Hacktoberfest 2026 : les issues que les mainteneurs ont marquées pour octobre, ouvertes et accessibles aux débutants. Parcourir les issues Hacktoberfest

DefaultRequestHandler (v2) ignores queue_manager, so the per-task event queue bound (1,024) cannot be configured

Ouverte
  • Quelque chose ne va pas ?
#1,339 2 commentaires 0 réactions 1 personne assignée Voir sur GitHub

Les mainteneurs répondent en général sous 2 jours

@ishymko y travaille déjà.

Depuis le 15/10/2026.

Évaluation

Cette issue n'a pas encore été évaluée.

Description

component: server status: needs review

Summary

Passing a queue_manager to DefaultRequestHandler produces

DeprecationWarning: A queue_manager was passed to DefaultRequestHandlerV2, but it is not used: v2 delegates
event streaming to an in-memory ActiveTaskRegistry, so custom or distributed QueueManager implementations
are ignored. ...

ActiveTask then builds its queues with the module default (EventQueueSource() /
DEFAULT_MAX_QUEUE_SIZE = 1024), and nothing on DefaultRequestHandler, ActiveTaskRegistry or
ActiveTask lets an application change it.

Why it matters

The bounded queue is what gives a streaming server backpressure: when a client connects and stops reading, the
queue fills, the producer (the agent executor) blocks and stops pulling tokens from the model, and a slow
subscriber is evicted. We verified this end to end and it works well. But 1,024 events is a fixed trade-off:

  • a deployment streaming single tokens may want a smaller bound (less memory per stalled client, faster
    eviction); one streaming large chunks may want a different one;
  • with many concurrent tasks the worst-case memory is tasks x 1,024 x event size, which the operator cannot
    tune.

Evidence

A client that opens POST /message:stream and never reads, against an agent producing 300,000 small events with
no delay: the SDK evicts the stalled sink ("queue full at delivery time"), process memory stays flat (about
80 MB), the server keeps answering other callers, and the task ends at its own deadline. So the mechanism is
sound; only its size is fixed.

Suggested change

Accept an optional size where the queues are created, for example DefaultRequestHandler(..., max_event_queue_size: int = DEFAULT_MAX_QUEUE_SIZE) passed to ActiveTaskRegistry and on to
EventQueueSource. Keep the default so nothing changes for existing users.

Environment

a2a-sdk 1.2.1 and 1.2.2, Python 3.11 to 3.13, FastAPI/Starlette routes via create_jsonrpc_routes and
create_rest_routes.

Found while building slow-consumer backpressure tests for a downstream adapter (lemonade-a2a).

Langage dominant
Python
Étoiles
2.2k
Forks
511
Merge moyen
3 j 2 h
PR mergées (30 j)
46

Préparer son environnement

Par où commencer

  1. Lisez l'issue en entier, puis le guide de contribution du projet.
  2. Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
  3. Forkez le dépôt et travaillez sur une branche.
  4. Ouvrez une pull request qui référence le numéro de l'issue.

Autres issues de a2aproject/a2a-python

Toutes les issues de a2aproject/a2a-python

Issues similaires

Plus d'issues Python

Recevez les nouvelles issues par e-mail

Un résumé court des issues GitHub adaptées aux débutants.