWindow: number;
@@ -90,10 +93,7 @@ export function getFaultInspectionParams(opts: {faultInspectionWindow?: number;
     SLOTS_PER_EPOCH
   );
   // allowedFaults should be < faultInspectionWindow, limiting them to faultInspectionWindow/4
-  const allowedFaults = Math.min(
-    opts.allowedFaults ?? Math.floor(faultInspectionWindow / 4),
-    Math.floor(faultInspectionWindow / 4)
-  );
+  const allowedFaults = Math.min(opts.allowedFaults ?? Infinity, Math.floor(faultInspectionWindow / 4));
   return {faultInspectionWindow, allowedFaults};
 }
 

package/src/metrics/metrics/beacon.ts

@@ -143,17 +143,17 @@ export function createBeaconMetrics(register: RegistryMetricCreator) {
         name: "beacon_builder_circuit_breaker_active",
         help: "Whether the builder circuit breaker is active (1) causing builder bids to be ignored",
       }),
-      faults: register.gauge({
-        name: "beacon_builder_circuit_breaker_faults",
-        help: "Count of blocks with unrevealed payloads in the fault inspection window",
+      canonicalBlocks: register.gauge({
+        name: "beacon_builder_circuit_breaker_canonical_blocks",
+        help: "Count of canonical FULL and EMPTY blocks in the fault inspection window",
       }),
-      blocksPresent: register.gauge({
-        name: "beacon_builder_circuit_breaker_blocks_present",
-        help: "Count of blocks present in the fault inspection window",
+      fullBlocks: register.gauge({
+        name: "beacon_builder_circuit_breaker_full_blocks",
+        help: "Count of canonical FULL blocks in the fault inspection window",
       }),
-      payloadsRevealed: register.gauge({
-        name: "beacon_builder_circuit_breaker_payloads_revealed",
-        help: "Count of blocks with revealed payloads in the fault inspection window",
+      emptyBlocks: register.gauge({
+        name: "beacon_builder_circuit_breaker_empty_blocks",
+        help: "Count of canonical EMPTY blocks in the fault inspection window",
       }),
     },
 

package/src/metrics/metrics/lodestar.ts

@@ -4,7 +4,10 @@ import {ArchiveStoreTask} from "../../chain/archiveStore/archiveStore.js";
 import {FrequencyStateArchiveStep} from "../../chain/archiveStore/strategies/frequencyStateArchiveStrategy.js";
 import {BlockInputSource} from "../../chain/blocks/blockInput/index.js";
 import {PayloadErrorCode} from "../../chain/blocks/importExecutionPayload.js";
-import {PayloadEnvelopeInputSource} from "../../chain/blocks/payloadEnvelopeInput/index.js";
+import {
+  PayloadEnvelopeInputPruneReason,
+  PayloadEnvelopeInputSource,
+} from "../../chain/blocks/payloadEnvelopeInput/index.js";
 import {JobQueueItemType} from "../../chain/bls/index.js";
 import {AttestationErrorCode, BlockErrorCode} from "../../chain/errors/index.js";
 import {
@@ -21,7 +24,7 @@ import {ExecutionPayloadStatus} from "../../execution/index.js";
 import {GossipType} from "../../network/index.js";
 import {CannotAcceptWorkReason, ReprocessRejectReason} from "../../network/processor/index.js";
 import {BackfillSyncMethod} from "../../sync/backfill/backfill.js";
-import {PendingBlockType} from "../../sync/types.js";
+import {DroppedItemReason, PendingBlockType} from "../../sync/types.js";
 import {PeerSyncType, RangeSyncType} from "../../sync/utils/remoteSyncType.js";
 import {AllocSource} from "../../util/bufferPool.js";
 import {DataColumnReconstructionCode} from "../../util/dataColumns.js";
@@ -31,6 +34,10 @@ import {RegistryMetricCreator} from "../utils/registryMetricCreator.js";
 
 export type LodestarMetrics = ReturnType<typeof createLodestarMetrics>;
 
+type BlsJobOutcome = "valid" | "invalid" | "prepError" | "verifyError" | "workerError";
+type BlsBufferFlushReason = "size" | "timeout";
+type BlsVerificationCallOperation = "batch" | "single" | "fallback" | "same_message";
+
 /**
  * Extra Lodestar custom metrics
  */
@@ -396,86 +403,108 @@ export function createLodestarMetrics(
     // BLS verifier thread pool and queue
 
     bls: {
-      aggregatedPubkeys: register.gauge({
+      aggregatedPubkeys: register.counter({
         name: "lodestar_bls_aggregated_pubkeys_total",
-        help: "Total aggregated pubkeys for BLS validation",
+        help: "Total validator public keys used as inputs to aggregate signature sets",
+      }),
+      pubkeyCacheSize: register.gauge({
+        name: "lodestar_bls_pubkey_cache_size",
+        help: "Current number of validator public keys in the native cache",
+      }),
+      pubkeyCacheCapacity: register.gauge({
+        name: "lodestar_bls_pubkey_cache_capacity",
+        help: "Number of validator public keys the native cache can hold without growing",
       }),
     },
 
     blsThreadPool: {
-      jobsWorkerTime: register.gauge<{workerId: number}>({
+      jobsWorkerTime: register.counter<{workerId: number}>({
         name: "lodestar_bls_thread_pool_time_seconds_sum",
-        help: "Total time spent verifying signature sets measured on the worker",
+        help: "Cumulative wall time spent processing BLS dispatch groups by worker",
         labelNames: ["workerId"],
       }),
-      successJobsSignatureSetsCount: register.gauge({
+      successJobsSignatureSetsCount: register.counter({
         name: "lodestar_bls_thread_pool_success_jobs_signature_sets_count",
-        help: "Count of total verified signature sets",
-      }),
-      errorAggregateSignatureSetsCount: register.gauge<{type: JobQueueItemType}>({
-        name: "lodestar_bls_thread_pool_error_aggregate_signature_sets_count",
-        help: "Count of error when aggregating pubkeys or signatures",
-        labelNames: ["type"],
+        help: "Count of signature sets completed without an operational error",
       }),
-      errorJobsSignatureSetsCount: register.gauge({
+      errorJobsSignatureSetsCount: register.counter({
         name: "lodestar_bls_thread_pool_error_jobs_signature_sets_count",
-        help: "Count of total error-ed signature sets",
+        help: "Count of signature sets that ended in an input, binding, or worker error",
       }),
-      jobWaitTime: register.histogram({
+      jobWaitTime: register.histogram<{type: JobQueueItemType}>({
         name: "lodestar_bls_thread_pool_queue_job_wait_time_seconds",
-        help: "Time from job added to the queue to starting the job in seconds",
-        buckets: [0.01, 0.02, 0.1, 0.3, 0.5, 1],
+        help: "Time from adding a BLS job until it is selected for a worker dispatch group",
+        labelNames: ["type"],
+        buckets: [0.005, 0.01, 0.025, 0.05, 0.1, 0.2, 0.5, 1, 2],
+      }),
+      jobDuration: register.histogram<{type: JobQueueItemType}>({
+        name: "lodestar_bls_thread_pool_job_duration_seconds",
+        help: "End-to-end duration of a BLS job from submission through completion",
+        labelNames: ["type"],
+        buckets: [0.005, 0.01, 0.025, 0.05, 0.1, 0.2, 0.5, 1, 2],
+      }),
+      jobResults: register.counter<{type: JobQueueItemType; outcome: BlsJobOutcome}>({
+        name: "lodestar_bls_thread_pool_job_results_total",
+        help: "Count of completed BLS jobs by cryptographic or operational outcome",
+        labelNames: ["type", "outcome"],
       }),
       queueLength: register.gauge({
         name: "lodestar_bls_thread_pool_queue_length",
-        help: "Count of total block processor queue length",
+        help: "Count of ready BLS jobs waiting for a worker",
+      }),
+      bufferedQueueLength: register.gauge({
+        name: "lodestar_bls_thread_pool_buffered_queue_length",
+        help: "Count of BLS jobs waiting in the intentional batching buffer",
       }),
       workersBusy: register.gauge({
         name: "lodestar_bls_thread_pool_workers_busy",
         help: "Count of current busy workers",
       }),
-      totalJobsGroupsStarted: register.gauge({
+      totalJobsGroupsStarted: register.counter({
         name: "lodestar_bls_thread_pool_job_groups_started_total",
-        help: "Count of total jobs groups started in bls thread pool, job groups include +1 jobs",
+        help: "Count of worker dispatch groups started by the BLS thread pool",
       }),
-      totalJobsStarted: register.gauge<{type: JobQueueItemType}>({
+      totalJobsStarted: register.counter<{type: JobQueueItemType}>({
         name: "lodestar_bls_thread_pool_jobs_started_total",
-        help: "Count of total jobs started in bls thread pool, jobs include +1 signature sets",
+        help: "Count of jobs started by the BLS thread pool",
         labelNames: ["type"],
       }),
-      totalSigSetsStarted: register.gauge<{type: JobQueueItemType}>({
+      totalSigSetsStarted: register.counter<{type: JobQueueItemType}>({
         name: "lodestar_bls_thread_pool_sig_sets_started_total",
-        help: "Count of total signature sets started in bls thread pool, sig sets include 1 pk, msg, sig",
+        help: "Count of signature sets started by the BLS thread pool",
         labelNames: ["type"],
       }),
       // Re-verifying a batch means doing double work. This number must be very low or it can be a waste of CPU resources
-      batchRetries: register.gauge({
+      batchRetries: register.counter({
         name: "lodestar_bls_thread_pool_batch_retries_total",
-        help: "Count of total batches that failed and had to be verified again.",
+        help: "Count of combined batches that failed and required per-job fallback",
       }),
       // To count how many sigs are being validated with the optimization of batching them
-      batchSigsSuccess: register.gauge({
+      batchSigsSuccess: register.counter({
         name: "lodestar_bls_thread_pool_batch_sigs_success_total",
-        help: "Count of total batches that failed and had to be verified again.",
+        help: "Count of signature sets accepted by combined batch verification without fallback",
       }),
-      sameMessageRetryJobs: register.gauge({
-        name: "lodestar_bls_thread_pool_same_message_jobs_retries_total",
-        help: "Count of total same message jobs that failed and had to be verified again.",
+      bufferFlushes: register.counter<{reason: BlsBufferFlushReason}>({
+        name: "lodestar_bls_thread_pool_buffer_flushes_total",
+        help: "Count of batching-buffer flushes by trigger",
+        labelNames: ["reason"],
       }),
-      sameMessageRetrySets: register.gauge({
-        name: "lodestar_bls_thread_pool_same_message_sets_retries_total",
-        help: "Count of total same message sets that failed and had to be verified again.",
+      bufferedSignatureSets: register.histogram({
+        name: "lodestar_bls_thread_pool_buffered_signature_sets",
+        help: "Count of signature sets included in each batching-buffer flush",
+        buckets: [1, 8, 16, 24, 28, 32, 36, 40, 44, 48, 64, 128],
       }),
       // To measure the time cost of main thread <-> worker message passing
       latencyToWorker: register.histogram({
         name: "lodestar_bls_thread_pool_latency_to_worker",
         help: "Time from sending the job to the worker and the worker receiving it",
-        buckets: [0.001, 0.003, 0.01, 0.03, 0.1],
+        // The dashboard treats 20ms as slow; values above 100ms only need severe-stall classification.
+        buckets: [0.00025, 0.001, 0.005, 0.02, 0.1],
       }),
       latencyFromWorker: register.histogram({
         name: "lodestar_bls_thread_pool_latency_from_worker",
         help: "Time from the worker sending the result and the main thread receiving it",
-        buckets: [0.001, 0.003, 0.01, 0.03, 0.1],
+        buckets: [0.00025, 0.001, 0.005, 0.02, 0.1],
       }),
       mainThreadDurationInThreadPool: register.histogram({
         name: "lodestar_bls_thread_pool_main_thread_time_seconds",
@@ -485,32 +514,40 @@ export function createLodestarMetrics(
       }),
       timePerSigSet: register.histogram({
         name: "lodestar_bls_worker_thread_time_per_sigset_seconds",
-        help: "Time to verify each sigset with worker thread mode",
+        help: "Average BLS worker dispatch wall time per original signature set",
         // Time per sig ~0.9ms on good machines
-        buckets: [0.5e-3, 0.75e-3, 1e-3, 1.5e-3, 2e-3, 5e-3],
-      }),
-      totalSigSets: register.gauge({
+        buckets: [0.5e-3, 0.75e-3, 1e-3, 1.5e-3, 2e-3, 5e-3, 10e-3, 20e-3, 50e-3, 100e-3],
+      }),
+      workRequestPreparationDuration: register.histogram({
+        name: "lodestar_bls_thread_pool_work_request_preparation_duration_seconds",
+        help: "Main-thread time spent converting one worker dispatch group to binding inputs",
+        // Normal requests take microseconds; the dashboard treats 1ms as a main-thread stall.
+        buckets: [0.000005, 0.000025, 0.0001, 0.0005, 0.001, 0.005, 0.025],
+      }),
+      verificationCallDuration: register.histogram<{operation: BlsVerificationCallOperation}>({
+        name: "lodestar_bls_thread_pool_verification_call_duration_seconds",
+        help: "Duration of one worker-side lodestar-z verification binding call, including input conversion",
+        labelNames: ["operation"],
+        // Verification ranges from sub-millisecond calls to large batches; above 500ms is uniformly severe.
+        buckets: [0.00025, 0.001, 0.005, 0.02, 0.1, 0.5],
+      }),
+      verificationCallSignatureSets: register.counter<{operation: BlsVerificationCallOperation}>({
+        name: "lodestar_bls_thread_pool_verification_call_signature_sets_total",
+        help: "Count of signature sets passed to worker-side lodestar-z verification binding calls",
+        labelNames: ["operation"],
+      }),
+      totalSigSets: register.counter({
         name: "lodestar_bls_thread_pool_sig_sets_total",
         help: "Count of total signature sets",
       }),
-      prioritizedSigSets: register.gauge({
+      prioritizedSigSets: register.counter({
         name: "lodestar_bls_thread_pool_prioritized_sig_sets_total",
         help: "Count of total prioritized signature sets",
       }),
-      batchableSigSets: register.gauge({
+      batchableSigSets: register.counter({
         name: "lodestar_bls_thread_pool_batchable_sig_sets_total",
         help: "Count of total batchable signature sets",
       }),
-      aggregateWithRandomnessAsyncDuration: register.histogram({
-        name: "lodestar_bls_thread_pool_aggregate_with_randomness_async_time_seconds",
-        help: "Total time performing aggregateWithRandomness async",
-        buckets: [0.001, 0.005, 0.01, 0.1, 0.3],
-      }),
-      pubkeysAggregationMainThreadDuration: register.histogram({
-        name: "lodestar_bls_thread_pool_pubkeys_aggregation_main_thread_time_seconds",
-        help: "Total time spent aggregating pubkeys on main thread",
-        buckets: [0.001, 0.005, 0.01, 0.1],
-      }),
     },
 
     // BLS time on single thread mode
@@ -643,9 +680,15 @@ export function createLodestarMetrics(
         name: "lodestar_sync_unknown_block_downloaded_blocks_error_total",
         help: "Total number of downloaded blocks errors in UnknownBlockSync",
       }),
-      removedBlocks: register.gauge({
+      removedBlocks: register.gauge<{reason: DroppedItemReason}>({
         name: "lodestar_sync_unknown_block_removed_blocks_total",
-        help: "Total number of removed bad blocks in UnknownBlockSync",
+        help: "Total pending blocks dropped from BlockInputSync without completing, by reason",
+        labelNames: ["reason"],
+      }),
+      removedPayloads: register.gauge<{reason: DroppedItemReason}>({
+        name: "lodestar_sync_unknown_block_removed_payloads_total",
+        help: "Total pending payloads dropped from BlockInputSync without completing, by reason",
+        labelNames: ["reason"],
       }),
       elapsedTimeTillReceived: register.histogram({
         name: "lodestar_sync_unknown_block_elapsed_time_till_received",
@@ -908,6 +951,22 @@ export function createLodestarMetrics(
         labelNames: ["error"],
       }),
     },
+    gossipExecutionPayloadBid: {
+      elapsedTimeTillReceived: register.histogram<{source: OpSource}>({
+        name: "lodestar_gossip_execution_payload_bid_elapsed_time_till_received_seconds",
+        help: "Time elapsed between the bid's slot start and the time the execution payload bid was received (negative = received before its slot, broadcast in the previous slot; positive = received during its own slot, late)",
+        labelNames: ["source"],
+        buckets: [-6, -3, -1, 0, 3],
+      }),
+    },
+    gossipPayloadAttestationMessage: {
+      elapsedTimeTillReceived: register.histogram<{source: OpSource}>({
+        name: "lodestar_gossip_payload_attestation_message_elapsed_time_till_received_seconds",
+        help: "Time elapsed between slot time and the time the payload attestation message was received",
+        labelNames: ["source"],
+        buckets: [3, 6, 9, 12],
+      }),
+    },
     // recovery in the case of specific blob rows required
     recoverBlobSidecars: {
       blobsReconstructed: register.counter({
@@ -1175,6 +1234,12 @@ export function createLodestarMetrics(
         name: "lodestar_oppool_attester_slashing_pool_size",
         help: "Current size of the AttesterSlashingPool",
       }),
+      deferredVoluntaryExitPool: {
+        size: register.gauge({
+          name: "lodestar_oppool_deferred_voluntary_exit_pool_size",
+          help: "Current size of the DeferredVoluntaryExitPool",
+        }),
+      },
       proposerSlashingPoolSize: register.gauge({
         name: "lodestar_oppool_proposer_slashing_pool_size",
         help: "Current size of the ProposerSlashingPool",
@@ -1379,6 +1444,16 @@ export function createLodestarMetrics(
         help: "Checkpoint state cache size",
         labelNames: ["type"],
       }),
+      persistentTierEpochs: register.gauge<{tier: number}>({
+        name: "lodestar_cp_state_cache_persistent_tier_epochs",
+        help: "Number of epoch keys retained in each on-disk checkpoint state retention tier",
+        labelNames: ["tier"],
+      }),
+      persistentTierStates: register.gauge<{tier: number}>({
+        name: "lodestar_cp_state_cache_persistent_tier_states",
+        help: "Number of persisted checkpoint states retained in each on-disk retention tier",
+        labelNames: ["tier"],
+      }),
       reads: register.avgMinMax({
         name: "lodestar_cp_state_epoch_reads",
         help: "Avg min max of all state cache items total read count",
@@ -1593,9 +1668,15 @@ export function createLodestarMetrics(
           name: "lodestar_seen_payload_envelope_input_cache_serialized_object_refs",
           help: "Number of serialized-cache object refs retained by cached PayloadEnvelopeInputs",
         }),
-        created: register.counter({
+        created: register.counter<{source: PayloadEnvelopeInputSource}>({
           name: "lodestar_seen_payload_envelope_input_cache_items_created_total",
-          help: "Number of PayloadEnvelopeInputs created",
+          help: "Number of PayloadEnvelopeInputs created by source",
+          labelNames: ["source"],
+        }),
+        pruned: register.counter<{reason: PayloadEnvelopeInputPruneReason}>({
+          name: "lodestar_seen_payload_envelope_input_cache_items_pruned_total",
+          help: "Number of PayloadEnvelopeInputs evicted by reason",
+          labelNames: ["reason"],
         }),
       },
     },
@@ -1705,6 +1786,10 @@ export function createLodestarMetrics(
         name: "lodestar_precompute_next_epoch_transition_waste_total",
         help: "Total number of precomputing next epoch transition wasted",
       }),
+      predictedReorg: register.counter({
+        name: "lodestar_precompute_next_epoch_transition_predicted_reorg_total",
+        help: "Predicted epoch-boundary reorgs where the strong parent was dialed instead of the weak head",
+      }),
       duration: register.histogram({
         name: "lodestar_precompute_next_epoch_transition_duration_seconds",
         help: "Duration of precomputeNextEpochTransition, including epoch transition and hashTreeRoot",

package/src/network/gossip/topic.ts

@@ -414,3 +414,41 @@ export const gossipTopicIgnoreDuplicatePublishError: Record<GossipType, boolean>
   [GossipType.execution_payload_bid]: true,
   [GossipType.proposer_preferences]: true,
 };
+
+/**
+ * Whether a publish that reached zero subscribed peers is an acceptable outcome for a topic.
+ *
+ * Publishing to a topic with no subscribed peers throws `PublishError.NoPeersSubscribedToTopic`, which
+ * propagates to the caller and, for messages submitted through the beacon API, is surfaced to the
+ * validator client. That is the right default: a message that reached nobody is a failed broadcast, and a
+ * VC configured with several beacon nodes can fall back to another one.
+ *
+ * Only topics that have a concrete reason to tolerate it opt out here, ie. are set to `true`, and the
+ * reason is named on the entry. Everything else is `false` and keeps raising. Note this is applied per
+ * publish, a topic set to `false` still honors the global `--network.allowPublishToZeroPeers` flag.
+ */
+export const gossipTopicAllowPublishToZeroPeers: Record<GossipType, boolean> = {
+  [GossipType.beacon_block]: false,
+  [GossipType.blob_sidecar]: false,
+  // data_column_sidecar: we ensure having all topic peers via prioritizePeers(). Even with 0 peers on the
+  // topic the overall publish can still succeed, because a supernode rebuilds and publishes the missing
+  // columns for us, so track sent peers as 0 instead of raising
+  [GossipType.data_column_sidecar]: true,
+  [GossipType.beacon_aggregate_and_proof]: false,
+  [GossipType.beacon_attestation]: false,
+  [GossipType.voluntary_exit]: false,
+  [GossipType.proposer_slashing]: false,
+  [GossipType.attester_slashing]: false,
+  [GossipType.sync_committee_contribution_and_proof]: false,
+  [GossipType.sync_committee]: false,
+  // light_client_finality_update and light_client_optimistic_update: non-mandatory route on most of the
+  // network as of Oct 2022, there may be no peer subscribed to the topic at all. Reason carried over from
+  // the handlers in network.ts, which used to swallow the error instead
+  [GossipType.light_client_finality_update]: true,
+  [GossipType.light_client_optimistic_update]: true,
+  [GossipType.bls_to_execution_change]: false,
+  [GossipType.execution_payload]: false,
+  [GossipType.payload_attestation_message]: false,
+  [GossipType.execution_payload_bid]: false,
+  [GossipType.proposer_preferences]: false,
+};

package/src/network/libp2p/index.ts

@@ -1,3 +1,4 @@
+import {getCiphers} from "node:crypto";
 import {bootstrap} from "@libp2p/bootstrap";
 import {identify} from "@libp2p/identify";
 import type {PrivateKey} from "@libp2p/interface";
@@ -9,12 +10,20 @@ import {Libp2pInit, createLibp2p} from "libp2p";
 import {Registry} from "prom-client";
 import {ENR} from "@chainsafe/enr";
 import {noise} from "@chainsafe/libp2p-noise";
-import {defaultCrypto} from "@chainsafe/libp2p-noise/crypto";
+import {asCrypto, defaultCrypto} from "@chainsafe/libp2p-noise/crypto";
 import {quic} from "@chainsafe/libp2p-quic";
 import {Libp2p, LodestarComponents} from "../interface.js";
 import {NetworkOptions, defaultNetworkOptions} from "../options.js";
 import {Eth2PeerDataStore} from "../peers/datastore.js";
 
+const noiseCrypto = getCiphers().includes("chacha20-poly1305")
+  ? defaultCrypto
+  : {
+      ...defaultCrypto,
+      chaCha20Poly1305Encrypt: asCrypto.chaCha20Poly1305Encrypt,
+      chaCha20Poly1305Decrypt: asCrypto.chaCha20Poly1305Decrypt,
+    };
+
 export type NodeJsLibp2pOpts = {
   peerStoreDir?: string;
   disablePeerDiscovery?: boolean;
@@ -122,7 +131,7 @@ export async function createNodeJsLibp2p(
       listen: localMultiaddrs,
       announce: [],
     },
-    connectionEncrypters: [noise({crypto: defaultCrypto})],
+    connectionEncrypters: [noise({crypto: noiseCrypto})],
     transports,
     streamMuxers: [mplex({disconnectThreshold})],
     peerDiscovery,