[LVS-DR + Keepalived] strange Netlink error

[Samuel Tran <stran@xxxxxxxx>]

Hi all,

I have setup LVS-DR + Keepalived using two directors and two real servers.
Here is a simple drawing of the architecture:

VIP = 192.168.1.10

director1       director2
192.168.1.101   192.168.1.102
|               |
|               |
--------|--------
        |
        |
--------|--------
|               |
|               |
realserver1     realserver2
192.168.1.111   192.168.1.112


Here is VRRP configuration for the MASTER (director1) and the BACKUP 
(director2):

## MASTER

vrrp_sync_group VG_1
{
   group {
     VI_1
   }
}

vrrp_instance VI_1 {
    state MASTER
    interface eth0
    priority 150
    advert_int 1
    virtual_router_id 50
    virtual_ipaddress {
        192.168.1.10
    }
    preempt
}

## BACKUP

vrrp_sync_group VG_1
{
   group {
     VI_1
   }
}


vrrp_instance VI_1 {
    state BACKUP
    interface eth0
    priority 100
    advert_int 1
    virtual_router_id 50
    virtual_ipaddress {
        192.168.1.10
    }
    preempt
}


Whenever I reboot the MASTER (director1) and start keepalived I get this error 
message in my logs:

Dec 22 10:11:26 idun Keepalived: Starting Keepalived v1.1.7 (12/08,2004)
Dec 22 10:11:26 idun Keepalived: Starting Healthcheck child process, pid=229
Dec 22 10:11:26 idun Keepalived: Starting VRRP child process, pid=230
Dec 22 10:11:26 idun Keepalived_healthcheckers: Using MII-BMSR NIC polling 
thread...
Dec 22 10:11:26 idun Keepalived_healthcheckers: Registering Kernel netlink 
reflector
Dec 22 10:11:26 idun Keepalived_healthcheckers: Registering Kernel netlink 
command channel
Dec 22 10:11:26 idun Keepalived_vrrp: Using MII-BMSR NIC polling thread...
Dec 22 10:11:26 idun Keepalived_vrrp: Netlink reflector reports IP 
192.168.1.101 added
Dec 22 10:11:26 idun Keepalived_vrrp: Registering Kernel netlink reflector
Dec 22 10:11:26 idun Keepalived_vrrp: Registering Kernel netlink command channel
Dec 22 10:11:26 idun Keepalived_vrrp: Registering gratutious ARP shared channel
Dec 22 10:11:26 idun Keepalived_vrrp: Configuration is using : 150397 Bytes
Dec 22 10:11:26 idun Keepalived_vrrp: ------< Global definitions >------
Dec 22 10:11:26 idun Keepalived_vrrp:  LVS ID = LVS_DEVEL_MASTER
Dec 22 10:11:26 idun Keepalived_healthcheckers: Configuration is using : 127104 
Bytes
Dec 22 10:11:26 idun Keepalived_vrrp:  Smtp server = xxxxxxxxxxxxxxxx
Dec 22 10:11:26 idun Keepalived_vrrp:  Smtp server connection timeout = 30
Dec 22 10:11:26 idun Keepalived_vrrp:  Email notification from = xxxxxxxx
Dec 22 10:11:26 idun Keepalived_vrrp:  Email notification = xxxxxxxxxxxxxxx
Dec 22 10:11:26 idun Keepalived_vrrp: ------< VRRP Topology >------
Dec 22 10:11:26 idun Keepalived_vrrp:  VRRP Instance = VI_1
Dec 22 10:11:26 idun Keepalived_vrrp:    Want State = MASTER
Dec 22 10:11:26 idun Keepalived_vrrp:    Runing on device = eth0
Dec 22 10:11:26 idun Keepalived_vrrp:    Virtual Router ID = 50
Dec 22 10:11:26 idun Keepalived_vrrp:    Priority = 150
Dec 22 10:11:26 idun Keepalived_vrrp:    Advert interval = 1sec
Dec 22 10:11:26 idun Keepalived_vrrp:    Preempt Active
Dec 22 10:11:26 idun Keepalived_vrrp:    Virtual IP = 1
Dec 22 10:11:26 idun Keepalived_vrrp:      192.168.1.10/32 brd 192.168.1.10 dev 
eth0 scope global
Dec 22 10:11:26 idun Keepalived_vrrp: ------< VRRP Sync groups >------
Dec 22 10:11:26 idun Keepalived_vrrp:  VRRP Sync Group = VG_1, BACKUP
Dec 22 10:11:26 idun Keepalived_vrrp:    monitor = VI_1
Dec 22 10:11:26 idun Keepalived_vrrp: Watchdog: Starting listener on /tmp/.vrrp 
wdog socket
Dec 22 10:11:26 idun Keepalived_vrrp: VRRP sockpool: [ifindex(2), proto(112), 
fd(9,10)]
Dec 22 10:11:26 idun kernel: IPVS: [wlc] scheduler registered.
[snip]
Dec 22 10:11:26 idun Keepalived_healthcheckers: Watchdog: Starting listener on 
/tmp/.healthcheckers wdog socket
Dec 22 10:11:26 idun Keepalived_healthcheckers: Activating healtchecker for 
service xxxxxx
Dec 22 10:11:26 idun Keepalived_healthcheckers: Activating healtchecker for 
service xxxxxx
Dec 22 10:11:26 idun Keepalived_healthcheckers: Activating healtchecker for 
service xxxxxx
Dec 22 10:11:26 idun Keepalived_healthcheckers: Activating healtchecker for 
service xxxxxx
Dec 22 10:11:26 idun Keepalived_healthcheckers: Activating healtchecker for 
service xxxxxx
Dec 22 10:11:26 idun Keepalived_healthcheckers: Activating healtchecker for 
service xxxxxx
Dec 22 10:11:26 idun Keepalived_healthcheckers: Netlink: filter function error
Dec 22 10:11:26 idun Keepalived_healthcheckers: Netlink: filter function error
Dec 22 10:11:26 idun Keepalived_healthcheckers: Netlink: error: Device or 
resource busy, type=(22), seq=1103728287, pid=0
Dec 22 10:11:27 idun Keepalived_vrrp: VRRP_Instance(VI_1) Transition to MASTER 
STATE
[snip]

Has anyone experienced this "Netlink: filter function error"?

However if I stop keepalived and restart it I don't see that error message 
anymore.
Besides failovers back and forth with the BACKUP were successful.

I would appreciate any input.

Thanks.
Sam