twoeths

for the record, `runHeartBeat()` takes 3.6% of cpu time on a lodestar's subscribeAllSubnets node, should take another profile when all of the improvements are merged [0522_contabo_20_subscribe_all_subnets.cpuprofile.zip](https://github.com/ChainSafe/js-libp2p-gossipsub/files/8905439/0522_contabo_20_subscribe_all_subnets.cpuprofile.zip)

need to see if this still happens in the latest release (4.0.0 as of Aug 2022)

Heartbeat duration is improved significantly (v0.14.0 vs v4.1.1)

This metric is incredible in v4.1.1

unstale

need to retest after #331

The metric is so nice in v4.1.1 (viewed in a 12h timeframe, new libp2p/gossipsub is deployed in the last 7h)

it looks like this is the beacon node's I/O lag issue, not the validator issue: ``` Apr-12 06:15:01.948[] ^[[36mverbose^[[39m: Found new chain head slot=2769075, head=0x4037009ea6dccfa854e2141bfa7aa4b4305d1d6614632475d79c09f49625381d, previouDuty=0x71b160050d26e63fe913fe3b090a9677ac0bc33139e1557e7a3d9dd4f160700b, currentDuty=0x60ea055172b06f1766fe4da73475b0ce5c178c211cfd7818ef87e59b3c17df2e Apr-12 06:15:01.949[] ^[[32minfo^[[39m:...

root cause is #4002

This shows that there are >100 messages in queue but there are no new jobs start running/finish as in the gap also the debug shows that at sometimes, there are...