cerc-io/lighthouse
16
0
Fork 0
Code Issues 1 Pull Requests Packages Projects Releases Wiki Activity

VC: avoid sending fee recipients until just before merge (#3064)

Browse Source 
## Issue Addressed

Presently if the VC is configured with a fee recipient it will error out when sending fee-recipient preparations to a beacon node that doesn't yet support the API:

```
Mar 08 22:23:36.236 ERRO Unable to publish proposer preparation  error: All endpoints failed https://eth2-beacon-prater.infura.io/ => RequestFailed(StatusCode(404)), service: preparation
```

This doesn't affect other VC duties, but could be a source of anxiety for users trying to do the right thing and configure their fee recipients in advance.

## Proposed Changes

Change the preparation service to only send preparations if the current slot is later than 2 epochs before the Bellatrix hard fork epoch.

## Additional Info

I've tagged this v2.1.4 as I think it's a small change that's worth having for the next release
This commit is contained in:
Michael Sproul 2022-03-09 06:36:38 +00:00
parent 267d8babc8
commit 65eaf01942
1 changed files with 30 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

27
validator_client/src/preparation_service.rs
View File

@ -11,6 +11,9 @@ use std::sync::Arc;
use tokio::time::{sleep, Duration};
use types::{Address, ChainSpec, EthSpec, ProposerPreparationData};
/// Number of epochs before the Bellatrix hard fork to begin posting proposer preparations.
const PROPOSER_PREPARATION_LOOKAHEAD_EPOCHS: u64 = 2;
/// Builds an `PreparationService`.
pub struct PreparationServiceBuilder<T: SlotClock + 'static, E: EthSpec> {
validator_store: Option<Arc<ValidatorStore<T, E>>>,
@ -122,15 +125,9 @@ impl<T: SlotClock + 'static, E: EthSpec> PreparationService<T, E> {
let log = self.context.log().clone();
let slot_duration = Duration::from_secs(spec.seconds_per_slot);
let duration_to_next_epoch = self
.slot_clock
.duration_to_next_epoch(E::slots_per_epoch())
.ok_or("Unable to determine duration to next epoch")?;
info!(
log,
"Proposer preparation service started";
"next_update_millis" => duration_to_next_epoch.as_millis()
);
let executor = self.context.executor.clone();
@ -138,6 +135,7 @@ impl<T: SlotClock + 'static, E: EthSpec> PreparationService<T, E> {
let interval_fut = async move {
loop {
if self.should_publish_at_current_slot(&spec) {
// Poll the endpoint immediately to ensure fee recipients are received.
self.prepare_proposers_and_publish(&spec)
.await
@ -145,10 +143,11 @@ impl<T: SlotClock + 'static, E: EthSpec> PreparationService<T, E> {
error!(
log,
"Error during proposer preparation";
"error" => format!("{:?}", e),
"error" => ?e,
)
})
.unwrap_or(());
}
if let Some(duration_to_next_slot) = self.slot_clock.duration_to_next_slot() {
sleep(duration_to_next_slot).await;
@ -164,6 +163,20 @@ impl<T: SlotClock + 'static, E: EthSpec> PreparationService<T, E> {
Ok(())
}
/// Return `true` if the current slot is close to or past the Bellatrix fork epoch.
///
/// This avoids spamming the BN with preparations before the Bellatrix fork epoch, which may
/// cause errors if it doesn't support the preparation API.
fn should_publish_at_current_slot(&self, spec: &ChainSpec) -> bool {
let current_epoch = self
.slot_clock
.now()
.map_or(E::genesis_epoch(), |slot| slot.epoch(E::slots_per_epoch()));
spec.bellatrix_fork_epoch.map_or(false, |fork_epoch| {
current_epoch + PROPOSER_PREPARATION_LOOKAHEAD_EPOCHS >= fork_epoch
})
}
/// Prepare proposer preparations and send to beacon node
async fn prepare_proposers_and_publish(&self, spec: &ChainSpec) -> Result<(), String> {
let preparation_data = self.collect_preparation_data(spec);