Hold a queued gossip verdict until its object settles, so a block imported moments later is still relayed
Les mainteneurs répondent en général sous 6 jours
Personne n'a encore pris cette issue.
Évaluation
- Difficulté
- 4/5
- Temps estimé
- 3-5 jours
- Accessibilité débutants
- 44/100
- Type d'issue
- Fonctionnalité
- Clarté
- Clairement spécifiée
- Activité
- Active
- Stack technique
- rust
Piste de recherche
Start with crates/net/p2p/src/beacon/verdict.rs and the gossipsub_config in crates/net/p2p/src/lib.rs, then trace the chain actor’s block import and envelope verification events alongside the API in crates/net/api. Check how stateful_checks, settle, and the seen cache work for queued messages. Done means a test queues a gloas block, verifies its parent envelope inside the gossip window, and confirms its verdict is reported as Accept exactly once; include bounded holds, expiry handling, and metrics.
Rédigé par le modèle d'indexation à partir du texte de l'issue.
Description
Problem. A beacon gossip message whose stateful checks answer Outcome::Queue is reported to gossipsub as a final IGNORE straight away (disposition in crates/net/p2p/src/beacon/verdict.rs:566, Outcome::Queue(_) => (MessageAcceptance::Ignore, true) at :569, reported by report at :577) and handed to the chain actor. The actor often imports it moments later, but by then gossipsub has its verdict, so this node never forwards it.
lambdaclass/ethlambda#802 makes this visible for gloas blocks. A block built on a FULL parent whose envelope is not verified yet is now queued (QueueReason::ParentPayloadUnverified, gossip::block::stateful_checks) and held by the actor until the envelope verifies (hold_block_for_parent_payload and release_blocks_awaiting_parent_payload, crates/blockchain/src/beacon_envelope.rs:959 and :1067). In the devnet trace in that PR, the envelope landed 40 ms after the block: the block now imports, but this node, and every other node that hits the same race, is a dead end for it. The block reaches the rest of the mesh only through nodes that did not race.
This applies to every topic that answers Queue, not only that rule: beacon_block (parent_unknown, parent_not_ready, shuffling_unavailable, parent_payload_unverified), data_column_sidecar (parent_unknown, parent_not_ready, shuffling_unavailable, and from gloas block_unknown and block_not_ready) and execution_payload (block_unknown, block_not_ready, state_not_cached).
Why it was out of scope for lambdaclass/ethlambda#802. That PR changes one rule's verdict from IGNORE to queue. This change is about how every queued verdict is reported. It also needs a signal that does not exist yet: nothing tells the p2p actor that a queued object has settled. And an Accept must not rest on the actor's import alone. Whatever gossip rules the queue returned before still have to pass, along with the seen cache. For parent_payload_unverified those are now only the finalized-ancestry walk's queue and the lookahead gate, since lambdaclass/ethlambda#802 moved every REJECT that does not read the parent's payload ahead of it. For parent_unknown they are nearly all of them.
The window. validate_messages() holds every beacon message until it has a verdict, and the message cache keeps it for history_length(6) heartbeats of 700 ms (gossipsub_config, crates/net/p2p/src/lib.rs:917, history at :933), about 4.2 s. A verdict reported after that propagates nothing (report_message_validation_result returns false, counted by lean_beacon_gossip_verdict_expired_total).
What a fix could look like.
- On
Queue, do not report yet. Park theGossipId(message id and propagation source) with the validated object, keyed by the root it waits on, under a deadline comfortably inside the window (say 3 s). Bound the parked set per root and in total. - Add a message from the chain actor to p2p (
crates/net/api) for the events that settle a queued object: a block imported, an envelope verified. On it, re-run the object's stateful checks against the store as it now is. Report the new verdict: Accept records the seen entry throughsettleand propagates; anything else reports as it would today. - At the deadline, report IGNORE, as today.
- Count holds, settles and expiries in metrics, so it is visible whether queued messages settle inside the window.
A test should queue a gloas block on an unverified parent payload, verify the envelope inside the window, and assert that the block's verdict is reported as Accept exactly once.
- Langage dominant
- Rust
- Étoiles
- 86
- Forks
- 34
- Merge moyen
- 9 j 20 h
- PR mergées (30 j)
- 77
Préparer son environnement
- Fournit un Dockerfile ou un fichier Docker Compose
- Propose un modèle de pull request
- Lire le guide de contribution
Par où commencer
- Lisez l'issue en entier, puis le guide de contribution du projet.
- Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
- Forkez le dépôt et travaillez sur une branche.
- Ouvrez une pull request qui référence le numéro de l'issue.
Autres issues de lambdaclass/ethlambda
-
Difficulté 2/5 1-3 heures Accessibilité débutants 76/100
lambdaclass/ethlambda#850 ·
Les mainteneurs répondent en général sous 6 jours
-
Difficulté 2/5 1-3 heures Accessibilité débutants 68/100
lambdaclass/ethlambda#777 · 1 réaction ·
Les mainteneurs répondent en général sous 6 jours
-
Run all gossip validation testsPeut-être à nouveau libre @dicethedev l’a pris il y a 9 jours, et aucune pull request n’est ouverte. Ouverte
Difficulté 2/5 1-3 heures Accessibilité débutants 62/100
lambdaclass/ethlambda#725 · 2 commentaires ·
Les mainteneurs répondent en général sous 6 jours
-
Handle malformed DB dataPeut-être à nouveau libre @dicethedev l’a pris il y a 115 jours, et aucune pull request n’est ouverte. Ouverte
Difficulté 2/5 1-3 heures Accessibilité débutants 76/100
lambdaclass/ethlambda#529 · 4 commentaires ·
Les mainteneurs répondent en général sous 6 jours
-
Document architectural choicesPeut-être à nouveau libre @dicethedev l’a pris il y a 115 jours, et aucune pull request n’est ouverte. Ouverte
Difficulté 2/5 1-3 heures Accessibilité débutants 68/100
lambdaclass/ethlambda#521 · 2 commentaires ·
Les mainteneurs répondent en général sous 6 jours
Toutes les issues de lambdaclass/ethlambda
Issues similaires
-
Difficulté 2/5 1-3 heures Accessibilité débutants 84/100
antithesishq/bombadil#473 ·
Les mainteneurs répondent en général sous 2 jours
-
bench/api.sh exec test fails with curl exit 52, and --ignore-failure still records the timingsOuvertebug
Difficulté 2/5 1-3 heures Accessibilité débutants 72/100
libracoder/kiri#669 ·
Les mainteneurs répondent en général sous 1 jour
-
Difficulté 1/5 Moins d'une heure Accessibilité débutants 88/100
Les mainteneurs répondent en général sous 5 jours
-
autofixer: follow-up
Difficulté 2/5 1-3 heures Accessibilité débutants 75/100
tursodatabase/turso#12217 ·
Les mainteneurs répondent en général sous 1 jour
-
Declare the CMS schema library in the optional recipients extraPeut-être pris Une pull request liée à cette issue est ouverte ou déjà fusionnée. Ouvertearea:python in-progress
Difficulté 2/5 1-3 heures Accessibilité débutants 82/100
yhay81/pyropdf#1005 · 2 commentaires ·
Les mainteneurs répondent en général sous 1 jour