First of all, excuse my limited knowledge of haproxy. I haven't had the time yet to thoroughly go over the provided documentation, but this is on the todo list ;)
Second, this is a project i have inherited from a colleague who no longer works for us, so pardon my maybe obvious questions.
The setup is haproxy with keepalived on 2 loadbalancers that serves http traffic to 2 realservers (apache). The problem is that response from the realservers isn't always optimal. The website is a complicated one, yes, but it should normally load in under 5-6 seconds. Sometimes this is well over 20s and sometimes even close to a minute. Also i've noticed that traffic goes back through the loadbalancer, is this normal and/or configurable? Lastly i've noticed a huge amount of errors in the stats page from haproxy:
(name: req, conn, resp)
Frontend: 841921, -, -
webserver1: -, 934, 3041
webserver2: -, 1680, 3059
Backend: -, 133, 6100
I don't (yet) understand what these errors all mean, but if i compare it with the demo stats running on the haproxy main site i do notice a huge difference, so that might have something to do with the above mentioned problems.
I don't expect a solution, but some pointers as to where i could start looking at the problem would be appreciated.
To be complete, here's the haproxy.cfg: (real ip's and passwd hidden)
log 127.0.0.1 local0 notice user haproxy group haproxy nbproc 4 maxconn 32000 defaults log global mode http option httplog option dontlognull retries 3 option redispatch contimeout 5000 clitimeout 50000 srvtimeout 30000 maxconn 32000 listen webservers x.x.x.10:80 log global mode http stats enable stats auth admin:password balance roundrobin option httpclose option forwardfor option httpchk HEAD /check.txt HTTP/1.0 server webs001 x.x.x.11:80 cookie A check server webs002 x.x.x.12:80 cookie B check
Thanks in advance,
Sebastian Received on 2008/04/16 15:44
This archive was generated by hypermail 2.2.0 : 2008/04/16 16:00 CEST