2 Remote Installation Services

This chapter introduces Remote Installation Services (RIS) and the ris utility, and explains the relationship between RIS servers and clients. The following topics are included:

Understanding RIS concepts and the benefits of using RIS (Section 2.1)

Starting RIS (Section 2.2)

Introducing RIS areas and product environments (Section 2.3)

Understanding RIS client characteristics (Section 2.4)

Registering RIS clients (Section 2.5)

Identifying a client's hardware network address (Section 2.6)

2.1 Overview

Remote Installation Services (RIS) uses the ris utility to set up a central computer system (a server) to service multiple computer systems (clients) on a local area network (a LAN) with required software.

With RIS, the server has a disk area set aside as the RIS area. The RIS area contains copies of software kits that are available for installation on to registered clients. Figure 2-1 shows how the RIS system works.

Figure 2-1: RIS Server and Client

The server maintains information in the RIS areas about what software kits clients can access. Kits are organized so that RIS can serve different versions of a software product to multiple hardware platforms and operating systems. The server's RIS area uses the Network File System (NFS) to provide read-only access to RIS clients.

Beyond verifying RIS clients' identities and managing their kit load requests, the RIS server does not interact directly with the clients. However, if the RIS server and clients are installing or using Secure Shell software that is installed when installing or upgrading your systems to Tru64 UNIX Version 5.1B or higher to implement a secure rutils environment (rlogin, rcp, etc.), host/server based authentication may be required. See the Security Administration manual, ssh2(1) and ssh2_config(4) for further information.

You do not have to set aside a system as a dedicated RIS server; you can use the same system to support local timesharing users.

A RIS client uses the setld utility to install software kits from the RIS server instead of from local distribution media.

The benefits and advantages of RIS include the following:

Installation and setup of servers and clients are done by scripts, thereby simplifying the server system administrator's task. Maintenance of the server's disk areas is similarly straightforward. The system interface is the same regardless of system type.

Because RIS supports different hardware platforms and different software versions, it is adaptable to a wide variety of client systems and requirements. Servers running a given version of an operating system can serve clients running the same version or an earlier version of the operating system. In addition, if the ris utility on the server is updated to the current version with the utilupdate utility available on distribution media, RIS servers running an earlier version of the operating system can make later versions of the operating system available to RIS clients.

RIS uses a single set of kit files for all clients having the same architecture.

You can perform a cloned installation on a RIS client, letting you duplicate a similar system installation or configuration. See the Installation Guide — Advanced Topics manual for information about installation cloning and configuration cloning.

2.2 Starting RIS

You always should run the ris utility as superuser. To start the ris utility, enter the following command:

# /usr/sbin/ris

When RIS starts up, it checks the status of the RIS areas.

If RIS can access all the products it was able to access the last time RIS was started, the ris utility displays the following message:

Checking accessibility of RIS areas...  done

If RIS cannot access all the products it was able to access previously, it displays the following message:

No Products Available in /var/adm/ris/ris0.alpha
 
Delete RIS environment? [y]:

This may occur because the source for this RIS environment is no longer mounted, and can be corrected by remounting the source. If the source is no longer available, you may delete this RIS environment. If you remount the source, you must restart RIS so that the environment is available.

If you try to start RIS without superuser privileges, the following message may be displayed:

Checking accessibility of RIS areas...
No permission to write /usr/var/adm/ris/ris0.alpha/ProdNames
done

Correct this problem by logging in as root or using the su command to gain superuser privileges before you start RIS.

2.3 RIS Areas and Product Environments

In addition to the server's normal disk area, an area is reserved to hold RIS software kits. This RIS area contains one or more product environments. Each product environment contains one or more product kits suitable for installation on a given hardware or software platform. Figure 2-2 shows a generalized illustration of a sample RIS area.

Figure 2-2: Sample RIS Area Overview

In Figure 2-2, the RIS area /var/adm/ris contains one product environment, ris0.alpha. Each product environment contains products for a specific platform. In Figure 2-2, the target platform is machines using Alpha processors. Multiple product environments can exist in a single RIS area. Each product environment contains one or more product directories, each product directory contains several product kit archives, called software subsets. Figure 2-2 shows a product environment named ris0.alpha containing directories called product_001 and product_002.

Figure 2-2 also shows the kit/isl directory. The kit/isl directory contains installation tools required by clients when they install software over the network. If your environment is in Direct CD-ROM (DCD) format, the kit/isl directory does not exist. An environment in DCD format is the same as a system disk format, it includes root, /usr, and so on.

The server itself usually does not use any of the RIS areas. System administrators can access the product area as required for maintenance and for installation or removal of product kits.

For more flexibility, you can establish multiple RIS areas in separate partitions. RIS areas on a given server can be exported to other servers using the Network File System (NFS). Servers that import such RIS areas can use them as if they were local, supplying the imported subsets to their own set of clients. Section 4.5 describes how to use NFS to mount a RIS area. The Network Administration: Services manual describes how to export and import file systems.

2.4 RIS Client Characteristics

A RIS installation uses the LAN as its installation media instead of a distribution CD-ROM. A RIS client can install any software kit for which it is registered on the server. The installation procedure runs entirely on the client and, after the necessary software is installed, no continuing relationship is required between the RIS server and client.

The operating system itself can be among the kits that are available from the server. To install the operating system, the client processor is booted across the network using a minimal generic kernel that is part of the software kit. The RIS area is NFS mounted and becomes the client's root file system during the installation.

When the client is booted, either the text-based or graphical installation interface is launched. After all installation responses are entered, the installation software configures the file system, and then uses the setld utility to load the software you selected. See setld(8) for more information.

After the installation is complete, the system reboots with the newly installed software. For information on installation procedures, see the Installation Guide.

2.5 Registering Clients

A client must be registered with only one server for the base operating system. If you register a client with more than one server for the base operating system, each server the client is registered on will attempt to respond to the client's network boot request with unpredictable results.

To change the server with which a client is registered for the base operating system, first remove the client from the current server's client database and then register it with the new server. See Chapter 6 for information about registering and removing RIS clients.

A client can be registered with multiple servers for optional subsets and products other than the base operating system. When you load optional subsets or layered products with the SysMan Menu, you specify the name of the server from which you will copy the kits.

If you are performing a rolling upgrade from a RIS server, you must register both the cluster alias and the lead cluster member as RIS clients before you execute the installation phase of the rolling upgrade. For information on rolling upgrade procedures, see the TruCluster Server Cluster Installation manual and the Installation Guide.

2.6 Identifying a Client Hardware Network Address

You need to know your client's hardware network address when you are registering a client to a RIS server. There are several ways to identify this information:

Log in to the RIS client as root or use the su command to gain superuser privileges, then shut down the system to the console prompt ( >>> ).
Use the show dev command to show all devices, and look for the hardware address of your network interface in the form xx-xx-xx-xx-xx-xx. For example:
```
>>> show dev
        
.
.
.
ewa0.0.0.0.1000.0	EWA0	xx-xx-xx-xx-xx-xx
        
.
.
.
```

Log in to the RIS client as root or use the su command to gain superuser privileges.
Use the uerf -r 300 command and look for the string hardware address in the ouput. Either that line or the next one contains the hardware network address. For example:
```
# uerf -r 300 | grep -i "hardware address" | uniq
_hardware address: xx-xx-xx-xx-xx-xx
```
If the hardware address is not on the line that contains the string hardware address, search the output from the uerf command to find the correct hardware address. For example:
```
# uerf -r 300 | more
        
.
.
.
_Interface, hardware address:
			_xx-xx-xx-xx-xx-xx 
        
.
.
.
```

Log in to the RIS server as root or use the su command to gain superuser privileges.

Use the ping and arp commands to determine the hardware address of a running client from the RIS server. For example, to determine the hardware address of the RIS client atlanta, enter a command similar to the following example:

# /usr/sbin/ping -q -c1 atlanta ; arp atlanta
PING atlanta.cities.xsamplex.com (nn.nn.nnn.nnn): 56 data bytes
----atlanta.cities.xsamplex.com PING Statistics----
1 packets transmitted, 1 packets received, 0% packet loss
round-trip (ms)  min/avg/max = 0/0/0 ms
atlanta (nn.nn.nnn.nnn) at xx-xx-xx-xx-xx-xx