Hi, I noticed with the output of ss -ti (example):
ESTAB 0 0 YYYYYY:40482 XXXXXX:imaps cubic wscale:3,7 rto:228 rtt:27.02/9.695 ato:40 mss:1400 pmtu:1500 rcvmss:1400 advmss:1448 cwnd:10 bytes_sent:1132 bytes_acked:1133 bytes_received:53005 segs_out:41 segs_in:57 data_segs_out:10 data_segs_in:49 send 4.15Mbps lastsnd:66520 lastrcv:66496 lastack:66496 pacing_rate 8.29Mbps delivery_rate 594kbps delivered:11 app_limited busy:288ms rcv_rtt:17.25 rcv_space:25313 rcv_ssthresh:128820 minrtt:18.859 rcv_ooopack:1 snd_wnd:6136 rcv_wnd:128896
that for every tcp socket, the kernel calculates and maintains a *pacing_rate* even if TCP pacing is turned off (which is the default). I could observe this on several servers and many different linux versions too.
Why does TCP do this? Isn't that a waste of memory and processing time to mantain when the default is not to use it?
Thanks, -- Richard _______________________________________________ Kernelnewbies mailing list Kernelnewbies@kernelnewbies.org https://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
