IBM Books

Administration Guide


Chapter 7. Managing file collections

The SP system installs file collection technology by default to simplify the task of maintaining duplicate files on multiple machines. In the delivered system, the files that are required on the control workstation, boot-install servers, and processor nodes belong to file collections. The master copy of each collection is on the control workstation. A process checks periodically, perhaps once each hour, if any file has been updated. Each file that was updated since the last time it checked, is propagated to the nodes that also have a copy.

This chapter explains the file collections that are delivered with the system. It shows you how to report on, update, and maintain the system file collections, as well as how to build file collections for your own purposes. It explains the basic concepts you need to understand before working with file collections, such as what file collections are, how they differ from standard directories and files, and how they are used in the SP system.

You can easily maintain the integrity of your files also by grouping them into file collections and using the tools provided to manage them. You can manage updates to your files with consistency and accuracy across the multiple nodes in your SP system. With a master version on the control workstation, you can be assured that any modifications are propagated to the nodes within the set time interval.

Usage note:

To run the commands specified in this chapter, you must be logged in as the root user with write access to the SDR.

You might want to take advantage of file collections to maintain files such as /etc/environment and /.kshrc. You need to read the entire chapter for complete information but briefly, the steps for a typical scenario might be as follows:

  1. Develop a list of files you want to update only on the control workstation and have distributed to the nodes of your SP system.
  2. Add that list to the list file in the file collection /var/sysman/sup/user.admin, one line for each file you want distributed, using the syntax:
    upgrade <full path name of the file>
    

    For example,

    upgrade ./etc/environment
    
  3. Within the set interval, all the files in the list file will be distributed to the nodes. If you do not want to wait, you can force an immediate update by using the command:
    dsh -a supper update user.admin
    

    If your PATH environment variable through dsh is not set to access the supper command, you might need to use:

    dsh -a /var/sysman/supper update user.admin
    

There might be cases when you want to distribute some files on the nodes but keep a different version on the control workstation for things you want to start on the nodes but not on the control workstation, for example file /etc/rc.tcpip. Use file collection node.root in that case, as in this example:

  1. To the list file in the file collection /var/sysman/sup/node.root, add the following:
    upgrade .
    
  2. Put the files you want distributed in subdirectories of /share/power/system/3.2 on the control workstation. For example /share/power/system/3.2/etc/rc.tcpip will be distributed to the nodes as /etc/rc.tcpip. After this, any file added to the /share/power/system/3.2 directory is automatically distributed to the nodes.
  3. To force immediate update, use the command:
    dsh -a supper update node.root
    

The file collection technology provides you with the means to perform both beginning basic tasks and optional advanced tasks. For the basic tasks, you need only understand the fundamental concepts and be familiar with the delivered file collections. The more advanced tasks require an understanding of the master files, what they contain, and how they work.

Information is presented on the following basic file collection tasks:

Information is presented on the following advanced file collection tasks:


[ Top of Page | Previous Page | Next Page | Table of Contents | Index ]