Go to the previous, next chapter.

System Tailoring

The basic system has now been installed and now needs to be configured for the CERN environment.

Setting up TCP/IP

Firstly the network needs to be configured and started. In order to do this first become super user and then invoke netsetup as follows:

	   #/etc/netsetup install

You will be presented with a series of questions and asked to confirm each answer. You will require the following information:

The name of your machine
An alias for your machine (if required)
The network number. This is the first part of the TCP/IP address (e.g. 128.141)
When asked if the network uses subnet routing answer "no"
Your host number which is the last two numbers of the above address.
Broadcast address - all '1's at CERN, the default. (IP Broadcast Address 128.141.255.255)
You will be asked for your ethernet device type. Unless you have more than one Ethernet interface you will normally be prompted with the correct device. For VAXstations and DECstations this is the DESVA, Busless LANCE network interface (ln0). If there is any doubt you may determine the network interfaces on your system by running the netstat command as follows:
1. Log on as root (or become superuser)
2. Invoke netstat with the -i option:
```
	   #netstat -i
```
  The system will output a table similar to the following:
```
Name   Mtu   Network  Address  Ipkts Ierrs  Opkts Oerrs  Coll
ln0    1500  none     none         0     0      0     0     0
lo0    1536  non      none         0     0      0     0     0
```
  The interface name is given in the left hand column. In this example the device name of the network adapter is ln and the unit number is 0. The device lo0, which is known as localhost, allows the system to simulate a network and is not one of the network adapters.
/etc/routed should not be started
The network name is "ether"
There is no alias for the network
You will then be asked to enter the host names, abbreviations, network addresses and host addesses for other hosts in the network. There are two name servers at CERN which centralise this information, however, you should add here the machines with which you are likely to communicate most often. The procedure used to configure the name servers is described later in this chapter. The following is a list of the central services that you may wish to add:
```
dxcern.cern.ch   dxcern   128.141.201.96  Central ULTRIX Service 
dxsoft.cern.ch   dxsoft   128.141.248.184 Remote Installation Server
springer.cern.ch springer 128.141.6.58    Central Print Server
dxmint.cern.ch   dxmint   128.141.1.113   Mail Gateway 
vxcern.cern.ch   vxcern   128.141.0.1     Central VAX Service
cernvm.cern.ch   cernvm   128.141.2.4     CERNVM             
```
The next step asks for the names of trusted hosts. Be VERY careful. Users on a trusted host who have an account on your system may log in without password verification.

Defining the External Gateway

The following line should then be added to the end of the file /etc/rc.local in order to define the external gateway (for access to other networks at CERN and to nodes outside CERN):

	   /etc/route add default 128.141.200.10 1

Configuring the BIND/Hesiod Naming Service

The BIND/Hesiod service is a lookup service for information on host names, IP addresses and user and network services. The following describes how to set up these services to use the CERN name servers.

Defining the Search Order

The file /etc/svc.conf defines the order in which the naming services are searched. Type the following to run the command:

	   #/usr/etc/svcsetup

The svcsetup procedure then presents a series of menus as follows:

The first menu asks whether you want to modify the file, print the file or exit from svcsetup.
```
     Select the modify option
```
The next menu lists the databases that you can modify.
```
     Select the hosts databases (3)
```
The next menu lists the possible naming service orders and will be presented twice, once for the hosts database and again for the networks database.
```
     Select local,bind (5)
```

The file svc.conf will then be updated and svcsetup will exit.

Adding the Names Servers

The Berkeley Internet Name Domain (BIND) is a network naming service. The bindsetup command can be used to add, modify or remove BIND servers on your system. Before running bindsetup check that your machine is in multiuser mode and the network is up. Type the following to run the command:

   #/usr/etc/bindsetup

The bindsetup command then steps through the configuration process as follows:

You are first presented with a menu asking whether you want to add to or modify the BIND/Hesiod environment, remove BIND/Hesiod or exit from bindsetup.
```
     Select the add option (a)
```
You are then asked for the default BIND domain name. This is:
```
     cern.ch
```
The configuration menu for domain "cern.ch" then appears. This asks whether your machine is a primary server, secondary server, caching server, slave server or a client. There is also an option to exit from bindsetup.
```
     Select the client option (c)
```
You are then prompted for the host name and Internet address of the server or servers. There are two such machines at CERN and their names and addresses are listed below. Enter them both.
```
     d-name-1    128.141.200.5
     d-name-2    128.141.200.6
```
In order to spread the load evenly over the two name servers you are asked to enter them in the order d-name-1, d-name-2 if your IP address is odd and d-name-2, d-name-1 if your IP address is even.

The bindsetup command will then list the files that it has modified and give a warning message concerning the file sendmail.cf. This message may be ignored at this point as sendmail.cf should be replaced later (See section Setting up Mail).

Setting up Mail

The following steps should be taken in order to tailor mail for the CERN environment.

ULTRIX Version 4.2 Systems

Make sure that you are not in the /etc directory. Copy the file ultrix.cf from the central server (dxsoft) using ftp.

	   #ftp dxsoft
	   #Name (dxsoft.xxxxx): anonymous
	   #use your machine name as the password
	   #cd pub/sendmail
	   #binary
	   #get ultrix.cf
	   #quit

Copy the existing /etc/sendmail.cf to /etc/sendmail.cfold (for example) and replace it with the new one as follows.
```
	   #mv /etc/sendmail.cf /etc/sendmail.cfold
	   #mv ultrix.cf /etc/sendmail.cf
```
Freeze the configuration file with the following command:
```
	   #/usr/lib/sendmail -bz
```
Either reboot the system or kill the current sendmail process and restart it with the following command:
```
	   #/usr/lib/sendmail -bd -q1h -om
```
Send a mail to mail-support@dxmint.cern.ch requesting that your node name be added to the SMTP data base. This is particularly important if your machine has a DECnet node name but mail should reach it via SMTP.

ULTRIX Version 4.3 Systems

Make sure that you are not in the /etc directory. Copy the file ultrix43.cf from the central server (dxsoft) using ftp.

	   #ftp dxsoft
	   #Name (dxsoft.xxxxx): anonymous
	   #use your machine name as the password
	   #cd pub/sendmail
	   #binary
	   #get ultrix43.cf
	   #quit

Copy the existing /etc/sendmail.cf to /etc/sendmail.cfold (for example) and replace it with the new one as follows.
```
  #mv /etc/sendmail.cf /etc/sendmail.cfold
  #sed "s/^DA.*/DA`hostname | cut -d. -f1`/" ultrix43.cf > /etc/sendmail.cf
```
WARNING: The ` in the above example is a backquote.
Freeze the configuration file with the following command:
```
	   #/usr/lib/sendmail -bz
```
Either reboot the system or kill the current sendmail process and restart it with the following command:
```
	   #/usr/lib/sendmail -bd -q1h -om
```
Send a mail to mail-support@dxmint.cern.ch requesting that your node name be added to the SMTP data base. This is particularly important if your machine has a DECnet node name but mail should reach it via SMTP.

Starting the Network File System (NFS)

In order to set up and start NFS invoke the command nfssetup as follows:

	   #cd /
	   #/etc/nfssetup

The command will step through the setup procedure and ask you to give any file systems that you would like to import or export. In most cases the default answers may be taken for this procedure. For more information on NFS see the "ULTRIX Guide to the Network File System"(17).

Time Setting

After the System Installation

After the installation of ULTRIX has terminated and the machine has rebooted the time should appear in the following format:

	   Wed Oct 30 12:08:36 MET 1991

The time zone should be given as MET or MET DST. You will need to do the following in order to ensure that your system will switch to and from Daylight Saving Time (DST) on the correct days.

	   #cd /etc/zoneinfo
	   #cp localtime localtime.old         (if it exists)
	   #ln MET localtime

If the time has not been set correctly during the installation it may be reset in either of the following ways:

	   #/bin/date  yymmddhhmm

or using a timeserver at CERN:

	   #/usr/etc/ntp -sf time-1

N.B. Do not modify the time in multi user mode unless the machine is idle.

Keeping the Correct Time

There are two timeservers at CERN, time-1 and time-2, which are linked to external institutes who have reliable time (atomic clock) sources. The easiest way to use this service is to add an entry similar to the following to the file /etc/crontab:

	   01 00 * * * /usr/etc/ntp -sf time-1       (or time-2)

Please use time-1 if your IP address is odd and time-2 if yor IP address is even.

The above entry will set the time at one minute past midnight every night. Please choose a random time for your entry so that not all the CERN machines will ask the time simultaneously.

Diskless Clients

When a Diskless Client machine reboots it uses the rdate command to check the current time. Unfortunately the standard release of rdate asks for the time from anywhere. This causes a broadcast storm on the network because over five hundred Unix systems at CERN send a brodcast in order to find the source of the rdate. It is therefore recommended that for these systems you modify the file /etc/rc to use the ntp command instead - see the example given below.

    if [ "$DISKLESS" ]
    then
           /etc/nfs_umount -b             >/dev/null 2>&1
           /etc/mount -a -t nfs           >/dev/console
    #       /usr/etc/rdate -s              >/dev/console
           /usr/etc/ntp -sf time-1        >/dev/console
    fi

Printer Setup

Printers may be added to the system by using the lprsetup command. At CERN there is a central print server which allows access to a large number of network connected printers. The server is called "springer" and its TCP/IP address is 128.141.6.58. The following shows how to use lprsetup to add one of the central printers (513-lps) to your machine.

First of all you must get your machine added to the list of print hosts on springer. To do this you should send a request via email to the printing support account printsp@springer.cern.ch. Do not forget to give the name of your machine.
Invoke lprsetup
When asked for the action to be taken enter : add
When asked for the printer name enter : 513-lps
When asked for the printer type enter : remote
Give a synonym for this printer or type return
When asked to set the spooler directory (sd) it is suggested that you do not take the proposed default but set the spooler directory to:
```
	   /usr/spool/lpd/513-lps
```
When asked for the remote system name (rm) enter : springer
When asked for the remote system printer name (rp) enter : 513-lps
You will then be prompted with a list of possible variables to be added/modified in the file /etc/printcap. Refer the "ULTRIX Guide to System Environment Setup"(18) for a description of these variables. Type q if you wish to continue without making any modifications.

You will now be able to print on the LPS20 in building 513.

Go to the previous, next chapter.