Failover. Reesetablishing contact after communications-interrupted

[Glenn Satchell]

>Date: Tue, 29 May 2007 08:46:21 +0200
>From: "Martin Ericsson" <martin.s.ericsson at gmail.com>
>To: dhcp-users at isc.org
>Subject: Failover. Reesetablishing contact after communications-interrupted
>
>Hi.
>I have experienced some problems with DHCP failover. The two servers
>occasionally (once every week or so) loses contact. I guess it is a
>network error. This is what happens:
>
>From syslog on server1:
>/var/log/syslog
>May 23 16:16:22 nadrdir01 dhcpd: failover: listener: no matching state
>May 23 16:16:57 nadrdir01 dhcpd: timeout waiting for failover peer 
dhcp-failover
>May 23 16:16:57 nadrdir01 dhcpd: peer dhcp-failover: disconnected
>May 23 16:16:57 nadrdir01 dhcpd: failover peer dhcp-failover: I move
>from normal to communications-interrupted
>
>From syslog on server2:
>failover peer "dhcp-failover" state {
>  my state communications-interrupted at 3 2007/05/23 14:16:17;
>  partner state normal at 2 2007/05/15 09:02:47;
>  mclt 600;
>}
>
>
>Both servers go into the "Communications-interrupted" state, but still
>think the other server is in "Normal" state.

The "partner state normal" is only set when the clients first conenct
and is then never updated, so you can't rely on it. I asked about this
once before.

If you only restart one server does it come back? This has been my
experience.

If you search the archives you may find an earlier post from me with
the same problem with 3.0.4 (I think). I am sure it used to reconnect
properly on prior releases. We have had the same problem where it
wouldn't automatically reconnect when doing disaster tests and breaking
the network between our two dhcp servers for a few hours.

Is there any traffic if you trace network traffic on ports 519 and 520?

I run 3.1.0a3 at home and that reconnects automatically, so when the
stable release comes out that may be an option for you.

regards,
-glenn