Add Peering Simulator

[macladson]

Issue Addressed

#6109

Proposed Changes

Adds a new simulator to CI which launches with 2 beacon nodes and every 2 epochs adds a new BN to the network. After 16 epochs it checks to see if all new BNs are synced to the head.

Additional Info

This might be overkill in the long term, but hopefully will be useful for DAS testing and CI.

[macladson]

From my initial local testing, if I disable the eip7594_fork_epoch, the nodes all sync fine and the simulation exits successfully. However if I leave in the eip7594_fork_epoch, the nodes consistently get stuck despite seemingly having peers.

[jimmygchen]

However if I leave in the eip7594_fork_epoch, the nodes consistently get stuck despite seemingly having peers.

Nice! that useful finding - will look into this tomorrow.

[jimmygchen]

Hi @macladson

I haven't been able to get to this, do you have some capacity to look into the issues? It would be useful to complete this change, as we're now picking up on PeerDAS development speed again.

The test will need a bit of update now we've switched to activating PeerDAS at the Fulu hard fork. Let me know if you're busy with something else and I can look into this.

[macladson]

So I've updated the code and rebased onto unstable. It looks like syncing still gets stuck. Simulator logging isn't working properly since the tracing migration but I seem to be seeing a lot of these logs:

Apr 09 20:14:40.606 WARN Could not publish message                     error: NoPeersSubscribedToTopic, kind: data_column_sidecar_125, service: "network"

I don't know PeerDAS enough to know for sure, but maybe in the simulator there aren't enough nodes to properly serve all subnets?

[ackintosh]

I'll take a look into this as well!

[ackintosh]

I created macladson#7. peering-sim has run successfully in my local tests, though NoPeersSubscribedToTopic still appears.

[ackintosh]

https://github.com/sigp/lighthouse/actions/runs/14413837149/job/40430090447?pr=6395

Created macladson#8 to fix the CI failure. 🙏