On my Linux server, I found abnormal memory usage and very large TCP in-use socket statistics in /proc/net/sockstat. The Linux version is 4.19 and the TCP congestion control algorithm is BBR without fq. The 'ss' command didn't list so many sockets, so it could be socket leak.
confirm from upstream: - https://lore.kernel.org/all/CANn89i+7-wE4xr5D9DpH+N-xkL1SB8oVghCKgz+https://lore.kernel.org/netdev/76fb2bc3-9f9c-cf5c-02f5-3a5da2312ad1@gmail.com/T/ - https://lore.kernel.org/all/20200602080425.93712-1-kerneljasonxing@gmail.com/ fix in Upstream: https://github.com/torvalds/linux/commit/0a70f118475e037732557796accd0878a00fc25a fxi in Anolis: https://gitee.com/anolis/cloud-kernel/commit/bed537da691b
This bug has been fixed in ANCK-4.19 release 4.19.91-25.7, as following commit: https://gitee.com/anolis/cloud-kernel/commit/bed537da691b