[KaloyanTanev]

Payload timeliness committee members should attest as soon as the payload and its blob data are seen, but only report either as missing once PAYLOAD_ATTESTATION_DUE_BPS (9s) has elapsed, since blob data may still arrive (spec). Charon fetched once at the payload due time (6s), locking blob_data_available=false in early, and Prysm's 503 before 9s failed the duty.

The fetcher now fetches again at the due time unless the first fetch reports both the payload and its blob data. PTC duties use the eager_ahead_split timer like gloas attesters, triggered at 6s with round 1 spanning to 10s, so a leader waiting for 9s needs no round change.

category: bug
ticket: #4324

[codecov]

Codecov Report

❌ Patch coverage is 97.14286% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 67.61%. Comparing base (957f2dd) to head (51cece3).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
core/fetcher/fetcher.go	96.29%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4797      +/-   ##
==========================================
+ Coverage   67.59%   67.61%   +0.02%     
==========================================
  Files         254      254              
  Lines       33589    33608      +19     
==========================================
+ Hits        22705    22725      +20     
+ Misses      10883    10882       -1     
  Partials        1        1              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[OisinKyne]

Things to note

1. Rolling-upgrade window. Round end times have to match across the cluster, and this changes the payload-attestation timer for every node. While old and new versions are mixed, those duties will likely round-change or fail. Ship it in the build everyone runs before the Hoodi fork epoch (Oct 26/27); Sepolia clusters will see it during their upgrade.
2. One refetch means waiting on the leader's beacon node. If the leader's beacon node hasn't seen the blobs at 6s, the cluster waits until 9s, even when every other node saw them at 7s. Polling every 250 to 500ms between 6s and 9s (or using the execution_payload_available event later) would let most slots attest early. It's a small change inside the same function.
3. Votes close to the 9s due time. In the fallback case, the message goes out at roughly 9.2 to 9.8s: consensus, then VC signing, partial-signature exchange and aggregation, all after 9s. That's inherent, since the spec doesn't allow reporting "missing" before 9s. It's still in time for the next proposer at 12s, and the Hoodi PTC metrics will show what it costs.
4. Side effects of moving the slot offset from 6s to 9s.
   • The duty's expiry deadline (core/deadline.go) moves 3s later, which is more lenient and fine.
   • core_bcast_delay_seconds{duty="payload_attestation"} will now record negative delays for early (6s) broadcasts. That's cosmetic, but worth knowing before anyone builds PTC dashboards from that metric.
5. Nit: when the first fetch errors, the debug log prints no_block from a zero-value struct. Harmless.