Construct and sign the Gloas SignedExecutionPayloadEnvelope, and broadcast it to beacon nodes by 0w3n-d · Pull Request #503 · gattaca-com/helix

0w3n-d · 2026-08-24T18:29:20Z

Issue: #489 (step 2 of 6)

Stacked on #501 — base branch is od/gloas-spec-link-cleanup, not develop. Retarget to develop once #501 merges.

What this PR does

BeaconClient/MultiBeaconClient::publish_execution_payload_envelope: POSTs a SignedExecutionPayloadEnvelope to a connected beacon node's /eth/v1/beacon/execution_payload_envelopes, mirroring publish_block's fan-out-and-return-on-first-success.
construct_signed_envelope: builds and signs the envelope from a proposer's signed block and a held payload, guarding against a missing held payload, a held-payload/bid block-hash mismatch, and a bid committing to a different builder_index than helix's configured identity.
GloasBuilderIdentity: helix's own Gloas builder identity (builder_index + key), signing under DOMAIN_BEACON_BUILDER — not the off-chain domain SignedBuilderBid uses. Reuses the relay's existing signing key; builder_index is a new RelayConfig field, defaulting to 0 as a placeholder until helix has a real on-chain builder registration.
Wires all of the above into submitSignedBeaconBlock: decodes the block, constructs and signs the envelope, and broadcasts it via MultiBeaconClient.
NoHeldPayloads: the GloasPayloadStore implementation used until step 3 lands a real one. Every real request currently gets a NoHeldPayloadForBlock error, which is correct: nothing has held a payload yet.

What this PR deliberately does not do

No real held-payload store yet (step 3). No support for external builders submitting their own signed bids/envelopes (deferred, see the issue's resolved design question).

Tests

Written before wiring, per the issue's tests-first convention:

BeaconClient::publish_execution_payload_envelope: correct headers/body, 200/202/error-response handling (httpmock).
MultiBeaconClient::publish_execution_payload_envelope: returns on first success, errors only if all clients fail.
construct_signed_envelope: correct construction, signature round-trip verification, and all three guard conditions.

Reviewer checklist

CI (lint, unit-test) is green
Matches the linked issue/step
No unexplained scope creep or unrelated files touched

ninaiiad · 2026-09-25T10:33:10Z

+/// Payloads held by a bid's committed block hash. Each payload is taken at most once.
+// TODO(gloas): populate from the auctioneer; see gattaca-com/helix#489 step 3.
+#[derive(Default)]
+pub struct GloasPayloadStore(DashMap<B256, HeldGloasPayload>);


why a dashmap?

This will be addressed by #635

ninaiiad · 2026-09-25T10:51:04Z

    pub blacklist_provider: Option<Url>,
+    /// This relay's on-chain Gloas (ePBS) builder_index. Placeholder until helix has a real
+    /// on-chain builder registration; signs under the relay's own key in the meantime.
+    #[serde(default)]


should this really have a default?

This will be addressed by #635

ninaiiad · 2026-09-25T10:56:16Z

+    /// success. Unlike `publish_block`, fans out via plain concurrent futures, not
+    /// `spawn_tracked!`.


The fan-out uses plain futures inside the axum handler: if the proposer disconnects, or the 5s TimeoutLayer fires, the handler future is dropped and every in-flight publish is cancelled. This is exactly why publish_block uses spawn_tracked!, and here the payload is already gone when it happens.

ninaiiad · 2026-09-25T10:59:24Z

    pub reg_handle: RegWorkerHandle,
    pub operator_api: Option<Arc<OperatorPubSub>>,
    pub ip_tracker: IpTracker,
+    pub gloas_builder_identity: Arc<GloasBuilderIdentity>,


don't need an arc here because it's not shared?

…dcast it to beacon nodes

…velope

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

This was referenced Aug 24, 2026

Gloas (ePBS) builder-API support #489

Open

Wire getExecutionPayloadBid to the auctioneer, checking parent_hash/parent_root #507

Merged

Wire submitBuilderPreferences to per-proposer per-slot storage on the auctioneer #509

Merged

0w3n-d force-pushed the od/gloas-step2-envelope-broadcast branch from 70da680 to 92bd1e1 Compare August 25, 2026 12:56

0w3n-d force-pushed the od/gloas-step2-envelope-broadcast branch 2 times, most recently from 92bd1e1 to 6c3aca0 Compare August 27, 2026 18:06

ninaiiad reviewed Sep 3, 2026

View reviewed changes

Comment thread crates/common/src/config.rs

ninaiiad reviewed Sep 3, 2026

View reviewed changes

Comment thread crates/relay/src/api/proposer/submit_signed_beacon_block.rs Outdated

Base automatically changed from od/gloas-spec-link-cleanup to develop September 3, 2026 17:14

0w3n-d force-pushed the od/gloas-step2-envelope-broadcast branch 2 times, most recently from e9a8df7 to 552e72c Compare September 3, 2026 17:34

0w3n-d force-pushed the od/gloas-step2-envelope-broadcast branch from e036e47 to 8379f97 Compare September 25, 2026 09:56