yorickdowne

Were you able to see something interesting in these traces? If there’s no way to find root cause even with trace logs, we can close the issue and just conclude...

We have metrics in Grafana. I am not sure of our retention time, but they should still be there. Which metrics would you like to see to narrow this down?

Here's peers and process_open_fds. I am not sure what to query to display mesh health and fanout health. process_open_fds just climbs steadily over these ~7 days. It wasn't running for...

We ran Nimbus to find a performance issue during sync committee and found this one instead. We are back on Lodestar. We are happy to test this again once the...

Will update Nimbus and turn on debug logs. Hopefully that sheds more light. If there's no clue, we can go to trace.

Delay is better by a good second after restart ``` eth-goerli-consensus-1 | INF 2023-08-19 16:53:49.695+00:00 Sync committee message sent message="(slot: 6329669, beacon_block_root: \"b8c7ba25\", validator_index: 394012, signature: \"8a5be233\")" delay=-2s304ms129us161ns eth-goerli-consensus-1 |...

Initial debug logs after restart on `23.7.0` [nimbus-debug.log.gz](https://github.com/status-im/nimbus-eth2/files/12386337/nimbus-debug.log.gz)

Checking participation on Goerli overall ... with this many blocks missing, we may not find a cause. 76% is in line with what you'd expect on a network with

Vouch 1.7.5. See first post :) Fellow NOs opine that Vouch in particular causes Nimbus to struggle with sync committee, whereas Vouch does not cause sync committee issues with Teku,...

You’re looking at the Goerli validator. Let’s get back to the original mainnet use case. I have validators running for Lido. Vouch is the VC, and then there are three...