[HELP] tproxy make network failed

[spencercjh]

Here is my cfg file.

proxy_ports: [ 8000 ]
rules: # option rule vec
  - target: Request # Request or Response.
    # Stand for target packet to select & take actions.
    # If target is Response & selecting request info such as method or path ,
    # proxy will select request and take actions on Response.
    selector:
      path: /test # option Match path of `Uri` with wildcard matches. Using [wildcard matches](https://www.wikiwand.com/en/Matching_wildcards).
      #    abc://username:[email protected]:123/path/data?key=value&key2=value2#fragid1
      #                                           |--------|
      #                                               |
      #                                             path
      method: GET # option string
      port: 8000
    actions:
      delay: 2s # option Duration

There is a hello-world python server running on 8000 port.

Why is my ssh connection to this machine affected when I start the chaos-tproxy? I can't ssh to the instance until I reboot.

I never saw logs after Uds listener listening on xxx because the ssh connection was closed...

Also I can't access the http server on port 8000 anymore.

[spencercjh]

中文版本：

我的预期是 chaos-tproxy 对 8000 端口的 GET /test 请求做 delay2s 的故障注入。但我发现 setenv_bridge 方法会把整台机器的网络给搞坏了，导致我无法再 ssh 连上它，8000 端口的服务也不可达了。

可能会是什么样的问题呢？

[Hexilee]

It's likely the chaos-tproxy exits by accident (not graceful shutdown), please check the status and logs by connecting to the physical machine.

[spencercjh]

I successfully ran chaos-tproxy on a brand new Debian VM running by vmware fusion and everything worked fine.

Thanks for your reply.

[spencercjh]

我为代码增加了更多的日志，并且使用 nohup 执行并将日志输出到文件中。发现了以下问题。

arp 包发送失败导致机器脱网，似乎是个非常合理的解释？

@Hexilee

[spencercjh]

I added more logging to the code and used nohup to execute and output the logs to a file. The following problem was found.

An arp packet sending failure causing the machine to disappear in the LAN seems like a perfectly reasonable explanation?