Retries when backend goes DOWN after was assigned

[Jose María Zaragoza]

Hello:

I'm using  haproxy-1.4.22 on TCP mode ( I paste haproxy.conf on bottom )
I use health checking with default values, so every 2 seconds, check id
server is available.

But I've observed that when a backed server is checked as UP but this
server goes DOWN when HA Proxy tries to connect to it ,
HA Proxy doesn't try to connect to other UP backend server

Is there any option to change this behaviour ?


I'm talking about balancing BEFORE to connect a backend server
I understand that if backend server goes DOWN during sent , HA Proxy
doesn't try with another backend server

Thanks and regards



#---------------------------------------------------------------------
# main frontend
#---------------------------------------------------------------------
frontend  main *:4100

    mode tcp
    log                     global
    option                  tcplog
    option                  dontlognull

    maxconn                 200
    timeout client          10s
    default_backend         tuxedo

#---------------------------------------------------------------------
# tuxedo backend
#---------------------------------------------------------------------
backend tuxedo
    mode tcp
    log                     global
    option                  tcplog

    retries 3
    timeout server  10s
    timeout connect 5s
    timeout check   5s
    timeout queue   20s

    balance roundrobin
    default-server maxconn 30 slowstart 3s
    server  tuxedo1-server tuxedo1-server:4100 weight 50 check
    server  tuxedo2-server tuxedo2-server:4100 weight 50 check