From 01a13fdddc404ab84e76a8bc0684bf126b96e4df Mon Sep 17 00:00:00 2001
From: matthewkeil <me@matthewkeil.com>
Date: Tue, 29 Oct 2024 13:55:26 +0700
Subject: [PATCH] feat: update metrics and remove stale comment

---
 .../src/chain/bls/multithread/jobItem.ts          | 15 +++++----------
 .../beacon-node/src/metrics/metrics/lodestar.ts   | 10 +++++++---
 2 files changed, 12 insertions(+), 13 deletions(-)

diff --git a/packages/beacon-node/src/chain/bls/multithread/jobItem.ts b/packages/beacon-node/src/chain/bls/multithread/jobItem.ts
index b03e6b17a48..ba636f6d5d6 100644
--- a/packages/beacon-node/src/chain/bls/multithread/jobItem.ts
+++ b/packages/beacon-node/src/chain/bls/multithread/jobItem.ts
@@ -61,16 +61,11 @@ export async function jobItemWorkReq(job: JobQueueItem, metrics: Metrics | null)
         })),
       };
     case JobQueueItemType.sameMessage: {
-      // This is slow code on main thread (mainly signature deserialization + group check).
-      // Ideally it can be taken off-thread, but in the mean time, keep track of total time spent here.
-      // As of July 2024, for a node subscribing to all subnets, with 1 signature per validator per epoch,
-      // it takes around 2.02 min to perform this operation for a single epoch.
-      // cpu profile on main thread has 250s idle so this only works until we reach 3M validators
-      // However, for normal node with only 2 to 7 subnet subscriptions per epoch this works until 27M validators
-      // and not a problem in the near future
-      // this is monitored on v1.21.0 https://github.com/ChainSafe/lodestar/pull/6894/files#r1687359225
-      const timer = metrics?.blsThreadPool.aggregateWithRandomnessMainThreadDuration.startTimer();
-      const {pk, sig} = await asyncAggregateWithRandomness(job.sets.map((set) => ({pk: set.publicKey, sig: set.signature})));
+      metrics?.blsThreadPool.aggregateWithRandomnessSets.inc(job.sets);
+      const timer = metrics?.blsThreadPool.aggregateWithRandomnessJobTime.startTimer();
+      const {pk, sig} = await asyncAggregateWithRandomness(
+        job.sets.map((set) => ({pk: set.publicKey, sig: set.signature}))
+      );
       timer?.();
 
       return {
diff --git a/packages/beacon-node/src/metrics/metrics/lodestar.ts b/packages/beacon-node/src/metrics/metrics/lodestar.ts
index f15e195faa2..477e6913304 100644
--- a/packages/beacon-node/src/metrics/metrics/lodestar.ts
+++ b/packages/beacon-node/src/metrics/metrics/lodestar.ts
@@ -497,9 +497,13 @@ export function createLodestarMetrics(
         name: "lodestar_bls_thread_pool_batchable_sig_sets_total",
         help: "Count of total batchable signature sets",
       }),
-      aggregateWithRandomnessMainThreadDuration: register.histogram({
-        name: "lodestar_bls_thread_pool_aggregate_with_randomness_main_thread_time_seconds",
-        help: "Total time performing aggregateWithRandomness on main thread",
+      aggregateWithRandomnessSets: register.gauge({
+        name: "lodestar_bls_thread_pool_aggregate_with_randomness_libuv_sets_total",
+        help: "Number of sets being asyncAggregateWithRandomness on libuv thread",
+      }),
+      aggregateWithRandomnessJobTime: register.histogram({
+        name: "lodestar_bls_thread_pool_aggregate_with_randomness_libuv_job_time_seconds",
+        help: "Total time performing aggregateWithRandomness on libuv thread",
         buckets: [0.001, 0.005, 0.01, 0.1],
       }),
       pubkeysAggregationMainThreadDuration: register.histogram({