Preset SMP relays silently drop all connections from a public SMP relay

未关闭
#2,017 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

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
35/100
Issue 类型
缺陷
描述清晰度
基本清楚
活跃度
活跃
技术栈
haskell
领域
networking

调研方向

The report names no source files or tests; start by reading the relay connection and block-list handling, then compare the reported direct TCP failures with the working Tor connections. Use the packet captures, traceroutes, and timestamps the reporter offers to investigate why these relay addresses are dropped. Done means identifying the cause and resolving or explaining the block.

由索引模型根据 Issue 内容生成。

描述

I run a public SMP relay (stock simplexmq v7.0.1, with an info page that lists operator and complaint contacts). For several days now my relay can't connect to any of the preset SimpleX SMP relays (smp4 to smp19.simplex.im). Every new TCP connection from it is silently dropped. It happens from all the addresses the relay has used, two IPv4 and one IPv6. I emailed and wrote in chat a few days ago but got no reply, so I'm asking here.

The odd part is that it only affects the SMP relays. From the same addresses I can reach ntf2/ntf3, all xftp1-6 servers and the website without any problem. Through Tor all 14 SMP relays work fine. Another server I have at the same hosting provider also reaches all 14 relays over IPv4 and IPv6.

I spent quite some time making sure it's not my side. I captured packets on the network interface before the firewall. The SYNs to your relays (for example 23.128.248.112 and .121, on both 5223 and 443) go out and get retransmitted, but nothing ever comes back: no SYN-ACK, no RST, no ICMP. Connections from the same address to other operators' relays get a SYN-ACK right away. I compared firewall counters before and after the test, and none of my rules touched your addresses; they aren't in any local ban list either. DNS answers match 1.1.1.1 and 9.9.9.9. Routing is normal: right gateway for each source address, no blackhole routes, nothing in mangle/raw/nat. A TCP traceroute to 23.128.248.112:5223 gets to 194.110.60.53, the last hop before the relay, and stops there. From my other server the path goes through the same 194.110.60.53 and the relay answers.

The block also seems to follow the relay's DNS name. When I moved the name to a new IPv4 and IPv6, both new addresses were blocked within about an hour. The old IPv4 started working again roughly 1.5 hours after the name moved away from it, and then about a day later it was blocked again (other names in the same domain still point there). Right now none of the three addresses can reach any of the 14 SMP relays.

For now I've set socks_mode: always, so private routing from my relay to yours goes through Tor and messages still get delivered. But it's slower and depends on Tor.

If the preset relays use a server block list (like the one described in the moderation RFC from December 2024), I'd really like to know whether my relay is on it and why. Nobody has sent me any complaint, and if there is a report about content on my server I'll deal with it right away. More generally, it would help operators to have some way to find out about a block and appeal it.

I can share the captures, traceroutes and exact timestamps privately if that helps.

主要语言
Haskell
星标
865
派生
125
平均合并
1 天 20 小时
30 天内合并 PR
64

环境准备

  • 提供 Dockerfile 或 Docker Compose 文件
  • 没有 Pull Request 模板
  • 阅读贡献指南

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。

simplex-chat/simplexmq 的其他 Issue

查看 simplex-chat/simplexmq 的全部 Issue

相似的 Issue

更多 Haskell Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。