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

Native ICE TURN failover fails when the servers use different credentials

未关闭
#445 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
3/5
预计耗时
半天
新手友好度
66/100
Issue 类型
缺陷
描述清晰度
描述清楚
活跃度
活跃
领域
networking

调研方向

Start at STUNRequestCallback_AllocateRelay in src/ice.cpp and trace where m_strTURNRealm, m_strTURNNonce and m_arrTURNKey are set during the 401 challenge and read when building the next Allocate. First fix is resetting those members before advancing to the next server; done means a second TURN server with different realm/credentials is reached with its own key after the first times out. Then evaluate the second part of the report: whether non-timeout error responses should also advance the server list instead of marking relay failed, and check existing TURN/ICE tests or a two-server repro for coverage.

由索引模型根据 Issue 内容生成。

描述

I think I found a problem with TURN failover in the native ICE client.

If the first TURN server answers the Allocate with a 401 and then stops responding, the client times out and moves on to the second server. But it seems to keep the realm, nonce and key from the first server (m_strTURNRealm, m_strTURNNonce, m_arrTURNKey), so the request to the second server has the right username but the wrong realm and key. The second server answers 401, the client treats that as wrong credentials, and relay is marked as failed.

To reproduce, configure two TURN servers with different credentials or realms, where the first one answers the initial 401 challenge but drops the authenticated Allocate. With peers that can only connect through a relay, the connection fails. If the first server never answers at all, failover works fine.

Clearing m_strTURNRealm and m_strTURNNonce before trying the next server in STUNRequestCallback_AllocateRelay fixed it for me.

Related: the second server is only tried after a timeout. If the first server rejects the request, for example because of a wrong password, relay is marked as failed and the second server is never contacted. That means a backup server only helps when the first one is down, not when it is up and refusing us. Shouldn't an error response also fall through to the next server?

主要语言
C++
星标
10k
派生
747
平均合并
5 天 19 小时
30 天内合并 PR
1

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。

ValveSoftware/GameNetworkingSockets 的其他 Issue

查看 ValveSoftware/GameNetworkingSockets 的全部 Issue

相似的 Issue

更多 C++ Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。