HAProxy community

Confused by stats page

Looking for some criticism/critique on my setup as it’s me and the walls here… Also trying to make sense of what I’m seeing on the stats page… Hopefully I’m not shooting myself in the foot somehow…

Stats page: https://imgur.com/a/u03CbsT

egrep -v “^$|^#” /etc/haproxy/local-haproxy.cfg
maxconn 32768
log local0 info
user haproxy
group haproxy
nbthread 2
stats socket /run/haproxy.sock # mode 660 level admin
pidfile /run/haproxy.pid
profiling.tasks auto
tune.maxaccept -1
server-state-file /var/cache/haproxy/state
maxconn 32768
backlog 16384
mode http
log global
http-reuse safe
option httplog
option dontlognull
option dontlog-normal
option tcp-smart-accept
option tcp-smart-connect
timeout connect 1024s
timeout client 10240s
timeout server 10240s
option http-keep-alive
option redispatch
option clitcpka
option http-no-delay
option logasap
balance static-rr
listen stats
stats enable
bind :9999 tfo
stats show-node
stats uri /stats
stats refresh 30s
frontend proxy
backlog 32768
bind tfo # tfo alpn h2,http/1.1
use_backend proxy_pool
backend proxy_pool
retry-on all-retryable-errors
server proxypy tfo check # send-proxy
server privoxy tfo check # send-proxy
server squid tfo check # send-proxy

running haproxy out of runit with a finish script to dump the stats…

the run file fwiw:

chpst -o 65536
chpst -e env
/usr/bin/haproxy -W -f /etc/haproxy/local-haproxy.cfg

I’m initially confused with the stats of privoxy vs squid if it’s in static-rr, Session rate Max looks balanced… but Sessions current and Sessions Max look very different from each other… What does that mean (in my case) of where I’m using haproxy in front of two forward proxies…

the haproxy machine is Archlinux with 4.19.72-1-lts kernel as is the machine. This machine is virtual, 0.33 is physical. I also have a tweaked ‘rc.local’ (Arch people seem to hate the ‘old ways’…)

I can share if it might be anything applicable…

Also on the frontend portion of the stats… does Max 60 mean 60 unique users/ips?

Open to questions or criticism…

Thanks in advance.