Bound how far Beacon API requests can advance a state before validating them

Open
#701 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 6 days

Nobody has claimed this yet.

Assessment

Difficulty
4/5
Estimated time
3-5 days
Newbie friendliness
52/100
Issue type
Bug
Clarity
Mostly clear
Activity status
Active
Tech stack
rust

Research direction

Read the implementations and related PRs #663, #671, and #646 for the attester and proposer duties, sync duties, and block publication paths. Trace where each path calls process_slots and how it determines the head, then inspect the relevant tests. Done means each path rejects requests that would advance too far when the head lags, while preserving the stated late or missing boundary block cases.

Written by the indexing model from the issue text.

Description

Several Beacon API paths advance a stored state with process_slots to answer a request, bounded only by the wall clock, not by the head. With the head lagging the clock (syncing, stalled, or behind after downtime) — or with an attacker choosing an old parent — one request can run hundreds of epoch transitions (~0.7–0.9 s each on mainnet, per #632/#633) on a blocking thread. The cache doesn't save it: its key includes the head root, which moves while syncing.

The existing 503 while-syncing guards don't cover this: SyncStatusController reports synced during a beacon catch-up (#636, "still open" #1).

Path PR Advance Status
POST /eth/v1/validator/duties/attester/{epoch}, GET /eth/v{1,2}/validator/duties/proposer/{epoch} #663 head → wall_epoch + 1 merged without a cap ("fix in another PR")
POST /eth/v1/validator/duties/sync/{epoch} #671 head → first epoch of clock_period (a period is 256 epochs) approved, cap requested
POST /eth/v2/beacon/blocks #646 any known parent → block slot, before the signature check cap requested

Proposal

  • Duties (#663): 503 when epoch > head_epoch + 2.
  • Sync duties (#671): only advance when first_epoch(requested_period - 1) <= head_epoch + 1, else 503. This also avoids answering a committee the chain may not produce: sync duties carry no dependent_root, so a VC can't detect a wrong answer.
  • Block publication (#646): 400 when the parent is below the finalized checkpoint or more than a couple of epochs behind the block's slot.

Each keeps the honest case these PRs were written for (a late or missing boundary block) and refuses the expensive, speculative ones. A test per path with the head several epochs behind the clock should assert the refusal.

Dominant language
Rust
Stars
86
Forks
34
Avg merge
9d 20h
Merged PRs (30d)
77

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from lambdaclass/ethlambda

All issues in lambdaclass/ethlambda

Similar issues

More Rust issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.