13.3 File System Support

13.3 File System Support

The physical file systems that are supported by the CFS are shown in Table 13-1. Note that we also list named pipes in this table even though a named pipe (or fifo) is not a file system, but a fifo is a file in a file system. Heck, we just wanted the word fifo in the document at least three times.

If a file system is mounted cluster-wide and the member serving the file system fails, as long as there is another member in the cluster with an active path to the device that contains the file system, then the CFS will automatically relocate the server function to that member. There are two exceptions to this rule:

• The file system is mounted as a partitioned file system (i.e., mount -o server_only). See section 13.4.

• The file system is an NFS file system.

If a cluster member is acting as the CFS server for an NFS-mounted file system (i.e., the cluster member is the NFS client) and it fails, the NFS-mounted file system is unmounted cluster-wide.

To prevent the loss of the NFS-mounted file system upon member failure, you can configure automount or AutoFS (autofs) for the NFS file systems that you want to remount.

For more information on automount, see the automount(8) reference page, the Network Administration: Services Guide for Tru64 UNIX version 5.1A, or the Network Administration Guide for Tru64 UNIX version 5.1 or version 5.0A.

For more information on autofs, see the autofs(8), autofsmount(8), and sys_attrs_autofs(5) reference pages as well as the Network Administration: Services Guide for Tru64 UNIX version 5.1A, or the Network Administration Guide for Tru64 UNIX version 5.1. AutoFS is not supported in V5.0A. See also section 20.4.