[OP Stack Deployment] Independent p2p network

[JossDuff]

Did you check the documentation?

• Yes

Did you check for duplicate questions?

• Yes

Issue Description

I want to run my own op-stack based network with a single sequencer and multiple rpc nodes - independent from the optimism network.
I was wondering what combination of flags I have to set in op-node and op-geth to get my own p2p network, independent of the rest of optimism but with node discovery internally. Op-geth keeps attempting to connect to outside nodes. I don't want to turn discovery off completely because I'd like to be able to discover nodes on this new p2p network.

Logs

No response

Additional Information

No response

Feedback

No response

[opfocus]

The information I’ve gathered indicates two aspects:

1. To add a p2p node for the op-node on a replica node, you can refer (with some limitations) to this discussion:
   [General] troubleshooting resources #17 (comment)

You can add the following flags to op-node:

--p2p.static=<sequencer node multiaddr> \
--p2p.listen.ip=0.0.0.0 \
--p2p.listen.tcp=9003 \
--p2p.listen.udp=9003

2. When using snap sync mode, we may need to add p2p peer information to op-geth. You can refer to the following resources:

• https://ethereum.org/en/developers/docs/nodes-and-clients/bootnodes/
• https://geth.ethereum.org/docs/interacting-with-geth/rpc/ns-admin#admin-addpeer

Additional : proxyd is an important RPC request router and proxy used within the OP Stack infrastructure. https://docs.optimism.io/operators/chain-operators/tools/proxyd
If this doesn’t solve your issue, please let me know and we’ll try to seek help from others.

[JossDuff]

Great thank you! This helped the sync a bunch! It's still attempting to attach to other peers outside my chain which worries me a bit but it doesn't seem to be impacting syncing / execution too much.

Also, previously I was trying to sync from L1 and then switch to the p2p network for finalized blocks when it was caught up to the head (head accoring to the l1), but was having timing issues. Is this assumed default behavior or is this not supported?

[opfocus]

It's still attempting to attach to other peers outside my chain which worries me a bit
Your node can recognize blocks from your chain. My understanding is that, in a testnet environment, because your chain ID is not unique, it may receive some blocks that don't belong to your chain. However, it can still distinguish them. In a production environment, make sure to obtain a unique chain ID (you can find guidance on how to do this in certain places).
previously I was trying to sync from L1 and then switch to the p2p network for finalized blocks when it was caught up to the head (head accoring to the l1), but was having timing issues. Is this assumed default behavior or is this not supported?
In theory, we cannot obtain finalized blocks via P2P. The op-node determines finalization status based on the L1 chain. All blocks received via P2P are treated as unsafe blocks until they are verified through L1. It's normal for finalized blocks to lag behind—sometimes by 20 minutes or even up to half an hour.
You should pay attention to the op‑node setting that chooses between execution‑layer sync and consensus‑layer sync—this option is mainly for snap mode. When you use consensus‑layer sync, the op‑node’s P2P configuration is critical, whereas op‑geth’s P2P matters much less. However, when you sync the execution layer in snap mode, op‑geth’s P2P is indispensable.