Due to the pectra blob schedule bug, sepolia and holesky chains with a sequencer running op-node/v1.11.2 or earlier, while L1 activated Pectra, were running into a bug where the blob base fee of the L1 Info block wasn't correctly calculated. See release notes of op-node/v1.12.0 for details.

We have since then released op-node v1.12.0 and v1.12.1 with a fix included. However, that fix needs to be explicitly scheduled. Node operators just upgrading from 1.11.2 to one of these releases without their chain having the fix scheduled (as part of the SR) would chain-split. The fix always has to be scheduled. This is being warned about at the top of the releases, but node operators might still miss it.

So I proposer that we add a kill-switch to op-node, where at startup it checks

• if the chain's L1 is Sepolia or Holesky
• if the pectra block schedule fix does not have a timestamp set
• if the L2 genesis timestamp is before the L1's activation of Pectra

If all of those conditions are true, the node exits with a warning explaining that this chain probably needs to schedule the fix. The warning can be deactivated with a hidden flag to force the node to start without the fix scheduled.