How to share IBM Spectrum Archive™ on a Linux system.
Before you begin
Note: IBM Spectrum Archive Library
Edition for Windows does not support file sharing by using NFS.
About this task
Beginning with version 2.1.2, IBM Spectrum Archive Library
Edition for Linux allows users to share Spectrum Archive by using the Network
File System (NFS). IBM Spectrum Archive Library
Edition
for Linux is implemented with File system in
Userspace (FUSE), which is a kernel module that provides functions for a user
space program to be able to implement a fully functional file system.
File sharing by using NFS requires the -o noforget option when you mount a
library. For more information, see Mounting a library by using the ltfs or
ltfs.exe command.
Procedure
Complete the following steps to export an Spectrum Archive medium by using NFS on
a Linux system.
- Stop the NFS server.
- Configure the NFS server to export the Spectrum Archive mount point.
- Mount Spectrum Archive to
the mount point as shown in the following example:
ltfs -o devname=/dev/IBMtape0 -o noforget [mount point]
Notes: - The opt option is required to resolve inode and file by using NFS.
- The sync time must not be close or many indexes are written. It is recommended that you use the
default setting of 5 minutes. For more information, see Sync operation.
- Start the
NFS server.
- Mount the exported Spectrum Archive mount point in the NFS
clients with the hard and intr options.
Note: By default, a Windows client mounts the NFS server
with the soft option, which might cause a timeout. In this case, it is recommended
that you use the hard option to mount. For example, use the mount command with the
o mtype=hard option.
- Access the exported Spectrum Archive mount point from the
NFS client.
- Unmount the exported Spectrum Archive mount point from all
NFS clients.
- Stop the NFS server.
- Unmount Spectrum Archive.
Results
It is possible that requests from the NFS clients might be out of order. In this case,
Spectrum Archive might return
an error if multiple operations are run against the same file in a short time, even if the requests
are serialized in the NFS clients.