[OP Stack Deployment] How to make nodes follow unsafe head

[fgimenez]

Did you check the documentation?

• Yes

Did you check for duplicate questions?

• Yes

Issue Description

Hello! I have a chain running with just a sequencer and a node used as an RPC endpoint. As part of the deployment, I also added op-exporter, and I can see that the node op_blocknumber is consistently behind the sequencer's, in some cases more than 50 blocks, this is how the difference looks like in a specific period:

According to the specs, I understand that the node's metric is showing the safe head as derived by its op-node and the sequencer's is showing the unsafe head, is that correct? If so, is it possible to configure the node to perform unsafe sync? I see some args in op-node that might be related, can be it configured to do so?

Thanks!

Logs

No response

Additional Information

Sequencer op-node args:

--l2=http://localhost:8551 \
--l2.jwt-secret=/scripts/jwt.txt \
--sequencer.enabled \
--sequencer.l1-confs=3 \
--verifier.l1-confs=3 \
--rollup.config=/scripts/rollup.json \
--rpc.addr=0.0.0.0 \
--rpc.port=8547 \
--rpc.enable-admin \
--p2p.disable \
--p2p.sequencer.key="$SEQUENCER_KEY" \
--l1="$L1_RPC_ENDPOINT" \
--l1.rpckind=basic \
--l1.trustrpc \
--metrics.enabled \
--metrics.addr=0.0.0.0 \
--metrics.port=7300 

Node op-node args:

--l1=$L1_RPC_ENDPOINT \
--l2=http://op-geth:8551 \
--rollup.config=/scripts/rollup.json \
--rpc.addr=127.0.0.1 \
--rpc.port=9545 \
--l2.jwt-secret=/scripts/jwt.txt \
--l1.trustrpc \
--l1.rpckind=basic \
--metrics.enabled \
--metrics.addr=0.0.0.0 \
--metrics.port=7300 

Feedback

No response

[sbvegan]

Hey @fgimenez,

I believe the issue is because of --p2p.disable on the sequencer node. Unsafe blocks are propagated to network via the p2p network.

You can try manually adding the replica op-node to the sequencer op-node. I'd also recommend protecting that peer as well. Please refer to our JSON RPC documentation for details.

[fgimenez]

Thanks a lot @sbvegan I had read in the specs that the unsafe sync is done through p2p and tried it before posting this, but didn't seem to work, I have enabled it again to verify and I still see the non-sequencer node a few blocks behind:

The difference seems to be lower and more stable now, is this ok or should the node be tracking the unsafe head more closely?

[sbvegan]

This looks normal. The sequencer op-node will always have the most up-to-date state because its the only component capable of changing the chain state. The p2p nodes would have almot the same state as the sequencer when they are healthy and not lagging behind. Somethings that may cause this lag are: slow disk, slow network, or syncing.

[fgimenez]

Awesome, thanks!

[sbvegan]

Happy to help!

[github-actions[bot]]

Hey @fgimenez, We hope your recent question was resolved to your satisfaction. Your feedback is invaluable and helps us improve our support services. Could you spare a moment to fill out a short feedback survey. Thank you for helping us improve our developer community.