To set up the Macintosh portion of Novell Native File Access Protocols in an environment running Novell Cluster Services:
Ensure that afptcp.nlm is loaded on all servers in the cluster by specifying MODULES at the server system console and reviewing the list of loaded modules.
Afptcp.nlm is loaded automatically on the server by the afpstrt.ncf file, which is automatically added to the autoexec.ncf file during the NetWare 6.5 installation.
Cluster enable the shared-disk pools or volumes by following the procedures described in Novell Cluster Services Administration Guide.
When you create and cluster enable an NSS pool or volume by following the above-referenced procedures, a screen appears that lets you choose the advertising protocols. Ensure AFP is selected on this screen. This will cause an AFPBIND command to be added automatically to the cluster-enabled pool volume load script, which ensures that your cluster-enabled pools are highly available to Macintosh clients.
AFPBIND allows AFP virtual server names to be advertised via SLP.
(Optional) Rename cluster-enabled volumes so Macintosh users will see the same volume name regardless of what server has the volume mounted.
For instructions, see Renaming Volumes.
Volumes are displayed as ServerName.VolumeName. If the server fails over, the user sees the next flivver server with the same volume name. For example, Server1.VOL1 becomes Server2.VOL1. Renaming each ServerName.VolumeName to a common name displays the common name regardless which server is providing the volume. For example, renaming Server1.VOL1 to Graphics, Server2.VOL1 to Graphics, and Server3.VOL1 to Graphics displays Graphics regardless which server is providing VOL1.
Macintosh clients should now be able to access files on the cluster by specifying the IP address or virtual server name of the cluster-enabled volume.
NOTE:Novell Native File Access Protocols does not support automatic reconnect for Macintosh computers. If the network connection between a Mac computer and one of the servers in the cluster fails, the user must reconnect using the same IP address for the cluster-enabled volume.