LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: Keepalived Director failover, LVS-TUN

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: Keepalived Director failover, LVS-TUN
From: Matt Chan <engineuity@xxxxxxxxx>
Date: Mon, 30 Jan 2006 13:08:55 -0500
DIP1: x.x.x.217 on eth0, VIP x.x.x.208 on eth1
DIP2: x.x.x.218 on eth0, VIP x.x.x.208 on eth1
On 1/26/06, Joseph Mack NA3T <jmack@xxxxxxxx> wrote:
>
> aliases were deprecated a long time ago (2.4 kernels?). Some
> programs may not work with aliases.
>
OK, I added another NIC so the director traffic and VIP traffic can be
separated...I have an eth0 for the DIP and eth1 for the VIP, on both the
director and the backup.
the vrrp part of my keepalived.conf on director1 now looks like:
## VRRP Instances Definitions
vrrp_sync_group VG1{
  VI_1
}
vrrp_instance VI_1{
  state MASTER
  interface eth1
  virtual_router_id 51
  priority 150
  smtp_alert
  advert_int 1
  virtual_ipaddress{
    165.230.111.208/32 dev eth1
  }
}
and director2 has state BACKUP and priority of 100

On 1/26/06, Dominik Klein <dk@xxxxxxxxxxxxxxxx> wrote:
>
> >     Try to start keepalived with -d and -D to see if it gives any
> >     error-messages in /v/l/m

When I restart keepalived, the backupdirector always transitions to the
MASTER state, but the director1 is also in the MASTER state. It did once
hold an election, but the election never finalized, and I never got it
again. The results are from /var/log/syslog.
Am I missing anything from my vrrp setup?
-Matt Chan

<Prev in Thread] Current Thread [Next in Thread>