|
Hello,
We currently have a cluster of 5 machines serving as a shared web cluster. The
load balancing is done with an older version of turboclusterd. One of the 5
machines is the load balancer (as well as a web server) and another of the 5 is
a backup load balancer (although that functionality has not worked since I've
been around--but it apparently did at one time).
I am in the process of modernizing this web cluster. I want to replace the load
balancing functionality with LVS. I have set up two new servers with Debian
(sarge), kernel 2.6.8 with IPVS 1.2.0 and ipvsadm v1.24. I have also installed
keepalived v1.1.7.
For now, I just want to be able to set up these two servers as LVS
directors--one primary, and one to take over the VIPs if the primary director
goes down. For real servers, I will be using the five web servers already in
our cluster. Eventually, once we are not reliant on turboclusterd and its
dependencies, these servers will be upgraded.
I have configured keepalived with one VIP using the 5 web servers as real
servers (using LVS-DR). I have health-checking working and am pretty pleased
with the setup. However, I am at a big loss when it comes to the load
balancing/director failover stuff (VRRP). The documented examples are for
setups much more complex than my simple test. I will eventually have 10 VIPS
or so, but in my testing setup now I have just one VIP that I want to belong to
the primary LVS directory, and which I want to fail over to the secondary if the
primary dies. I have tried to extrapolate from the documentation I could find
and create the relevant sections of keepalived.conf but I think I'm way off
track. I also can't seem to figure out if I need to configure anything for the
multicast to work... when I run keepalived with VRRP turned on and run tcpdump
either locally or on the secondary, I don't see any "heartbeat" packets or
anything.
This is the VRRP section of my keepalived.conf on my primary director:
vrrp_sync_group VG1 {
group {
VI_1
}
}
vrrp_instace VI_1 {
state MASTER
interface eth0
lvs_sync_daemon_interface eth0
virtual_router_id 51
priority 150
advert_int 1
smtp_alert
authentication {
auth_type PASS
auth_pass <<<<EDITED>>>>
}
virtual_ipaddress {
<<<<EDITED>>>>
}
}
IP numbers and passwords have been edited out...
The section is the same on the secondary except "state" is set to BACKUP.
I hope I'm not too far off base. Can anyone give me a nudge in the right
direction?
Thanks,
John
----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
|
