I hope you can help.
I currently have haproxy 1.7.9 compiled with Tproxy on centos 7.
This is used to load balance 6 proxy servers for internet access on the back end. It is setup as a one arm nat with a 10gb vmnet3 nic. The 6 proxy servers are cisco WSA. keepalived is setup with a vip. IPtables is utilized to encapsulate the traffic to ensure the back end cisco wsa proxy receives the source ip address rather than the haproxy ip address.
We are getting performance/latency issues with users reporting slow internet access etc, this is since the user count has increased, can anyone see or suggest anything on the load balancer that may be causing issues?
config:
global
daemon
log 127.0.0.1 local4
maxconn 40000
ulimit-n 81000
defaults
log global
timeout connect 5000
timeout client 300000
timeout server 300000
mode http
frontend http-in
bind *:80,*:443,*:8021
maxconn 50000
default_backend backend_servers
option forwardfor
#option http-server-close
#option forceclose
backend backend_servers
#option http-server-close
#option forceclose
balance leastconn
stick-table type ip size 20240k expire 5m
stick on src
server edu-wsa01 10.80.10.111:80 check inter 30000 fall 3
server edu-wsa02 10.80.10.112:80 check inter 30000 fall 3
server edu-wsa03 10.80.10.113:80 check inter 30000 fall 3
server edu-wsa04 10.80.10.114:80 check inter 30000 fall 3
server edu-wsa06 10.80.10.116:80 check inter 30000 fall 3
server edu-wsa07 10.80.10.117:80 check inter 30000 fall 3
#server web2 10.80.10.248:80 check inter 30000 fall 3
#server web3 10.80.10.247:80 check
source 0.0.0.0 usesrc clientip
Below this is just stats config on port 7000
I was looking to do direct server return however i am having issues with the WSA supporting loopback.