go-eth shows a significant delay in returning latest data about finalized block.

[Mummyhap]

Hello, I am from BlockPI Network. We recently encountered an issue on the OP mainnet.

Geth version: op-node/v1.11.1
op-geth/v1.101500.0

The finalized blocks returned by two requests show significant differences, with the parameter set to 'true' resulting in block heights lagging behind by several hundred to several thousand blocks.

{
        "id": 4,
        "jsonrpc": "2.0",
        "method": "eth_getBlockByNumber",
        "params": ["finalized", true]
}

and

{
        "id": 4,
        "jsonrpc": "2.0",
        "method": "eth_getBlockByNumber",
        "params": ["finalized", false]
}

[Mummyhap]

This is the script running on our node:

serviceName=optimism
serviceCmd="/node/full/optimism/bin/op-geth --datadir=/node/full/optimism/geth --gcmode=full --syncmode=snap --op-network=op-mainnet --state.scheme=path --rollup.sequencerhttp=https://mainnet-sequencer.optimism.io/ --rollup.disabletxpoolgossip=true --rollup.sequencertxconditionalenabled --rpc.gascap=600000000 --rpc.evmtimeout=10s --history.transactions=0 --txpool.globalslots 1000000 --txpool.globalqueue 3000000 --ws --ws.addr=0.0.0.0 --ws.origins=* --http --http.addr=0.0.0.0 --http.api 'eth,net,web3,debug,txpool' --http.vhosts=* --http.corsdomain=* --authrpc.addr=localhost --authrpc.jwtsecret=/node/full/optimism/jwt.txt --authrpc.vhosts=* --port=31360 --http.port=31361 --ws.port=31362 --authrpc.port=31363"

mkdir -p /node/full/optimism/node
cat > /etc/systemd/system/optimism-node.service << EOF
[Unit]
Description=daemon
After=network-online.target

[Service]
User=root
WorkingDirectory=/node/full/optimism/node
ExecStart=/node/full/optimism/bin/op-node --network=op-mainnet --syncmode=execution-layer --l2.enginekind=geth --l1.rpckind=standard --l1=https://ethereum.blockpi.network/v1/rpc/--l1.beacon=https://ethereum-beacon.blockpi.network/rpc/v1/--l2=ws://127.0.0.1:31363 --rpc.addr=127.0.0.1 --rpc.port=31366 --p2p.listen.tcp=31367 --l2.jwt-secret=/node/full/optimism/jwt.txt
Restart=on-failure
RestartSec=5
LimitNOFILE=200000

[Install]
WantedBy=multi-user.target
EOF

systemctl enable optimism-node
systemctl restart optimism-node

[Mummyhap]

We ran OP Sepolia with the same parameters, but encountered no issues.