Is this DHCP 3.0.5 failover recovery behavior normal?

[Daniel Loughlin]

Hi,

I have two 3.0.5 DHCP servers configured for failover.

---- snip from primary ---
# failover configuration
failover peer "dhcp-failover" {
        primary;
        address 10.1.28.34;
        port 520;
        peer address 10.1.28.35;
        peer port 520;
        max-response-delay 30;
        max-unacked-updates 10;
        load balance max seconds 3;
        mclt 1800;
        split 128;
}

--- snip from secondary ---

# failover configuration
failover peer "dhcp-failover" {
        secondary;
        address 10.1.28.35;
        port 520;
        peer address 10.1.28.34;
        peer port 520;
        max-response-delay 30;
        max-unacked-updates 10;
        load balance max seconds 3;
}


I was forced to poweroff both servers yesterday.  When I powered them 
back on, I noticed something I wasn't expecting.
Both servers in their dhcplog entered the normal state after synching 
the leases.  I could see the daemon listening on both servers. I could 
see the dhcp request frames on both servers via tcpdump. However, my 
dhcp servers were not responding to the requests. 

I was able to get them to respond after doing the following:

1) Start primary daemon
2) Start secondary daemon
3) Turn off primary daemon
4) The secondary would then hand out leases
5) Then I turned the primary back on, and both began to hand out leases.

Is this normal? I assumed I should have been able to simply turn both of 
these on and have them synchronize the leases, enter the normal state, 
and hand them out to clients.

Thanks in advance for any insight.
-Danny