[Bug]: 在200ms延时和15%丢包率网络情况下拉流测试，长时间后出现速度闪崩且无法恢复

[shungo2020]

What happened?

我在局域网中进行长时间拉流测试（使用bbr拥塞控制）
客户端：192.168.6.30
服务端：192.168.6.42 其中服务端网卡使用linux的tc模拟了200ms延时和15%丢包的网络
使用的tc命令：tc qdisc add dev ens37 root netem delay 200ms loss 15%

客户端连接服务器后，服务端不间断向客户端的stream写入数据，客户端打印接收的速度，前面一直维持在20mbps-30mbps左右
在20分钟后，速度突然闪崩到1mbps以下，并且长时间不能恢复，网络状况没有改变，同时多次测试能复现

闪崩过程速度日志：
2024-10-24 05:27:58.180 speed: 14.4mbps/s
2024-10-24 05:27:59.180 speed: 32.6mbps/s
2024-10-24 05:28:00.180 speed: 21.57mbps/s
2024-10-24 05:28:01.180 speed: 14.33mbps/s
2024-10-24 05:28:02.179 speed: 32.37mbps/s
2024-10-24 05:28:03.180 speed: 25.89mbps/s
2024-10-24 05:28:04.180 speed: 23.11mbps/s
2024-10-24 05:28:05.180 speed: 22.11mbps/s
2024-10-24 05:28:06.179 speed: 28.59mbps/s
2024-10-24 05:28:07.181 speed: 25.92mbps/s
2024-10-24 05:28:08.181 speed: 25.14mbps/s
2024-10-24 05:28:09.179 speed: 32.67mbps/s
2024-10-24 05:28:10.179 speed: 31.26mbps/s
2024-10-24 05:28:11.180 speed: 21.83mbps/s
2024-10-24 05:28:12.179 speed: 11.61mbps/s
2024-10-24 05:28:13.181 speed: 9.68mbps/s
2024-10-24 05:28:14.179 speed: 6.9mbps/s
2024-10-24 05:28:15.179 speed: 4.53mbps/s
2024-10-24 05:28:16.179 speed: 1.98mbps/s
2024-10-24 05:28:17.180 speed: 2.26mbps/s
2024-10-24 05:28:18.179 speed: 739.69kbps/s
2024-10-24 05:28:19.177 speed: 1.82mbps/s
2024-10-24 05:28:20.178 speed: 908.10kbps/s
2024-10-24 05:28:21.181 speed: 1.19mbps/s
2024-10-24 05:28:22.179 speed: 711.49kbps/s
2024-10-24 05:28:23.178 speed: 767.63kbps/s
2024-10-24 05:28:24.178 speed: 795.83kbps/s
2024-10-24 05:28:25.180 speed: 355.88kbps/s
2024-10-24 05:28:26.181 speed: 673.89kbps/s
2024-10-24 05:28:27.184 speed: 486.95kbps/s
2024-10-24 05:28:28.180 speed: 206.0kbps/s
2024-10-24 05:28:29.178 speed: 552.22kbps/s
2024-10-24 05:28:30.179 speed: 805.23kbps/s

代码和速度日志在 https://github.com/shungo2020/quic

同时也在同样环境下测试TCP bbr，能一直保持在15mbps左右的速度，不会出现速度突然下降的情况

Steps To Reproduce

Information and Steps to reproduce the behavior.

Relevant log output

No response