Hacktoberfest 2026：维护者为十月标记出来的 issue，仍然开放、适合新手。 浏览 Hacktoberfest issue

WebSocket API: signature user data subscription is lost after reconnect

未关闭
  • 哪里不对？
#578 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

还没有人认领这个 Issue。

评估

难度
4/5
预计耗时
3-5 天
新手友好度
45/100
Issue 类型
缺陷
描述清晰度
描述清楚
活跃度
活跃
技术栈
python
领域
backend, networking

调研方向

Start with common/src/binance_common/websocket.py and reconnect() around line 916, then follow how userDataStream.subscribe.signature stores subscription state and how callbacks are keyed by subscriptionId. Reproduce with the described local aiohttp WebSocket server and the failing pytest mentioned in the issue. Done means signature subscriptions are restored after reconnect, callbacks use the new subscription ID, and resubscribe failures emit an error.

由索引模型根据 Issue 内容生成。

描述

Summary

A user data subscription made with userDataStream.subscribe.signature (no session.logon) is not restored when the WebSocket API connection reconnects, including the planned reconnect before the 24h limit and after serverShutdown. The bot stops receiving execution reports and balance updates, and no event or error is raised.

Where

common/src/binance_common/websocket.py, reconnect() (line 916 at a54ec0b): user streams are only resubscribed inside if connection.session_logon_request and self.configuration.session_re_logon. A signature subscription stores no logon request, so nothing is sent on the new connection, and user_stream_connections still points to the closed one.

Reproduction

Local aiohttp WebSocket server, no Binance network:

  1. userDataStream.subscribe.signature on a WS API client in single mode; the server replies with subscriptionId: 0.
  2. The server sends serverShutdown (or the reconnect timer fires).
  3. The client opens connection 2, but the server receives no subscribe request on it, and events pushed there never reach the callback.

I have this as a failing pytest and can share it.

Suggested fix

  • Store the subscribe request per subscription, as session_logon_request is stored, and replay it, re-signed with a fresh timestamp, after reconnect.
  • Remap the callbacks to the subscriptionId returned on the new connection instead of assuming it is unchanged. _resubscribe_user_streams on the logon path assumes the same id and does not await the reply.
  • If the resubscribe fails, emit an error event so the application can react.

I'm happy to open a PR along these lines if this approach works for you.

Related, found while testing (each has a failing test, I can open separate issues or PRs)

  • A reply that arrives after asyncio.wait_for(future, 20) timed out raises InvalidStateError in receive_loop (future.set_result on a cancelled future), which ends the loop: the connection stays open but nothing else is received. Fix: skip futures that are done() and pop the id on timeout.
  • In pool mode, subscribe_user_data registers the callback on the next connection in the round robin, not the one that sent the subscribe, so events are dropped.
  • send_request retries signed DELETE (order cancel) after a ReadTimeout; the first request was already delivered, so the retry returns -2011 Unknown order sent. for an order that was cancelled. Retrying only on connect errors and 5xx would avoid it.
主要语言
Python
星标
3k
派生
712
平均合并
1 天 5 小时
30 天内合并 PR
4

环境准备

这个项目没有提供开发容器、Dockerfile 或贡献指南，环境需要你自己搭建：先看它的 README，通用步骤见我们的新手贡献指南。

从这里开始

  1. 先读完整个 Issue，再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库，在一个分支上完成修改。
  4. 提交 Pull Request，并在描述里引用这个 Issue 编号。

binance/binance-connector-python 的其他 Issue

查看 binance/binance-connector-python 的全部 Issue

相似的 Issue

更多 Python Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。