I don’t know if this is helpful or not, but I went in and set the net.ipv4.tcp_ecn=1 instead of 2 and my bufferbloat testing showed an improvement.
What has improved?
It helped to improve the initial spike and there are no more subsequent spikes I see in bufferbloat on the speedtest, as well as the bufferbloat ping times seems to have gone down.
I really need to setup a test environment that doesn’t use dslreports to bench test the bufferbloat.
Pre change
Post change test result
The setting only affects TCP connections. A ping is an ICMP echo request which cannot have changed because of this.