Giters

ObolNetwork / charon-distributed-validator-cluster

Run a Distributed Validator Cluster locally using docker-compose

Home Page:https://docs.obol.tech/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Fix issues with nimbus BN

xenowits opened this issue · comments

commented

Problem to be solved

The nimbus beacon node has some issues:

  • The sync-distance keeps on increasing instead of decreasing.
  • The latency for most of the endpoints is high.

Proposed solution

Spike into the issue and dig into the root cause and fix the issue. Ensure the whole cluster is working properly including synced nethermind.

commented

Is this issue still occuring @xenowits? Should we allocate this bug a severity level? If this repo is broken without this fix we should flag that.

commented

Actually, if I remember correctly, I think Jim McDonald said that the Nimbus REST API runs on a single thread, so maybe that is part of the latency issues. This if true would definitely be unfriendly for this repo that has 6 VCs/DVCs calling it simultaneously

commented

Actually, if I remember correctly, I think Jim McDonald said that the Nimbus REST API runs on a single thread, so maybe that is part of the latency issues. This if true would definitely be unfriendly for this repo that has 6 VCs/DVCs calling it simultaneously

Yeah probably, will dig into more specific details. But, yeah it's probably too much to have 6 VCs asking for data from one single-threaded nimbus BN process

commented

Turned out it was not a performance issue but the real issue was docker needs more resource capabilities. Checkpoint sync works just fine