Hi, all.

I have a rather odd situation at a single client site with a 2-node
cluster. I have configured SLP, DNS, and DHCP for clustering, and
failover/failback seems to be working quite well.

There are two logical networks on the LAN:

DHCP is configured for the subnet. The other subnet is
static only, and used for the cluster nodes, other servers, printers, etc.

NetWare servers have IPs on both networks assigned, though the cluster
objects (including DHCP) are all on the subnet.

Routing on the network is handled by an Astaro v7 gateway, and is
configured as the default router in DHCP.

Clients renewing addresses are able to renew without a problem, but new
clients requesting address assignments (or whose leases have expired)
return to the DHCP server:

DHCP client <xx:xx:xx:xx:xx:xx> did not select this server.

There should be no reason to configure DHCP forwarding in the Astaro
gateway, as the cluster members themselves have valid IPs on both
subnets. However, could the fact that the clustered DHCP object's
address is on the subnet and is attempting to hand off
addresses on the subnet without an address itself on that
subnet be the cause of confusion to the clients?

I first noticed this issue with my own ThinkPad running OS/2, and
figured that it might be due to some bug in the OS/2 stack (ported from
AIX), but I see this on Win32 and Linux stations, as well.


Lewis G Rosenthal, CNA, CLP, CLE
Rosenthal & Rosenthal, LLC
Accountants / Network Consultants
New York / Northern Virginia www.2rosenthals.com
eComStation Consultants www.ecomstation.com
Novell Users Int'l www.novell.com/openenterpriseserver
Need a managed Wi-Fi hotspot? www.hautspot.com