speeding up failover

[Will Glass-Husain]

Hi,

(running 1.5dev19)

I have a load balanced back end.  I want to have fast failover response to
a bad server (e.g. < 30 seconds) but allow for long timeouts for other
requests.

Here's an excerpt from my haproxy.cfg.   When my web server hangs, it seems
to take a long time before haproxy marks that server as down and switches
to the other one.  Is it using the srvtimeout for the status request?

defaults
...
   maxconn        2000
   contimeout     10000
   clitimeout     180000
   srvtimeout     180000


backend mybackend
    option httpchk OPTIONS /status
    stick-table type string len 40 size 5M expire 120m peers balancers
    stick store-response set-cookie(SIMULATE_STICKY_SESSION) table mytable
    stick on cookie(SIMULATE_STICKY_SESSION) table mytable
    stick on url_param(SIMULATE_STICKY_SESSION,;) table mytable

    server app1 10.0.2.99:8080  cookie app1 check inter 10000
    server app2 10.0.3.99:8080  cookie app2 check inter 10000


Thanks in advance, WILL