Event adapter refactor to be fork based by zilm13 · Pull Request #10361 · Consensys/teku

zilm13 · 2026-02-11T16:14:15Z

PR Description

Fixed Issue(s)

Documentation

I thought about documentation and added the doc-change-required label to this PR if updates are required.

Changelog

I thought about adding a changelog entry, and added one if I deemed necessary.

Note

Medium Risk
Changes core validator timing/scheduling and fork-transition behavior (including new EIP-7732 duties), which could impact block/attestation production timing if incorrect despite added test coverage.

Overview
Refactors validator time-based duty scheduling to be fork-aware by splitting the previous monolithic TimeBasedEventAdapter logic into Phase0TimeBasedEventAdapter and GloasTimeBasedEventAdapter, selected via a new ForkAwareTimeBasedEventAdapter wrapper.

GloasTimeBasedEventAdapter adds EIP-7732 (“Gloas”) scheduling, including the new payload timeliness attestation event and explicit transition behavior across the fork boundary, while TimeBasedEventAdapter becomes an abstract base providing shared scheduling/utilities. All beacon node API entrypoints now instantiate ForkAwareTimeBasedEventAdapter, and new/updated tests cover EIP-7732 scheduling and pre-/post-fork behavior.

^{Written by Cursor Bugbot for commit a0b37cf. This will update automatically on new commits. Configure here.}

tbenr

some suggestions and an additional AI generated test to verify the transition

  @Test
  void shouldTransitionFromPhase0ToEip7732DutiesAcrossForkBoundary() {
    // Spec: minimal config, EIP7732 fork at epoch 1 (slot 8)
    // SLOTS_PER_EPOCH=8, SECONDS_PER_SLOT=6, millisPerSlot=6000
    // genesisTime=100 sec, eip7732 starts at slot 8 = 148 sec
    final UInt64 genesisTime = timeProvider.getTimeInSeconds(); // 100 sec
    when(genesisDataProvider.getGenesisTime()).thenReturn(SafeFuture.completedFuture(genesisTime));

    // Start at mid-slot 5 (133 sec), so next slot is 6 (pre-EIP7732)
    timeProvider.advanceTimeBySeconds(33); // 100 + 33 = 133 sec

    assertThat(eventAdapter.start()).isCompleted();
    asyncRunner.executeDueActionsRepeatedly();
    verifyNoMoreInteractions(validatorTimingChannel);

    // --- Pre-EIP7732 slot 6: duties at 1/3 and 2/3 timing ---

    // Advance to slot 6 + 1/3 (138 sec). Slot 6 start at 136 sec also fires.
    timeProvider.advanceTimeByMillis(5000); // 133000 + 5000 = 138000 ms
    asyncRunner.executeDueActionsRepeatedly();
    verify(validatorTimingChannel).onSlot(UInt64.valueOf(6));
    verify(validatorTimingChannel).onBlockProductionDue(UInt64.valueOf(6));
    verify(validatorTimingChannel, times(1)).onAttestationCreationDue(UInt64.valueOf(6));
    verify(validatorTimingChannel, never()).onAttestationAggregationDue(UInt64.valueOf(6));

    // Advance to slot 6 + 2/3 (140 sec)
    timeProvider.advanceTimeByMillis(2000); // 140000 ms
    asyncRunner.executeDueActionsRepeatedly();
    verify(validatorTimingChannel, times(1)).onAttestationAggregationDue(UInt64.valueOf(6));
    verify(validatorTimingChannel, never()).onPayloadAttestationDue(UInt64.valueOf(6));

    // --- Slot 7: last pre-EIP7732 slot, old duties expire after firing ---

    // Advance to slot 7 + 1/3 (144 sec). Slot 7 start at 142 sec also fires.
    // After attestation fires, nextDue (150000) > expiration (148000) -> expires!
    // Expiration callback schedules new attestation at slot 8 start + 1/4
    timeProvider.advanceTimeByMillis(4000); // 144000 ms
    asyncRunner.executeDueActionsRepeatedly();
    verify(validatorTimingChannel).onSlot(UInt64.valueOf(7));
    verify(validatorTimingChannel, times(1)).onAttestationCreationDue(UInt64.valueOf(7));

    // Advance to slot 7 + 2/3 (146 sec)
    // After aggregation fires, nextDue (152000) > expiration (148000) -> expires!
    // Expiration callback schedules new aggregation + payload attestation at slot 8
    timeProvider.advanceTimeByMillis(2000); // 146000 ms
    asyncRunner.executeDueActionsRepeatedly();
    verify(validatorTimingChannel, times(1)).onAttestationAggregationDue(UInt64.valueOf(7));
    verify(validatorTimingChannel, never()).onPayloadAttestationDue(UInt64.valueOf(7));

    // --- Post-EIP7732 slot 8: fork boundary! Duties at 1/4, 2/4, 3/4 timing ---

    // Advance to slot 8 start (148 sec)
    timeProvider.advanceTimeByMillis(2000); // 148000 ms
    asyncRunner.executeDueActionsRepeatedly();
    verify(validatorTimingChannel).onSlot(UInt64.valueOf(8));
    // No attestation/aggregation/payload duties should have fired yet at slot 8 start
    verify(validatorTimingChannel, never()).onAttestationCreationDue(UInt64.valueOf(8));
    verify(validatorTimingChannel, never()).onAttestationAggregationDue(UInt64.valueOf(8));
    verify(validatorTimingChannel, never()).onPayloadAttestationDue(UInt64.valueOf(8));

    // Advance to slot 8 + 1/4 (149.5 sec = 149500 ms) - EIP7732 attestation timing
    timeProvider.advanceTimeByMillis(1500); // 149500 ms
    asyncRunner.executeDueActionsRepeatedly();
    verify(validatorTimingChannel, times(1)).onAttestationCreationDue(UInt64.valueOf(8));

    // Advance to slot 8 + 2/4 (151 sec = 151000 ms) - EIP7732 aggregation timing
    timeProvider.advanceTimeByMillis(1500); // 151000 ms
    asyncRunner.executeDueActionsRepeatedly();
    verify(validatorTimingChannel, times(1)).onAttestationAggregationDue(UInt64.valueOf(8));

    // Advance to slot 8 + 3/4 (152.5 sec = 152500 ms) - payload attestation (new in EIP7732!)
    timeProvider.advanceTimeByMillis(1500); // 152500 ms
    asyncRunner.executeDueActionsRepeatedly();
    verify(validatorTimingChannel, times(1)).onPayloadAttestationDue(UInt64.valueOf(8));
  }

tbenr · 2026-02-11T16:44:29Z

...nnode/src/main/java/tech/pegasys/teku/validator/beaconnode/Eip7732TimeBasedEventAdapter.java

+          slot);
+      return;
+    }
+    LOG.info("EVENT *** onPayloadTimelinessAttestationDue");


we should remove this noise left from the original :)

tbenr · 2026-02-11T18:48:37Z

...nnode/src/main/java/tech/pegasys/teku/validator/beaconnode/Eip7732TimeBasedEventAdapter.java

+public class Eip7732TimeBasedEventAdapter extends TimeBasedEventAdapter {
+  private static final Logger LOG = LogManager.getLogger();
+
+  private UInt64 genesisTime;


we could have genesisTime in the abstract class

tbenr · 2026-02-11T18:49:46Z

...r/beaconnode/src/main/java/tech/pegasys/teku/validator/beaconnode/TimeBasedEventAdapter.java

+  final ValidatorTimingChannel validatorTimingChannel;
+  final Spec spec;
+  final RepeatingTaskScheduler taskScheduler;


protected?

do we need to extend it over package-private?

tbenr · 2026-02-11T18:50:16Z

...ode/src/main/java/tech/pegasys/teku/validator/beaconnode/ForkAwareTimeBasedEventAdapter.java

+
+public class ForkAwareTimeBasedEventAdapter implements BeaconChainEventAdapter {
+
+  final TimeBasedEventAdapter delegate;


tbenr · 2026-02-11T20:27:55Z

and, obviously, as pointed out by @StefanBratanov , 7732 -> GLOAS

zilm13 · 2026-02-11T23:27:45Z

Addressed all feedback
I've renamed to Gloas.. but it's still a fork where there is no Gloas Milestone

tbenr · 2026-02-12T08:23:38Z

...ode/src/main/java/tech/pegasys/teku/validator/beaconnode/ForkAwareTimeBasedEventAdapter.java

+      final TimeProvider timeProvider,
+      final ValidatorTimingChannel validatorTimingChannel,
+      final Spec spec) {
+    if (spec.isMilestoneSupported(SpecMilestone.EIP7732)) {


EIP7732 ?

tbenr

there are still a lot of references to EIP7732 milestone

oh.. but this is not master... why??

tbenr

LGTM.
nit: some renames on the new test are needed

cursor

Cursor Bugbot has reviewed your changes and found 1 potential issue.

cursor · 2026-02-13T19:02:45Z

...connode/src/main/java/tech/pegasys/teku/validator/beaconnode/GloasTimeBasedEventAdapter.java

+
+    final UInt64 nextSlotStartTimeMillis = secondsToMillis(nextSlotStartTime);
+
+    final UInt64 millisPerSlot = secondsToMillis(secondsPerSlot);


Duplicated initialization logic across fork-specific subclasses

Low Severity

Both GloasTimeBasedEventAdapter.start() and Phase0TimeBasedEventAdapter.start() contain an identical block of initialization: setGenesisTime, computing currentSlot/nextSlotStartTime/secondsPerSlot, scheduling onStartSlot, and converting to millis — plus the same NOTE comment. This duplicated logic could live in the abstract TimeBasedEventAdapter base class (e.g., as a common setup method or template method), reducing maintenance burden and eliminating the risk that a future subclass forgets to call setGenesisTime, which would cause a NPE.

Additional Locations (1)

validator/beaconnode/src/main/java/tech/pegasys/teku/validator/beaconnode/Phase0TimeBasedEventAdapter.java#L36-L52

StefanBratanov · 2026-02-16T17:46:51Z

@zilm13 can we raise this PR against master

zilm13 · 2026-02-18T15:25:50Z

ci is stalled here, i don't understand why
master pr is #10385

Refactor TimeBasedEventAdapter to ForkAwareBasedEventAdapter

a7ed59d

tbenr reviewed Feb 11, 2026

View reviewed changes

zilm13 added 3 commits February 12, 2026 00:12

Rename EIP7732 -> GLOAS in event adapter classes and methods

d4ad822

add test checking transition of scheduled events during gloas

07c9c2e

feedback addressed

4d74b06

zilm13 requested a review from tbenr February 11, 2026 23:27

tbenr reviewed Feb 12, 2026

View reviewed changes

tbenr approved these changes Feb 12, 2026

View reviewed changes

zilm13 added 2 commits February 12, 2026 11:15

clean up test

2d0b4f3

Resurrect CI

a0b37cf

cursor bot reviewed Feb 13, 2026

View reviewed changes


		public class ForkAwareTimeBasedEventAdapter implements BeaconChainEventAdapter {

		final TimeBasedEventAdapter delegate;


		final UInt64 nextSlotStartTimeMillis = secondsToMillis(nextSlotStartTime);

		final UInt64 millisPerSlot = secondsToMillis(secondsPerSlot);

Comments

Conversation

zilm13 commented Feb 11, 2026 • edited by cursor bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Description

Fixed Issue(s)

Documentation

Changelog

Uh oh!

tbenr left a comment

Choose a reason for hiding this comment

Uh oh!

tbenr Feb 11, 2026

Choose a reason for hiding this comment

Uh oh!

tbenr Feb 11, 2026

Choose a reason for hiding this comment

Uh oh!

tbenr Feb 11, 2026

Choose a reason for hiding this comment

Uh oh!

zilm13 Feb 11, 2026

Choose a reason for hiding this comment

Uh oh!

tbenr Feb 11, 2026

Choose a reason for hiding this comment

Uh oh!

tbenr commented Feb 11, 2026

Uh oh!

zilm13 commented Feb 11, 2026

Uh oh!

tbenr Feb 12, 2026

Choose a reason for hiding this comment

Uh oh!

tbenr left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tbenr left a comment

Choose a reason for hiding this comment

Uh oh!

cursor bot left a comment

Choose a reason for hiding this comment

Uh oh!

cursor bot Feb 13, 2026

Choose a reason for hiding this comment

Duplicated initialization logic across fork-specific subclasses

Uh oh!

StefanBratanov commented Feb 16, 2026

Uh oh!

zilm13 commented Feb 18, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

zilm13 commented Feb 11, 2026 •

edited by cursor bot

Loading

tbenr left a comment •

edited

Loading