Fast Confirmation Rule (FCR)

Lighthouse supports the fast confirmation rule since v8.3.0. To enable fast confirmation rule, use the flag --enable-fast-confirmation on the beacon node. It is disabled by default. With fast confirmation, most blocks can be confirmed within 1-2 slots (12-24 seconds) instead of two epochs (~13 minutes).

SSE Event

If you have subscribed to the beacon node Server-Sent-Event stream for fast confirmation, you should see the following:

event:fast_confirmation
data:{"block":"0x335216055caedd9c20f8e4c5a3f314e1fc11f1031a42ffd030f344a7dfb28988","slot":"4046741","current_slot":"4046743"}

where block is the block root of slot, slot is the most recent confirmed block, and current_slot is the wall-clock slot. It is normal to see the fast confirmation event being emitted a few times for the same block/slot.

Logs

When blocks are produced normally, you should see the following debug log every 12 seconds:

DEBUG FCR advanced                                  confirmed: 0xd97ec7d7bfdf842e3b41a136fa4d55cba0586abec9724af517ab9504e75c2d8e, prev: 0xe1ce20d19963d332ba6a084de278fc71e2b0cdaac8c2e43a50ab5a0b0f

where confirmed is the block root of the most recent confirmed slot and prev is the block root of the previously confirmed slot. If there is a skipped slot, the next block will take a longer time to confirm. The above log will not be observed if the beacon node is syncing or the execution engine is syncing or offline, indicating that no new block has been confirmed.

After a restart of Lighthouse, the following log is usually seen:

DEBUG FCR restored a root confirmed before the restart  root: 0xcdfb8947cf88870b6e4721f27f7470f07478644419d6347fa01f871b00b25ca2

At the first epoch boundary after the restart, the following logs are usually seen:

DEBUG FCR fell back to finalized                    prev_confirmed: 0x1f92a52bf914720564776a1a392a7e10ed7cd998141fbcbd708622ba8ff5cc0d, finalized: 0xb8dfdf0b53d01b5197c058bf6b23891a826a5d3f606da195b4b71d4ad78891e1, slot: 4052704, reason: "epoch_too_old"
DEBUG FCR restarted from observed justified         prev_confirmed: 0xb8dfdf0b53d01b5197c058bf6b23891a826a5d3f606da195b4b71d4ad78891e1, justified: 0xd5b68e49a23eced36dc5693374660aaf63d707a88efb0d52363c4d10795034a4, justified_epoch: 126646

The above logs are harmless and expected after a restart.

If the beacon node has been offline for some time (e.g. one hour), when it is back online, you may see the following error log:

ERROR Error running FCR: NodeNotFound(0xb471d7f9b3e760aea6a364dad4987fd033db672ec8f1efdc46a79f2bdd5427be)  current_slot: Slot(4046686), slot: 4046686

The error should go away once the beacon node is back in sync.

Metrics

A Grafana dashboard for FCR is available in the lighthouse-metrics repository.