We have set up NFS in Active/Passive mode on our NetWare 6.5 cluster.

This works properly for all our (Red Hat) Linux servers on the LAN, but
there are a couple of servers that have been placed in a DMZ and have to
access NFS through a firewall.

We have opened up the apropriate ports to allow communications back and
forth through the firewall to/from the Linux host and the cluster
resource IP address.

The Linux host was unable to mount the filesystems that were exported
for it and mount attempts eventually time out.

After some troubleshooting, we discovered that even though we have
NFSCLUST in the resource load script with the correct IP-address,
packets from the clusternode running the NFS resource to the clients are
sent out from the mail IP-address of the node instead of the resource IP
address.

Is there any way to force NFS on NetWare to use the cluster resource IP
address?

The start script is:

nss /poolactivate=POSTBUS
mount POSTBUS VOLID=236
CLUSTER CVSBIND ADD C02-POSTBUS-SERVER 192.168.20.129
NUDP ADD C02-POSTBUS-SERVER 192.168.20.129
add secondary ipaddress 192.168.20.129

NFSCLUST 192.168.20.129 POSTBUS C02-POSTBUS-SERVER
POSTBUS:\SYSTEM\NFSSTART

--
Greetings/Groetjes,
Maurice Snellen