[Keepalived-devel] Failover problem

[Alexis J. <aja...@pr...>]

Hi !

I'm trying to use Keepalived with my cluster, which is a quite 
"specific" cluster... In fact, I'm trying to use only two servers, which 
are at the same time directors and real servers :
- Srv 1 :
* IP : 10.0.0.232
* Service : Squid, on port 3128
* Has to be backup server
- Srv 2 :
* IP : 10.0.0.233
* Service : Squid, on port 3128
* Has to be master server
- VIP : 10.0.0.234

I'm not interested for the moment by load balancing but only fail over 
with services check. Keepalived seems to be really excellent for this, 
but I've a bad comportment. Maybe due to my configuration...

Squid is running on each server. I start Keepalived on each server : no 
problem, Srv2 answers to telnet on port 3128. I disconnect Srv2 from 
network : Srv1 becomes Master and answers to connections. Good at this 
point :)
Now, I reconnect Srv2 to network : Srv1 TCP Check is ok, but Srv1 stays 
Master. Is it normal ? It seems to me that Srv2 should become master. 
And, finally, I disconnect Srv1 : of course, no server is master, and 
consequently, cluster is dead :(
Is there a problem with me configuration ?

For your information, when I reconnect Srv1, cluster stays dead.
It seems there is a problem, so any help will be appreciated :)

Thanks !

----------------------------------------------------
Config file on Srv1 :

global_defs {
	notification_email {
		my...@my...
	}
	notification_email_from sr...@sr...
	smtp_server 10.0.0.10
	smtp_connect_timeout 30
	router_id ClsSquid
}

vrrp_instance VI_1 {
	state BACKUP
	interface eth0
	virtual_router_id 51
	priority 100
	advert_int 1
	authentication {
		auth_type PASS
		auth_pass 1111
	}
	virtual_ipaddress {
		10.0.0.234
	}
}

virtual_server 10.0.0.234 3128 {
	delay_loop 6
	lb_algo wlc
	lb_kind NAT
	persistence_timeout 600
	protocol TCP

	real_server 10.0.0.232 3128 {
		weight 50
		TCP_CHECK {
			connect_timeout 3
		}
	}

	real_server 10.0.0.233 3128 {
		weight 100
		TCP_CHECK {
			connect_timeout 3
		}
	}
}
----------------------------------------------------

Config file on server 2

global_defs {
	notification_email {
		my...@my...
	}
	notification_email_from sr...@sr...
	smtp_server 10.0.0.10
	smtp_connect_timeout 30
	router_id ClsSquid
}

vrrp_instance VI_1 {
	state MASTER
	interface eth0
	virtual_router_id 51
	priority 100
	advert_int 1
	authentication {
		auth_type PASS
		auth_pass 1111
	}
	virtual_ipaddress {
		10.0.0.234
	}
}

virtual_server 10.0.0.234 3128 {
	delay_loop 6
	lb_algo wlc
	lb_kind NAT
	persistence_timeout 600
	protocol TCP

	real_server 10.0.0.232 3128 {
		weight 50
		TCP_CHECK {
			connect_timeout 3
		}
	}

	real_server 10.0.0.233 3128 {
		weight 100
		TCP_CHECK {
			connect_timeout 3
		}
	}
}


PS : Log files are joined to this mail
-- 
Alexis