Appendix F. Initial Configuration Checklist

Following is a checklist of the steps you must perform when installing and configuring a CXFS system.


Caution: CXFS is a complex product. To ensure that it is installed and configured in an optimal manner, you must purchase initial setup services from SGI.

This checklist is not intended to be used directly by the customer, but is provided for reference. It is intended to be used only as an aid; you must be certain to read this entire manual, especially Chapter 18, “Troubleshooting”, before attempting to complete these procedures.


[ ]

Understand the application use, storage configuration, and I/O patterns at the site. (Not all applications benefit from CXFS; see “Comparison of XFS and CXFS” in Chapter 1.)

[ ]

Connect the SAN hardware. See the RAID documents.

[ ]

Is there a private network? This is a requirement. SGI recommends a network switch rather than a hub for performance and control.

 

[ ]

Are there reset lines or I/O fencing? One of these solutions is required to ensure data integrity for all nodes. Clusters should have an odd number of server-capable nodes. See “Isolating Failed Nodes” in Chapter 1, and “Reset” in Chapter 1.

 

[ ]

Read this book and any README files provided with the release. If you have clients running operating systems other than IRIX or SGI ProPack for Linux 64-bit, also read the CXFS MultiOS Client-Only Guide for SGI InfiniteStorage.

 

[ ]

Verify that the network is usable. See Chapter 6, “IRIX CXFS Installation”.

[ ]

Install the CXFS software. See Chapter 6, “IRIX CXFS Installation”, and Chapter 7, “Linux 64-bit CXFS Installation”. If you have clients running operating systems other than IRIX or SGI ProPack for Linux 64-bit, also see the CXFS MultiOS Client-Only Guide for SGI InfiniteStorage.

 

[ ]

Modify the configuration files and perform other tasks as needed. See Chapter 8, “Postinstallation Steps”.

 

[ ]

Completely configure a small cluster (3 nodes). See Chapter 9, “Initial Configuration of the Cluster”.

 

[ ]

Look for errors in the daemon log files in the /var/cluster/ha/logs directory.

 

[ ]

If all is well, add the rest of the nodes. If there are problems, see Chapter 18, “Troubleshooting”.

 

[ ]

Set up the filesystems. See Chapter 9, “Initial Configuration of the Cluster”.