On Tue, May 15, 2012 at 8:48 PM, Foreman, Tim <[email protected]> wrote: > Using an 'option httpchk' service check, what happens to the existing > connections to the server when it is marked down? > > We are trying to take servers out of the backend by making them fail the > service check and it appears to be taking more than 120 seconds to be marked > down on occasion. > > We are running v1.4.15 and this is my config: > > global > log 127.0.0.1 local0 > log 127.0.0.1 local1 debug > log-send-hostname prodlb01-dal05 > maxconn 4096 > daemon > > defaults > log global > option dontlognull > retries 3 > option redispatch > maxconn 2000 > contimeout 5000 > timeout client 60s > timeout server 60s > timeout check 5s > > listen cae-preview-servers > bind xx.xx.xx.xx:80 > default_backend caepreview_http > > backend caepreview_http > mode http > option httplog > stick on src table caepreview_https > option httpchk GET /base/monitor/status HTTP/1.1\r\nHost:\ monitoring > server prodcaepreview01-dal05 10.41.135.140:80 check inter 5s rise 2 > fall 3 > server prodcaepreview02-dal05 10.41.135.141:80 check inter 5s rise 2 > fall 3 > > -- > Timothy W. Foreman > Security Administrator/Sr. Systems Administrator > Internet Broadcasting > P: 651.365.4181 | F: 651.365.4430 > [email protected] | AIM: timfibsys > > Visit us at: www.ibsys.com > > >
Hi, First of all, you're using HAProxy in tunnel mode. So even if the health check is failing, as long as the server does not shut the connection, HAProxy will maintain it opened, fowarding data as long as the client sends it. Two ways to avoid this behavior: 1. shut down the service to force a failover 2. enable the option "http-server-close" cheers
