Dear all,
I am asking your opinion to help track an issue I am having with a web
front. The situation is the following:
The same haproxy (1.4.13) is load balancing our web fronts and our
middleware servers (2 servers with Apache + 2 servers with Tomcat) For
business reason, the call to the middleware can be quite long which is
why we increased some timeout values.
The problem is the following: the web pages can be very slow to load
in the front. To be more accurate, the main html loads a few small
files (images, css, js...). And - sometimes, one this downloads takes
a long time, like 4-5 seconds. The browsers are using keep-alive for
these downloads.
However, if I connect directly on the web front without haproxy, it is
much faster.
Here are my settings:
defaults
log global
mode http
option httplog
option dontlognull
retries 3
option redispatch
maxconn 2000
contimeout 20000
clitimeout 500000
srvtimeout 500000
stats hide-version
stats scope .
listen FrontsWeb :80
mode http
stats enable
option httplog
option http-server-close
option persist
option forwardfor
balance roundrobin
server web1 X.X.X.X cookie server01 check inter 3000 rise 2
fall 4 weight 1
server web2 Y.Y.Y.Y cookie server02 check inter 3000 rise 2
fall 4 weight 3
listen MiddleWare :8080
mode http
stats enable
option httplog
option persist
option forwardfor
balance roundrobin
server middle1 X.X.X.X cookie mw01 check inter 3000 rise 2 fall 4
server middle2 Y.Y.Y.Y cookie mw02 check inter 3000 rise 2 fall 4
Do you see any problems in this setup that would cause this behavior ?
Is there a performance reason to switch to 1.4.19 ?
Thanks in advance for the feeback,
Matthieu
