Setting up the home cluster

Use this procedure to set up the home cluster.

Set up a home cluster. For more information, see Creating your GPFS cluster.
Configure a cache relationship for AFM filesets by using the home cluster that you created.

The relationship between the home cluster and the cache cluster is set up by using NFS exports that are defined on home cluster. The home cluster exports NFS mount points that AFM cache cluster uses to synchronize data.
Create a file system and mount this file system on the home cluster. For more information, see Managing file systems.
The home side of the relationship is defined at the NFS share level. The home contains all the data available from the NFS mount point.
At the home cluster, create and link one or more filesets. For more information, see Filesets.
These filesets are used to set up NFS exports at the home cluster. These export paths are fileset junction paths where filesets are linked.
Export the fileset, one fileset at a time. Do one of the following steps:

Note: In this example, the IP address of the Gateway node of the cache cluster is 192.168.1.2. As an Administrator, ensure that NFS exports are accessible only to nodes at the cache.
- If you are using KNFS, complete the following steps for KNFS export of the home file system:
  1. Add fileset junction path to the /etc/exports file of home cluster export servers. Exported path must have the necessary permissions. no_root_squash and rw are mandatory and fsid is optional. An example of an NFS export entry with fsid at home side is as follows:
```
/gpfs/fs1/sw2 192.168.1.2(rw,nohide,insecure,no_subtree_check,sync,no_root_squash,fsid=1069)
```
  2. Restart the NFS services on home cluster export servers. Because you restarted the NF services, all the necessary NFS services start and the specified path to be used by the AFM cache is exported.
  3. If the file system goes down, you must export the file system again. The gateway nodes at the cache site must have access to the exported directory and can be mounted by using NFS.
- If you want to configure a file system on the home cluster with protocols nodes, complete the following steps:
  1. Use the mmnfs export add command to create export on junction path or the gpfs path of user choice.
```
mmnfs export add /ibm/gpfs0/nfsexport --client \
    "192.168.1.2(Access_Type=RW,Squash=no_root_squash,SecType=sys)"
```
    Use the mmnfs export list command to list NFS exports:
```
     mmnfs export list 
    
```
    A sample output is as follows:
    Path Delegations Clients
    
    ----------------------------------------
    
    /ibm/gpfs0/nfsexport none *
    Do not edit /etc/exports or any other NFS configuration files manually, and do not restart NFS services after the export is created.
After you export filesets at the home cluster, run mmafmconfig enable /ibm/gpfs0/nfsexport. For more information about the command, see mmafmconfig command.
Note:
1. Ensure that you add the IP addresses of all gateway nodes of the cache cluster. Multiple IP addresses can be indicated by a comma-separated list. Update the list of IP addresses whenever you add or remove a gateway node.
2. Ensure that the KNFS or NFS server at home is restarted, and home exports are available. CES NFS does not require a restart.
3. If both NFS and GPFS start automatically at the start-up time, ensure that GPFS starts before NFS as NFS can export GPFS only if it is loaded. If NFS starts before GPFS, run exportfs -r.
4. If an Ubuntu server is used as an AFM DR secondary server, configure the AFM secondary server so that the NFS server service rpc.mountd does not start with --manage-gids. rpc.mountd --manage-gids is not applicable for CES NFS or Native GPFS protocol.
5. When an AFM-DR secondary fileset is exported by using an NFS server, set the manage_gids=FALSE option. The server does not start because of this configuration. To change the option value, see Making bulk changes to NFS exports.
  This option is not applicable when an AFM-DR secondary fileset is connected over the native GPFS protocol.