The

UNIX Install

Manual

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

August 8, 1995

 

The UNIX Install Manual

 

Table of Contents

Pre-requisites......................... 3

Procedures for a New Software Install.. 4

Phase I........................... 4

Phase II.......................... 5

Phase III......................... 7

Communications Set-Up.................. 9

Test the Network Functions............. 12

Customizing a Staged Unix System....... 13

Create Standard UserIDs.and

Establish Kshell Defaults.... 13

Set X Windows Font Size........... 15

Set the Default Profile........... 15

Create Standard Directories....... 16

Configure E-Mail.................. 17

Complete the Install Process........... 19

Routing................................ 24

UNIX Token Ring........................ 26

Locally Administered Token Ring Address

UNIX Integrated LAN Driver........ 28

SNA Services/UNIX System 5, Release 4.. 29

Install Instructions for REEL

Backup/Librarian.................. 31

Oracle Install - Part I................ 34

Oracle Install - Part II............... 36

Tunables/Parameters.................... 38

Oracle Tunables................... 38

Semaphore Tunables................ 40

TIMEZONE Parameter................ 41

SPX/IPX Tuning Parameters......... 41

Patrol Install......................... 42

XCOM/UNIX System 5, Release 4.......... 44

Computer Manager....................... 54

OA-Admin............................... 57

Installation Verification.............. 59

Removing WIN-TCP....................... 60

Kernel Link Failure.................... 60

Additional Commands.................... 61

Patch Install Instructions............. 62

 

 

 

August 8, 1995

 

Pre-Requisites for a UNIX Install

 

1) Install diskettes appropriate to the machine being updated.

2) Administrative Guide: Command Line Interface General

Administration (Vols 1-3).

3) Specifications for the machine to be installed include the following:

a) Machine name (format USATLxxx),

b) TCP/IP address (from System Staging Document),

c) System address (from System Staging Document),

d) Selection of software to be installed,

e) Description of how the available disks are to be sliced.

The resource management group will eventually be responsible for providing this.

4) Labels. As you identify drives or machines, please place a

label on them for the next persons convenience.

 

If you need install assistance the AT&T GIS help line is

1-(800) 531-2222, our PIN ID is C O K E (or 2653).

Another important manual is the UNIX Install Guide, titled

Installing NCR UNIX SvR4 MP-RAS, Release 2.03.

NOTE: Do not attempt sections or portions of sections where you use vi (UNIX editor) unless you have experience, an experienced person with you, or you have a vi manual.

 

 

 

 

 

 

 

 

 

 

This document can be found in SAROS Document Manager (SDM). To access this Install Manual, enter SAROS Document Manager and open the following:

1) Information Processing,

2) Client Services,

3) IP Manuals,

4) UNIX Install Manual.

 

Procedures for a New Software Install

 

Verify where the root directory is located. To do this, boot (if it is not already active) the "factory" system. Then, go into sysadm, file system, and list to verify where the root directory is located.

 

PHASE I...

Insert Boot Diskette 1

Power off the machine and power back on. A system test will then be performed.

If this is a 3430 install, you will not get this next message. It will continue loading all boot diskettes and the order of the below listed prompts will be different. Otherwise...

WATCH CAREFULLY for the message:

"To interrupt autoload sequence press Space Bar within 5

seconds"

Press the Space Bar.

A menu "Startup Subsystem Services Main Menu" will appear,

Select option 7 for System Boot Services

A menu will appear for System Boot Services,

Select option 5 for System Processor Boot from Flex Disk,

"Booting the UNIX system".

In 5 minutes it will ask for disk 2.

In 1 minute it will ask for disk 3. Leave diskette (Flex

Disk) in the drive until prompted to remove it.

A prompt for the date will appear,

Press Enter if correct, follow the instructions if the date

should be changed.

An unnamed menu will appear for Installation, Maintenance, or Restore.

Choose option 1 "Perform Installation"

Another panel will appear for installation options,

Select option 1 for interactive install

There will be a prompt for System Name

Enter the System Name from your System Staging Documentation

(SSD). James Cook will provide this document.

A list of disks will be presented. Choose the first disk unless you have instructions to select a different disk. Check the SSD for the boot drive address.

Enter the disk that will contain root directories.

Record the disk device name for later reference.

A list of install media will be presented.

Choose the media that you have.

If a cartridge tape, it will be labeled "Base Install &

Packages, Tape 1.

A list of drives for the install media devices will be presented. If you do not know which drive is which, choose one. You will be given another chance to specify the drive.

Choose a drive, and place the media in the drive.

There will be a prompt stating that the hard disk will be reformated. "Do you want to re-format hard disk... (y/n)?"

Choose "NO" to continue the install.

If you do NOT reply ‘n’ it will take approximatley 20

minutes to reformat the hard disk. This is NOT necessary.

This prompt may appear at this point,

"Please press ENTER when ready or DEL to cancel the

installation,

Press Enter.

A screen which lists the partitions and cylinders per partition will be presented with a prompt to slice (Partition) the hard disk.

Respond by choosing "4". Exit. This will let the install

process slice the disk.

Another prompt will appear,

"Please press ENTER when ready or DEL to cancel the

installation,

Press Enter.

The system may re-boot at this time. The process to list the packages will take several minutes. If the system prompts

you to re-boot, follow the instructions it gives you.

 

PHASE II...

A list of packages to be installed will appear. This list nearly fills the screen. Use the arrow keys to move the cursor, follow the instructions at the bottom of the screen to install or spool packages.

Using the System Staging Documentation (SSD), select all the software to be installed. (If this machine will use Token-Ring TCP/IP, do NOT install WIN-TCP at this time.) When finished, select "X" for save and exit.

There may be an automatic re-boot at this point.

The following prompts will appear for the specified packages.

If SAT was selected

A prompt to respond to audit failures with a system shutdown will be presented.

Reply "No".

A prompt to relink the kernel after the SAT install will appear.

Reply "No".

If CSP was selected

A prompt to say that you must read the install documentation for CSP will be presented.

Reply "Yes" to continue the install.

A prompt about setting up mlsadmin will be presented.

Reply ‘Yes".

A prompt about setting up administrative groups for the LPA feature will be presented.

Reply "Yes".

A prompt to relink the kernel after the CSP install will be presented.

Reply "No".

If MSSF was selected for installation

A prompt for the prefix for an outside phone connection will be presented.

Press enter to take the default of 9.

Press enter again to confirm.

After the separate package prompts, the install will continue.

There will be a display of slicing information on the root disk.

A prompt to slice the root disk will be presented.

Check the root disk layout against the staging document (SSD). If there are differences, zero-out or reallocate the slices that are incorrect. Re-slice these to match the SSD*.

Enter Choice (slice number),

Enter size 0,

Enter Choice (slice number),

Enter size (from SSD),

Enter Starting sector,

Enter Path Name for Mount Point,

Enter File system type,

Repeat the above for each slice that needs editing.

* Note: Start at the bottom of the addresses (highest number),

so that the slices do not overlap. If they overlap,

the system will not change the addresses. You cannot

zero-out or reallocate swap, just enter the slice

number and size from the SSD.

Enter "Q" to confirm

You may get a prompt to re-edit the slices. If so, and there are no mistakes,

Enter "N" to prompt to re-edit slices.

Once the root disk layout matches the SSD, respond by choosing:

Enter "Q" to confirm.

There will be an 8 minute delay for the Automatic Boot Procedure.

 

PHASE III...

After the re-boot, a message may appear that 45 minutes will elapse before the install continues. This process can take up to 2 hours to complete. The first part will be for the products selected to INSTALL (PKGPLUS - Package Install), the second part is for the products selected for SPOOL (PKGPLUS - Package Spool).

Take a break.

The kernel will be re-built at this point and the system will need to be re-booted. The prompts will be automatically responded too. You will not be able to respond to them.

If you receive the following messages, see the "Kernel Link Failure" section later in this manual:

Error: cannot link edit unix

Error: kernel build failed

Trace/Log data saved in /var/spool/install/log.install

 

 

If WIN-TCP was installed and it usually is due to dependencies, reply ‘n’ to the re-configure message, and when the system is up, go to the section on "Removing WIN-TCP" in this manual.

When rebuilding the Kernel, if you get messages pertaining to CTKDB, you can install the package from the flex disk (1 of 11) which comes with the original boot disks. Use the PKGPLUS process to complete the install. CTKDB can put an overload on the system, so you may not want to install it.

Communications Set-Up

 

You will need to LOGON to the system at this point. The LogonID is ROOT, with the same password.

You will need to install LLC (Token-Ring LLC) and WIN-TCP in that order. The complete process is as follows:

1) Go to state 1,

2) Remove WIN-TCP,

3) Re-build the kernel,

4) Re-boot the system,

5) Go to state 1,

6) Install LLC,

7) Install WIN-TCP

Use the pkgadd command to add llc. llc is on a flex diskette. The command pkgadd -d /dev/rdsk/f03ht will show you what is on the diskette. When the message "Select package(s) you wish to install?" appears, enter "1" or "llc".

Use the pkgadd command to add WIN-TCP. WIN-TCP is on the install tapes. If the WIN-TCP package is on the spool simply enter

the pkgadd command with WIN-TCP. If you get a message indicating

WIN-TCP is NOT on the spool then do the following.

The command pkgadd -d /dev/rmt/device number will show you what is on the tape, where device number is the name of the tape drive. When the message "Select package(s) you wish to install?" appears, enter either the number that is associated with WIN-TCP or WIN-TCP.

Reply "y" to the build kernel message.

After installing WIN-TCP and the rebuilding of the kernel. Following the bootup will be a series of questions.

There will be a prompt to re-configure the network

Enter "Y" to re-configure network.

You may get a message asking if you want to use the saved

networking data files. Reply "n".

There will be a prompt to choose Domain Name Service

Enter "Y"

There will be a prompt to choose local or remote domain name server.

Enter "2" for remote server

 

There will be a prompt to create the resolver configuration file.

Enter "Y"

There will be a prompt for the complete Domain Name Server (DNS).

Enter the DNS HOST name from the SSD.

 

There will be a prompt for the IP address of the machine to send DNS queries.

Enter the DNS TCP/IP address from the SSD.

There will be a prompt to configure "slot 5" for the LAN connection. "Will this network interface be used for TCP?

Respond by entering "y".

(The slot number will vary by machine. Use the first slot for the TCP/IP connection.)

Follow the instructions in the SSD.

For an example if "slot 5" were given a reply of "yes", here are the prompts

- Enter the HOST NAME for the interface.

Use the SSD to enter the correct TCP/IP HOST name

- Enter TCP/IP (application) address

Use the SSD to enter the correct address

- Enter the network mask

Use the SSD to find the Subnet Mask

Enter "y" to confirm

- Enter the network name

Enter FIRST TWO INDEXES (octates) of the TCP/IP ADDRESS in

the SSD from the APPLICATION entry (i.e. 111.111 or 111.22)

At this point, you may be prompted with the question, "Will this network interface be used for TCP" again. This is for additional slots. If there are additional network cards , (i.e. there is a network card in "slot x", where x is the slot #) then reply "Y", for it’s configuration and follow the instructions in the SSD. If the system sees a Token Ring Interface in a slot, it will continue prompting. At this time, the maximum number of cards in any system is three (3). Therefore, the maximum number of prompts will be three (3). Please configure all slots. It takes a system down to configure the slots at a later time.

- Enter the SNDCF interface

Reply "N"

- Enter the SLIP

Reply "N"

- Enter the name to address translation library search list.

Enter "1,2"

This will choose the local host file first and then the DNS

file.

Go to the "UNIX Token Ring" section in this manual to complete the network configuration.

 

The Network Configuration and BOOT is now complete.

-----------------------------------------------------------------

Test the Network Functions

 

Test by the configuration by entering from the shell prompt:

Go to a PC with LAN WP and select the PING icon.

Enter the system name of the system being staged.

The message "USATLUxx is alive" should appear.

On the system being staged enter the following:

/usr/etc/ping 161.162.91.94

This command should repeat over and over with a message about the number of bytes transferred.

To halt the command use CNTL Z or the Delete key.

**This may not work until after the static setup. The ROUTING

must be done before the test from the ACC.

After the Token Ring cable(s) have been connected, have a UNIX team member at the ACC and one at TCP "PING" the box from their LAN WorkPlace menu for workstations. Use the TCP/IP address rather than relying on the Domain Name Server (DNS). One team member should also try using the TCP/IP address provided by the DNS. Use the box name rather than the TCP/IP address. The TCP/IP address should never need to be added tot he HOSTS file on a workstation.

If this fails,

1) And you get the message, "ping: socket: Protocol not supported", check to make sure the Token-Ring cable is connected, or that the network configuration (Token-Ring) is properly set-up for this system.

2) Or, you can re-configure the network by entering:

/opt/net/tcpip/script/tcpconfig

The prompts for network configuration will be re-displayed, reply as stated above.

3) You may get messages indicating PPA 0 notup, network startup failed. If you do one of the things to check is the status of configuration. Issue the following:

cd /etc

./ildconfig

Then make sure the configurations are set correctly. If you have to go into sysadm to correct the configuration, you will have to re-boot.

 

Customizing a STAGED UNIX SYSTEM

 

Enter ROOT

Assign the current ROOT password.

Call Resource Management (James Cook at 6-8522) and tell him that communication is available, and to slice the DASD as needed, or see the SSD for a complete disk layout. Use sysadm to slice the DASD and start the mirroring to the alternate boot disk. See the Administration Guide OA&M Menu Interface Guide for slicing, Chapter 11, page 17.

-----------------------------------------------------------------

Create Standard UserIDs and

Establish Kshell Defaults

 

To establish the minimum UserIDs, enter sysadm. For a list of the Function keys, see the "Additional Commands" section of this manual.

*Mirror boot requires Disk Array installed.

For mirroring, select storage devices. Then select disk array. And then, select mirror boot and follow the requests.

Select the users option and defaults suboption

Update shell to ksh, (use choices to select the directory). It should be /usr/bin/ksh. Update Login Inactivity to "365".

Press F3 to save, followed by F6 to cancel subpanel

Select the users option and the add suboption

1) Enter the name lname,fname in the Comments

2) Set the System Admin. privilege to "yes" for those users

requiring the System Admin. priviledge.

3) After saving, select the PASSWORD screen and set the

PASSWORD STATUS field to "password" and the MAX DAYS

field to "45"

The following people from IP and the DBA group are responsible for UNIX and/or UNIX/ORACLE support and need access to all UNIX hosts. Add the following list to this system under sysadm:

 

 

 

 

 

IP-Technical Assistance DBA's

Bourchier, John.....z8078a Anthony, Smith......z8416a

Clemons, Willie.....z9061c Bengochea, Gus......z8462a

Cleveland, Tom......o9061t Borna, Ramin........a00824

Hanger, Bill........o9080h Cohen, Barry........z84752

O'Neal, Fyphine.....z8213a Cooper, Claudia.....a00273

Russell, Greg.......z8090a Fearnley, David.....a00975

Minter, Charles.....o08117

IP-SSD, OS Mollen, Jim.........a01387

Hammill, Ken........z9061j Smith, Paul.........a01260

Sturdivant,Debra....a01275

IP-SSD, DTP Taylor, Kim.........a01187

Belton, Blair.......o9080m

Rosenblit, Isaiah...o9080r IP-Client Services

Cook, James.........o9061c

IP-BBSS Czekalla, Gunter....z9003h

Atchley, Gary.......z9061a Moody, Steve........z9061m

Dentice, Vince......o08290

Dorsey, Sarah.......o9061d IP-Provisioning

Harris, Mark........z8091a Dorsey, Sarah.......o9061d

Hunter, Ramona......z9061h Gibbs, Duncan.......z9080q

Johnson, Tami.......z9003j* Klein, Randy........z9061k

Stoller, David......z9080s O’Quinn, Dennis.....o9061o

IP-SSP

Ibarra, Ed..........z9061i

 

* Note: Tami Johnson’s user number is 2000.

 

Select the users option and change it to group. Add the group name xcomadm. This is required for the install of XCOM later in this process.

Press F7 for the Command menu, then choose EXIT

 

Set X windows Font size

 

Do not attempt this section unless you have vi (UNIX editor) experience.

To enlarge the font size on the console:

Change directory to the Xwindows directory

cd /usr/lib/X11/xdm

vi Xsession

Near the bottom, find the line that looks similar to:

"exec xterm -geometry 80 x 24 + 10 + 10 -ls"

and change it to:

exec xterm -fn 9x15 -geometry 80x24+1+1

(make a note of what the line read before you changed it.)

Also change the twm statement above this line to mwm to run

motif windows.

To prevent problems with the video resolution (i.e. Xwindows will not come up or the window is off center) and with the /etc/profile, edit the servers file in /usr/lib/X11/xdm. There is only one line in this file.

vi Xservers

:0 local /usr/bin/X11/Xnovt /usr/bin/X11/XR5vga :0 -lines 600

-virtes 800 600

This file says it is "read only". To override this, use the vi command w!. Quit the console session and log back on. Use LAN WP to logon and correct any syntax errors if Xwindows fails.

-----------------------------------------------------------------

Set the Default Profile

vi /etc/profile

Add the following lines at the bottom.

Typing notes

these lines are case sensitive

` is accent grave on the top left above the Tab key

" are double quotes. Word makes them look different

@ to enter the @, type \@

# is the comment indicator

The first line is only valid on machines that have a CD drive.

Move the export path statement to the last line in the file.

PATH="$PATH:/opt/onlib" #UNIX online library: "onlib &"

rev=`tput rev`

norm=`tput sgr0`

PS1="$rev$LOGNAME@`uname`$norm \$PWD>"

EDITOR=vi

export EDITOR

export PATH

The default key for the KILL function is "@". By adding the following statement, the KILL function is moved to the <Ctrl>K key and this allows the Backspace key to work as well. Also, this will solve some inconsistency problems. This statement can be added anywhere in the mainline code. The statement to add is as follows:

stty erase <control>V<control>H kill <control>V<control>K

What appears in /etc/profile will be: stty erase ^h kill ^k.

After typing these changes log back on. The prompt should be significantly different.

-----------------------------------------------------------------

Create Standard Directories

 

Refer to the SSD and create any file systems on the specified slices with the mount point names. Use the "File System Create" option to create the mount points. Otherwise, create them as directories off of / (root). If below has been created as file systems, do not create them as directories.

mkdir /zsoe

mkdir /zsoe/local

mkdir /zgenpk

mkdir /zgenpk/local

mkdir /zdatab

mkdir /zdatab/local

mkdir /zdatab/oracle

Configure E-Mail

 

To configure an E-Mail client do the following:

In /etc/mail/mailcnfg, add the following:

SMARTERHOST=mail.atl.ko.com

(Note: If you have to create mailcnfg, make sure to set decent permissions:)

chown bin /etc/mail/mailcnfg

chgrp mail /etc/mail/mailcnfg

chmod 644 /etc/mail/mailcnfg

Edit /etc/mail/mailsurr. Follow the instructions below in this part of the file to uncomment the line and remove the -N to make it look like so: (note, you should also comment out the /usr/bin/uux line since you don't want to use uucp).

# If using smtp, un-comment the second line.

# If running DNS name server, delete the "-N" option to smtpqer

# (it disables MX queries).

#'.+' '!([^!]+)!(.+)' '< /usr/bin/uux -a%R - \\1!rmail (\\2)'

'.+' '!([^!]+)!(.+)' '< /usr/lib/mail/surrcmd/smtpqer %R \\1 \\2'

# else

#'.+' '!([^!]+)!(.+)' '< /usr/bin/uux - \\1!rmail (\\2)'

Similarly, you want to uncomment the line that handles forwarding to a smarter host further in the file. Here is an example:

#If none of the above work, then ship remote mail off to a smarter host.

# Make certain that SMARTERHOST= is defined within /etc/mail/mailcnfg.

# If there is no smarter host, then routed mail fails here.

#

# ifdef NCR_V3

'.+' '!(.+[!@%].+)' 'Translate R=!%X!\\1'

# else

#'.+' '!(.+)' 'Translate R=!%X!\\1'

# endif NCR_V3

Now, run this command:

crontab -l >/tmp/ctab

Add the following to root's crontab (command is: crontab -e):

1,16,31,46 * * * * /usr/lib/mail/surrcmd/smtpsched

0 9 * * * /usr/lib/mail/surrcmd/smtpsched -c -w 1 -r 7

Note that any mistake in "crontab -e" can cause major problems. If something goes wrong that you don't know how to fix, get help as soon as possible.

Make your changes active by bouncing the smtp server:

/etc/init.d/smtpd stop

/etc/init.d/smtpd start

Send yourself a message on another server. The command is:

mailx server!userid (ex. usatlua0hb!z9061m)

To end entry of the message text, type <enter> and <Ctrl>d.

Verify that you received the message on the other server.

-----------------------------------------------------------------

 

Complete the Install Process

 

If a package is already installed, there is no need to install it again. Some of the following packages may be on Install Tape 2 or Bundle Tape 2. If they cannot be found on either of these tapes, use the diskettes. Be aware of package dependencies. Check the Product Install List or read the Install Manual - Appendix A for this information. Also, the System Transmittal Sheet may contain this information. An example of dependencies: If you are installing rdac, then dau must be installed first. If you must remove dau, then rdac must be removed before attempting to remove dau. Also, if LifeKeeper is on the system, it must be shutdown before removing rdac and it must be left down until everything is re-installed. Use pkgplus for dependency inform-ation instead of pkgadd. Pkgplus cannot be used if input is on multi-input devices, or requires responses not selected by pkgplus at pkginfo time.

Install the following packages as directed:

* Cacheopt, scde, and xcontrib.

From Tape labeled packages Tape 2, install the following

packages: Cacheopt, scde, and xcontrib.

Enter the following command pkgplus -d /dev/rmt/device

number or diskette drive name.

This will give a list of software on the tape.

Record the number for each package you wish to add.

Take next page option to list any additional software on the tape.

Record the numbers for each package you wish to add.

Once you've recorded number for all packages you wish to add, at prompt enter the number of each package you wish to

add separated by a comma.

Take defaults to prompts. However do not rebuild kernel after every install. This should be done after every 4 or 5 packages.

You can rebuild the kernel by issuing the following command:

/etc/conf/bin/idbuild and then shutdown -y -i6 -g0

* Onlib (diskette or Tape 1)

Place diskette containing onlib software in diskette drive. Enter pkgadd -d /dev/rdsk/f03ht (This will be referred to as repeat1)

Install onlib with defaults. You must reply "y" after

defaults.

* user32 (User license package) diskette

Repeat1

Enter "n" to rebuild kernel

* dau (RAID manager) diskette (must be installed before rdac)

If using EMC drives, rdac and dau must not be on the system.

Check SSD to see if you are installing EMC drives, if so,

skip this section. Otherwise...

Repeat1

At prompt for hour of the day to check parity

Enter "22" for hour

Enter "0" for minute.

* rdac (diskette)

See note above on EMC drives. Otherwise...

Repeat1

For rdac, a prompt for hour/minute will appear.

Enter "22" for hour

Enter "0" for minute.

* lk, lk3450b, lkcomm-ip, and sanity (LifeKeeper diskette)

Check the SSD to see if the LifeKeeper software needs to be

installed. If so,

Additional software may be located on a cartridge labeled

Bundle Tape 2. Place this tape in the cartridge drive.

Enter pkgadd -d /dev/rmt/device number

This will give a list of software on the tape.

Record the number for each package you wish to add.

Take next page option to list any additional software on the tape.

Record the numbers for each package you wish to add.

Once you've recorded number for all packages you wish to add, at prompt enter the number of each package you wish to

add separated by a comma.

Take the defaults on these packages.

* susErrpt (diskette)

Repeat1

Take defaults

 

* lk-oracle (diskette)

Install only if LifeKeeper (lk) is being installed. If so,

Repeat1

Select version 7

* vxfs (diskette or Tape 2)

Repeat1

Enter "y" for ? would you like vxfs interface and menus.

Enter "Y" for ? online manuals.

Rebuild kernel "y" or "n" (this is up to you)

* nwu (diskettes) - NETWARE (StarGroup Netware/X Server)

Note: NETWARE is required only if ORACLE has been installed.

There is currently a total of nine (9) diskettes for nwu.

Repeat1

Enter Server name?

Take default (same as machine name)

Enter path?

Take defaults

Enter # NVT clients?

Take defaults

Enter Internal Network Number? (last 8 characters)

Get from staging document (SSD)

Enter Network Number?

Get from staging document (SSD)

Enter the number of network boards to be used?

Enter "1"

Should board 1 be configured?

Enter "Y"

Is Board 1 Ethernet or Token Ring?

Enter "T" for Token Ring

Enter the Network Number for board #1

Get from staging document (SSD)

Should board 2-4 be configured?

Enter "no"

Ignore the warning about MP Activator.

Enter "y" to continue.

You will be prompted for a password for id console.

We are currently using password as the password

Enter password lower case.

If beaconing occurs, follow the instructions below:

/opt/netware/sconsole (menu system)

Select ADMINISTRATOR TRANSPORT.

Shutdown TRANSPORT & SERVICES.

Go to Config menu and correct network number (8 characters).

Then, restart TRANSPORT.

* nwu+mp (diskette) - MP Activator, install after nwu

Repeat1

Take defaults

* crypt (diskette)

Repeat1

Take defaults

* diskarray (diskette)

Repeat1

Take defaults

* diskarray patch (diskette) Install after diskarray

Repeat1

Take defaults

This package will automatically rebuild the kernel.

* MF-COBOL (Cartridge)

Check the Software List to see if this product is to be

installed. If so,

Pkgadd -d /dev/rmt/device number

Take defaults

Edit with vi and add the following to /etc/profile after

export EDITOR:

COBDIR=/opt/lib/cobol

export COBDIR

PATH="$PATH:/opt/lib:$COBDIR/bin"

export PATH

LD_LIBRARY_PATH=$COBDIR/coblib

export LD_LIBRARY_PATH

 

 

If this box has an EMC Disk Array attached, install the EMC Disk Array Support Software that was supplied as a separate document.

 

Routing

 

ROUTING is the process of mapping a path to send a packet to its destination internet address. Routing can be direct or indirect. Direct Routing is used when the source and destination nodes are on the same physical network. When direct routing is used, the source node sends the packet on the network with the destination hardware address in the link layer protocol header. The destination node hardware detects its own address in the packet header and accepts the packet.

Indirect Routing (Static Routing) is used when the source and destination nodes are not on the same physical network. The source node uses its routing tables to determine which gateway will forward packets to the destination node. The packet is put on the network with the gateway’s hardware address in the network header, and the destination’s internet address in the IP header. The gateway receives the packet from the network, and the gateway IP software determines the packet’s destination from the IP header.

Adding a Static Route

Within the AT&T System V Release 4 and AIX IP environments there can only be one default gateway per system. A gateway is the method to communicate with another network. There are multiple IP networks at The Coca-Cola Company. The AT&T UNIX and AIX systems in the TCP computer room are located on the 161.162 network. ACC workstations are located on the 185.153 network and the UNIX system, USATLUZ0, is located on the 185.88 network.

The UNIX systems at TCP within the 161.162 network need to have a gateway address configured which will allow the system to communicate with the ACC 185.153 and 185.88. This is performed by adding a static route. A static route must be added to each and every UNIX box that must communicate with this network.

For AT&T Machines:

If you are at TCP, do this section. Else, go to the ACC section below.

From the command line on the TCP 161.162 network, type:

/usr/etc/route add net NETWORKADDRESS GATEWAYADDRESS METRIC

/usr/etc/route add net 185.153.0.0 161.162.11.10 1

This informs the system to utilize Corporate gateway.

/usr/etc/route add net 185.88.0.0 161.162.11.10 1

this informs the system to utilize gateway address 161.162.11.10 to communication with the 185.153 and 185.88 networks This will do the trick until the system is booted.

The method to have this static route active after boot time is: vi /etc/gateways.eth and add the following lines:

# Corporate router address to access network 185.153 via one hop.

/usr/etc/route add net 185.153.0.0 161.162.11.10 1

/usr/etc/route add net 185.88.0.0 161.162.10.1 1

/usr/etc/route add default router 1

to this file. Please insert comments into this file describing the network and the gateway. The 1 after the gateway address is the metric (or hop count).

At the ACC network (and etc. for any box on any new sub-network), type:

/usr/etc/route add net 161.162.0.0 185.88.1.10 1

/usr/etc/route add net 185.153.0.0 185.88.1.10 1

vi /etc/gateways.eth and add the following line:

/usr/etc/route add net 161.162.0.0 185.88.1.10 1

/usr/etc/route add net 185.153.0.0 185.88.1.10 1

/usr/etc/route add default router 1

Then run the /etc/gateways.eth script.

 

For AIX Machines:

SMIT

Communications Applications and Services

TCP/IP

Further Configurations

Static Routes

Add a Static Route

Destination Address NETWORKADDRESS

Gateway Address GATEWAYADDRESS

For Both AT&T and AIX Machines:

You will need to vi /etc/hosts file to include a line that says either: 161.162.1.1 router (TCP), or 185.88.1.10 router (ACC).

ex.

vi /etc/hosts

161.162.1.1 router (TCP)

185.88.1.10 router (ACC)

vi /etc/gateways.eth

/usr/etc/route add default router 1

 

UNIX Token Ring

 

Display Adapters

/etc/ildconfig (Issue this command to get the PPA number.)

Provides the slot number that contains the adapter; the relative physical position of the adapter port beginning with zero; current state of the PPA (UP, DOWN, DISABLED, or BAD; MCA board ID; MAC type of physical media being used; RATE transmission rate; DESCRIPTION type of adapter; IO I/O address used by the adapter; IRQ interrupt request level; MEM MCA shared memory at start only; ARB bus arbitration level; NODE address of the adapter - will display the burn-in address if no change has been made to utilize a locally administered address.

Display Default Configuration

/etc/ildcfg -d -p ppa

Display Current Configuration

/etc/ildcfg -D -p ppa

Change Configuration Values

Create a file which contains the token ring timing and adapter address values to be changed

vi /tmp/infile

nodeaddr=400069xxxxxx # where xxxxxx=ring number plus

unique number specific to segment

stationmax=5 # the total number of SNA links which

will be open at one time

t1_1=50 # response timer

t2_1=60 # receive acknowledge timer

ti_1=200 # inactivity timer

Verify changes:

/etc/ildcfg -T -i /tmp/infile -p ppa # substitute the ppa number you wish to change

Write to configuration file:

/etc/ildcfg -c -i /tmp/infile -p ppa # substitute the ppa number you are changing

Load new configuration:

Ensure any packages using the PPA are shutdown (SNA, WIN-TCP, NETWARE, etc.)

/etc/init.d/ild stop # stops all PPAs

/etc/ildinit -u ppa # Stop a single PPA by substituting the

PPA number

/etc/init.d/ild start

If you need to restore the original configuration:

/etc/ildbldcf > /etc/ild/ildcf

ln /etc/ild/ildcf /etc/ildcf

After the Token Ring adapters have bee configured, verify that each Token Ring adapter microcode has been configured with the speed (4Mb vs. 16Mb), etc., to match the local ring network.

Contact a UNIX network specialist if you need assistance.

Make sure the Token Ring cable(s) are connected to the Token Ring cards which have been configured. If the Token Ring cards have not been configured, do not connect them. They may "beacon" the network which slows or stops all other network traffic on that ring.

 

Locally Administered Token Ring Address

UNIX Integrated LAN Driver

 

Show current status of interface

/etc/ildconfig

Create file with required changes

vi /tmp/infile

nodeaddr=400069xxxyyy where 69=UNIX xxx=ring number and yyy=unique address on segment

stationmax=5

t1_1=50

t2_1=60

ti_1=200

Verify changes

/etc/ildcfg -T -i/tmp/infile -px where x=PPA number (-p0 for card 0, -p1 for card 1, -p2 for card 2). There must be a space between I and /tmp/infile. The word document would not allow a lowercase I with a space

Write changes to the configuration file

/etc/ildcfg -c -i/tmp/infile -px where x=PPA number. There must be a space between I and /tmp/infile. The word document would not allow a lowercase I with a space following.

Shutdown interface

/etc/init.d/ild stop

Load new configuration

/etc/init.d/ild start

Check changes

/etc/ildcfg -D -px where x=PPA number

-----------------------------------------------------------------

SNA Services/UNIX System 5 Release 4

 

* snasvc (Cartridge)

Note: SNA is required for XCOM. snasvc must be installed first, then the rest of the SNA packages can be installed.

 

Pkgadd -d /dev/rmt/device number

Insert a cartridge tape into Tape Drive.

Type [go] when ready,

or [q] to quit: "go"

Note: snasvc must be installed before any other modules on this tape. snasvc NCR sna services: PU - T2.1 Base, APPC/LU6.2, sna manager (i486, MCA) is the current full name of the product.

Select package you wish to process (or 'all' to process all packages.) (default:all) [?,??,q]: "4" (is the current number)

Processing package instance <snasvc> from <tape device name>

A message expressing concerns about the current run level of the machine will be displayed if the network is up. The message will make recommendations to install the package at a single user run level. Do NOT be concerned about this message. There should be concern only if the package is already installed and clients are using the software.

Do you wish to continue with the installation of this package [y,n,?} "y"

 

A warning message will be sent:

WARNING: This release of HDLC driver 1.04 for SDLC and QLLC link types which use the MCIA adapter board is not compatible with earlier releases of the HDLC driver. This new HDLC driver is not required if you use an MPCA adapter. Refer to this release definition (/usr/RELDEF/snasvc.rd) for further information. (Currently, as token ring is the interface being utilized and not X.25, this message is nothing to be concerned about).

Do you want to continue with the installation (y/n)? "y"

 

 

 

* snamgr (Cartridge)

The default directory for snamgr is /home/snamgr. Do you want a different home directory for snamgr? (y/n) (default=n): "n"

Kernel tuning parameters for LU6.2 subsystem.....

Enter choice (default=50): "50"

Do you wish to rebuild the kernel now?

Yes or No [y,n,?q] "n"

 

* csv (Cartridge)

Press enter and wait for prompt. Continue with installing module number 1: Common Service Verb API

Insert a cartridge tape into Tape Drive.

Type [go] when ready

or [q] to quit: "go"

Select packages you wish to process (or 'all' to process all packages). (default: all) [?,??,q]: "1" (is the current number)

Installation of <csv> was successful.

Insert a cartridge tape into Tape Drive.

Type [go] when ready,

or [q] to quit: "q"

An idbuild and a boot of the system is required to activate these packages. See Additional Commands on how to do an idbuild (rebuild kernel).

 

 

Install Instructions for

REEL Backup and REEL Librarian

 

There are two tapes for the REEL products, one is labeled REEL Backup and the other is REEL Librarian. The tapes are in a gray box on a shelf below the other install tapes. Please put them back in the box when you are finished.

- Logon to ROOT

- Create directories.

"mkdir /tmp/reelb"

"mkdir /tmp/reell"

- copy the tapes to the new directories.

"cd /tmp/reelb"

load the REEL Backup tape in the drive

"cpio -icdvB < /dev/rmt/c-t-d-s-" fill in the drive name

After this completes, execute the install script to expand

the install files.

Remove the tape from the tape drive.

"sh RUNME"

"cd /tmp/reell"

load the REEL Librarian tape in the drive

"cpio -icdvB < /dev/rmt/c-t-d-s-" fill in the drive name

After this completes, execute the shell script to expand the

files.

Remove the tape from the tape drive.

"sh RUNME"

- The tapes have now been copied. The install from this point

is like most other package adds. REEL Backup must be

installed first.

"pkgadd -d /tmp/reelb rlb"

- Answer the install script questions.

Is this a Crash Recovery? "N"

Is this a New Installation? "Y"

Enter path for the REEL environment file: /etc

press enter to accept the default.

The contents of the environment file will be displayed.

Library Directory: /zgenpk/local/reel/lib

Binary Directory: /zgenpk/reel/bin

Server Machine: usatluaz

Program Number: 673095718

Client Name: HOST

Is this correct? (y/n)

On a new install the display should not match, reply "N"

You will be prompted for each of these items:

/zgenpk/local/reel/lib Is the directory for machine

specific parms,

/zgenpk/reel/bin Is the directory for LOADLIBS,

usatlua2 Is the name of the master machine,

673095718 The remote procedure call id on

usatluaz,

HOST Tells REEL that this is a client

install.

The script will redisplay the contents and ask:

Is this correct (y/n) "y"

There will be two (2) prompts:

Directories will be created? (y/n) "y"

Install on-line manual pages (y/n) "y"

Should the kernel be rebuilt and the system rebooted (y/n)?

Reply "n"

This install is for a Client, ok? (y/n) "y"

The install will now proceed for about 50 messages. The last message is:

Installation of <rlb> was successful

 

REEL Backup is installed now install REEL Librarian.

"pkgadd -d /tmp/reell rll"

- Answer the install script questions.

Is this a separate installation? "N"

Is this a Crash Recovery? "N"

Enter path for the REEL environment file: /etc

press enter to accept the default.

The contents of the environment file will be displayed.

Library Directory: /zgenpk/local/reel/lib

Binary Directory: /zgenpk/reel/bin

Server Machine: usatluaz

Program Number: 673095718

Client Name: HOST

Is this correct? (y/n)

Since we just installed REEL Backup these parms should

match, reply "y"

Install on-line manual pages (y/n) "y"

This install is for a Client, ok? (y/n) "y"

The install will now proceed for about 50 messages. The last

message is:

Installation of <rll> was successful

 

- The REEL products have been installed, but they have not been

started. Also the system startup needs to be modified to

always start REEL after every IPL. While still in root:

cd /etc/rc3.d

ftp usatluaz (must have ID on UAZ to to transfer)

get /etc/rc3.d/S90Reel

<CTRL D>

You can now do an ls to verify that the file was transfered.

To start REEL and test the startup shell.

sh S90Reel

The REEL environment file will be listed.

A message about daemon started, will appear.

A message saying 1 process started will appear.

 

Congratulations! REEL is installed and started.

ORACLE Install - Part I

 

Oracle Support 1-(415)-506-1500 CSI for ATT=632-372

This procedure is for a NEW Oracle 7.1.4.1.1 install

on an AT&T Unix.

These are the products installed:

Oracle 7.x.x.x Dist. Data

Oracle 7.x.x.x PL/SQL

Oracle 7.x.x.x RDBMS

Oracle Terminal

Pro*C

Pro*Cobol

SPX/IPX Protocol

SQL*NET SPX; SQL*NET TCP; SQL*NET V2

SQL*Plus

Multi-Protocol Interchange

Oracle Names

Oracle Network Mgr.

Oracle Server Mgr.

TCP/IP

1) Prereqs:

TCP/IP address - from Staging Doc;

root passwd

Insure 1 gig mount point of /zdatab/oracle (dfspace)

Insure id=oracle group=dba exists (under sysadm)

(with a home of /zdatab/oracle)

2) Update System Files

A) vi /etc/default/login -

set ULIMIT=unlimited (as root)

B) vi /etc/services - add

orasrv 1525/tcp #oracle tcp server

C) Insure /etc/conf/cf.d/stune is OK by comparing it to the

Oracle Tunables, (see Section on Oracle Tunables in this

manual). Make sure mtune is also OK. Copy mtune and

stune to .old, then ftp these files from another machine.

3) On the Source Machine (where the data you want to copy exists,

in this case USATLUA1, USATLUA5 or USATLUZ0 can also be used;

UZ0 takes extra time).

go into sysadm

select network_services

select remote_files

select local_resources

select share

select nfs

enter /zdatab/oracle

f3, f3

should see messages directory is shared

exit sysadm (f7)

log off Source Machine

4) On Install Machine, issue a cd/

Create temp mount point = mkdir /tmpora

(ensure oracle id,dba grp id own tmpora dir)

chown oracle tmpora

chgrp dba tmpora

5) On Install Machine

go into sysadm

select network_services

select remote_files

select remote_resources

select mount

select nfs

enter Server = usatlua1

Remote path = /zdatab/oracle

Mountpoint = /tmpora

f3, f3

should see messages directory is mounted

exit sysadm (f7)

6) cd to /tmpora

the following command will copy all data from

usatlua1:/zdatab/oracle which is mounted as /tmpora

find . -print|cpio -pudmv /zdatab/oracle

Don't worry about error messages.

if necessary: change file ownership, at /zdatab/oracle

(shouldn't be - verify oracle,dba owner)

enter the following:

find . -user xxx -exec chown yyy {} \;

where xxx is uid currently owning the file

where yyy is uid you want to own the file

find . -group xxx -exec chgrp yyy {} \;

where xxx is grpid currently owning the file

where yyy is grpid you want to own the file

7) unmount from install machine (reverse of step 5)

unshare from source machine (reverse of step 3)

--------------------------------------------------------------

- End of ORACLE Part I Staging - Notify Database Team Leader -

--------------------------------------------------------------

ORACLE Install - Part II

 

This section is to be completed by David Stoller, Isaiah Rosenblit, or Blair Belton.

1) do a du compare of the directories

2) mkdir /var/opt/oracle -

ftp a /var/opt/oracle/oratab

edit SIDs

(ensure oracle id,dba grp id own oracle dir)

chown oracle tmpora

chgrp dba tmpora

3) login as oracle (su - oracle (from root)

you might get errors - don't worry)

hit enter at the two prompts

4) vi .profile

ensure that the ORACLE_HOME variable is correct

cntrl d out of shell, reenter (su - oracle)

this will run the modified .profile

5) login as root for the following:

cd /zdatab/oracle/product/7.1.4.1.1/orainst

execute the following pgm

./root.sh

the following variables are checked:

1. ORACLE_HOME, ORACLE_OWNER, ORACLE_SID

enter "y" if correct

2. checks for /opt/bin

press enter

3. ORACLE_HOME will not match, this is fine.

press "y"

Now check results:

1. cd /zdatab/oracle/product/7.1.4.11/bin

enter ls -l|grep rws

2. you should see four programs:

ntisbsdm, orasrv, osh, spxsrv

6) login as oracle again (su - oracle)

7) execute the following programs:

spxctl start

tcpctl start

8) Verify the instance can be brought up. If the instance comes

up the install was successful.

run sqldba,

connect internal

startup nomount

run sqlplus,

for SPX at prompt>system/manager@X:usatluz0:IPDB

for TCP/IP at prompt>system/manager@T:185.88.2.61:IPDB

9) Now create the database. DBA will provide information to

complete the INIT.ORA modifications.

10) Update the Oracle Server Document to include information

for installed server.

11) Neccessary on Non-LifeKeeper Systems Only.

Incorporate autostart on machine /etc/rc3.d

S91OraListen

12) place oraenv, dbhome in /usr/bin

from $ORACLE_HOME/bin

-----------------------------------------------------------------

 

Tunables

 

ORACLE

04/28/95

These TUNABLES include Oracle and CA Unicenter. This is the lowest acceptable TUNABLE level. A higher level is generally better.

*ident "@(#)master:master.d/stune 1.4.1.4"

MAXUP 400

SCPULIM 0x7FFFFFFF

HCPULIM 0x7FFFFFFF

SFSZLIM 0x7FFFFFFF

HFSZLIM 0x7FFFFFFF

SDATLIM 0x7FFFFFFF

HDATLIM 0x7FFFFFFF

SSTKLIM 0x7FFFFFFF

HSTKLIM 0x7FFFFFFF

SCORLIM 0x7FFFFFFF

HCORLIM 0x7FFFFFFF

SFNOLIM 0x400

HFNOLIM 0x400

SVMMLIM 0x7FFFFFFF

HVMMLIM 0x7FFFFFFF

NPROC 3080

NCALL 700

UFSNINODE 3000

SEGMAPSZ 0

BUFHWM 0

NBUF 512

SEMMAP 1000

SEMMNS 2500

SEMMNI 100

SEMMNU 1000

SEMMSL 25

NINODE 500

SHMMNI 500

SHMMAX 198427136

NUMTIM 2009

NUMTRW 2000

PRFMAX 10240

ILDMAXOPENS 4096

ARP_NDEV 33

ARP_TABSIZE 2000

IP_NDEV 256

TELNET_NDEV 1250

WTTY_NDEV 700

ARP_TIMEOUT 5

IP_NROUTES 64

RAW_NDEV 32

SLIP_NDEV 4

TCP_KEEPINT 600

NAIOSYS 512

TCP_NDEV 2000

UDP_NDEV 912

NFILE 1200

STRTHRESH 0

FLCKREC 1000

NHBUF 1024

SEMUME 20

SEMOPM 14

ARG_MAX 51200

MAXAIOS 20

MSGMAX 16500

MSGSSZ 128

MSGSEG 1024

SHMSEG 30

MSGMNB 65535

MSGTQL 1000

MSGMAP 200

LISTIO_MAX_CNT 1024

This last one is machine dependant (until post OA Admin configuration):

SHM_NAILED_GID1 803

the 803 is the group-id of the dba group (id oracle runs under).

 

UNIX SEMAPHORES SETTINGS

There is currently no standard on the Semaphore resources on the UNIX machines. These resources are used to help performance in allocation during execution. The Semaphores are located in the /etc/conf/cd.d directory and the member is stune. The default tunables are in the mtune member. There are five Semaphores which should be changed to increase performance of the UNIX machines:

1. SEMMAP: This parameter specifies the number of entries in the control map used to manage semaphores. This map is used to keep track of free areas in the system pool of semaphores.

2. SEMMNS: This parameter specifies the maximum number of semaphores permitted in the system.

3. SEMMNU: This parameter specifies the maximum number of undo structures in the system.

4. SEMMSL: This parameter specifies the maximum number of semaphores per semaphore identifier.

5. SEMMNI: This parameter specifies the maximum number of semaphore identifiers in the kernel. This is the number of unique semaphore sets that can be active at any given time.

Resource Management makes the following recommendations for standard settings for these options:

SEMMAP 1000

SEMMNS 1000

SEMMNU 1000

SEMMSL 25

SEMMNI 40

The formula for these calculations is:

SEMMAP = SEMMNS

SEMMSL * SEMMNI = SEMMNS

SEMMAP = 1000 (MAXIMUM allowed)

If there are any questions, please contact the Resource Management group.

 

 

 

 

TIMEZONE

The TIMEZONE parameter specifies the number of minutes that Eastern Standard time differs from Greenwich Mean Time (GMT). This parameter is set in /etc/conf/cf.d/mtune.

· Range: 0 -1440

· Default: 480

The default should be changed to 300 to indicate Eastern Standard Time.

 

 

SPX AND IPX TUNING PARAMETERS

SPX and IPX tuning parameters should be set to allow a production level of concurrent users. The directory is /etc/conf/pack.d/spx and the member names are spx_tune.h and ipx_tune.h. The following are the ORACLE recommended values:

vi spx_tune.h

Change SPX_MAX_CONNECTIONS to equal 550

:wq

vi ipx_tune.h

Change IPX_MAX_SOCKETS to equal 400

Change SPX_MAX_SOCKETS to equal 200

:wq

The kernel must be rebuilt after these changes are made for them to take effect.

-----------------------------------------------------------------

 

Patrol Install Instructions

 

1) Create a patrol logonid if one does not already exist.

Attributes - group = dba (you may have to create dba group)

sysadm = yes

password = (check with a patrol knowledgeable person for the

standard password)

2) Logon as root

mkdir /zgenpk/patrol

chown patrol patrol

chgrp dba patrol

3) Logon as patrol

vi .profile

Add these lines

PATROL_HOME=/zgenpk/patrol

PATH=$PATH:/zgenpk/patrol

export PATH PATROL_HOME

:wq

cd /zgenpk/patrol

ftp usatluu2

logon as patrol

mget *.tar

reply "y" to numerous confirmation messages.

(there may be an mget option that will suppress these

prompts)

^D (control-D) when complete to exit ftp

On the target system

Check that /zgenpk/patrol is where the files ended up, then

tar xvf install.tar

sh install.sh

You are now at the Patrol install menu.

Select "change source" and set the source directory to

/zgenpk/patrol and then return to the Main Menu

Select "view" products

Select some products, (use "+" to go to the next page)

Choose the "agent" for "SVR4-i386"

Backup to the main menu

Select "install"

Reply "y" to install products

Reply "n" to the prompt to remove tar after unpacking

At the end of the install is a comment about a

"configure" shell.

Logon to root

/zgenpk/patrol/configure

Enter 3 when menu appears

While still logged on as root:

cd /etc/rc3.d

pwd to check directory

ftp usatlua1 as yourself

get /etc/rc3.d/S92patrol

^D

ls to check

sh S92patrol

 

Check the install

logon to usatluaz (the master console for Patrol)

cd /zgenpk/patrol

Patrol -p 2023

logon to patrol as yourself

Be patient, initialization is very slow

On the Patrol display select Edit

Edit Computer, select Add, select SVR4-i386, set the name to

the new machine name.

A new icon should appear for the machine you are staging.

-----------------------------------------------------------------

 

XCOM V02.05.02b

UNIX System V Release 4

 

Installation

Requirements:

Installation diskette

Super user privilege (root)

User Modification

The xcompp script, which renames the arguments of the XCOM process has been modified. The script xcompp uses the case command to execute application scripts/ programs based on the local file name passed in the arguments. The file argument is checked (all or part) and processing is passed to the script/program defined (full path name is used).

When you are installing a new version of XCOM you must remember to save xcompp script which is found in /usr/lib/xcom.

You also will want to save the configuration, session, and global members of /usr/spool/xcom/config in order to maintain the current production parameters.

Make sure the group name xcomadm has been added via sysadm before installing the XCOM package.

If you are using 6298 Disk Array you will need to edit the line /etc/conf/pack.d/vdisk/space.c

change

int vd_rdac =0

to int vd_rdac =1

 

Load Procedure(s)

a. pkgadd -d /dev/rdsk/f03ht

Insert diskette into 3.5 inch Mbyte (High Density)

Type [go] when ready,

or [q] to quit: go

b. Select package(s) you wish to process (or 'all' to process

all packages). (default: all) [?,??,q]: all

Press Enter

c. The installation continues, when complete Type q and press

enter.

d. The documentation is in error. LU6RECV, LU6SEND, XCOMRECV,

and XCOMSEND must be moved to /usr/lbin from /usr/lib/xcom.

 

Configuration

File which controls the number of sessions

Parallel session usage: Edit the xcom.ses file. Add the LU6.2 LU name and the number of sessions desired.

vi /usr/spool/xcom/config/xcom.ses

Example: ARxxx#xx=10

File which controls UNIX outbound

Change/add default parameters in the xcom.cnf file.

vi /usr/spool/xcom/config/xcom.cnf

CODE_FLAG=EBDIC

NUMBER_OF_RETRIES=25

RETRY_TIME=600

TMPDIR=/dev/null

XLUNAME=ARxxx#xx Replace the x's with the LU name.

Notify Gary Atchley to set-up the Network Parameters on the production or test mainframe systems.

 

Testing

xcom62 -c1 -f /usr/spool/xcom/config/xcom.cnf \

LOCAL_FILE=full path of a local file to be sent REMOTE_FILE=dataset name \

USERID=your userID PASSWORD=your password

xcomqm -La will list all queue entries. Notice that there is a transmission number associated with the display.

xcomqm -Dxxxxxx were x=transmission number

XCOM Configuration Files

 

List of Parameters

/usr/spool/xcom/config/xcom.cnf and xcom.glb. Please ensure /usr/spool/xcom/config is in the path of your .profile or shellscript

ALLOCATION_TYPE CYL,TRK, or BLK [default CYL] CYL - Cylinders, TRK - Tracks, BLK - Blocks

AT0E_FILENAME 0 - 256 characters [default none] The name of the file containing the ASCII-to-EBCDIC character conversion set.

BLKSIZE 0 - 32767 [default 00800] Used when FILE_OPTION=CREATE. Specifies the block size of the dataset. For a fixed or fixed block record format, this value must be a multiple of the Record Length. For a variable record format, it must be 4 bytes larger than the Record Length. For undefined record formats, it must be larger than the largest Record Length.

CARRIAGE_CONTROL_CHARACTERS ASA, IBM, or OTHER [default OTHER] For report transfers only. Indicates the type of printer carriage control codes, if any, that are included in the report file. ASA - ASA control codes in column 1. IBM - IBM Machine Characters (valid only for IBM mainframes). OTHER - No carriage control codes.

CARRIAGE_FLAG YES, NO, MPACK, LPACK, TPACK, or VLR [default YES] YES - Indicates that the transferred file is a text file and a new line character should be added to the end of incoming records. Also, new line characters will be removed from the ends of lines before an outgoing record is sent. NO - Indicates no special processing. MPACK - Indicates a text file with record packing. LPACK - Not supported in this release. TPACK - Not supported in this release. VLR - Indicates a binary file of variable length records with a 4 byte length field preceding each record. NOTE: The parameter MPACK does not support MAXRECL (actual record length) over 2k.

CHECKPOINT_COUNT 0 - 9999 [default 1000] Defines how often (based on record count) the sending systems will request a checkpoint to be taken. 0000 denotes no checkpointing.

CLASS 1 character [default none] For report transfers only. Indicates the print class to be assigned to this report on the remote system. If the remote systems is an IBM mainframe, then this field designates the JES SYSOUT class. If this is the case, enter A to print the report through SYSOUT=A.

CODE_FLAG ASCII, BINARY, or EBCDIC [default ASCII] Identifies the type of data being transferred. Choose from the following: ASCII - Indicates that an ASCII file is being transferred. BINARY - Indicates that a binary file is being transferred. This indicates to a remote system that it is not to translate the data it is exchanging with your system. EBCDIC - Indicates that an EBCDIC file is being transferred. The transferred data will translate from EBCDIC to ASCII format when receiving the data, and from ASCII to EBCDIC format when sending the data.

COMPRESS YES or NO [default YES] Indicates the compression type. YES - Indicates run length encoding of binary zeros and blanks only. NO - Indicates no compression. NOTE: When communicating with an IBM mainframe, if the data file contains only empty lines, COMPRESS must be set to YES.

CONVERT_CLASSES 1 - 64 characters [default none] For incoming report transfers only. A character string containing print classes for which EBCDIC to ASCII conversion will be performed.

COPIES 001 - 999 [default 001] For report transfers only. The number of copies of the report to be printed.

DEBUG_FLAG YES or NO [default NO] Specify YES to cause the transaction to print a trace to strerr.

DESTINATION 0 - 16 characters [default none] For report transfers only. Indicates the printer or other device on the remote system to which the report is to be sent. If not specified, the remote system will send the report to the system's default printer.

DESTINATION_TYPE HOST, MIDRANGE, or OTHER [default OTHER] Indicates the type of target system.

DISPOSITION DELETE, KEEP, or HOLD [default DELETE] For report transfers only. Indicates what the remote systems should do with the report file after the report has been printed. This field is not used when the remote system is an IBM mainframe. DELETE - Delete after printing. KEEP - Keep after printing. HOLD - Hold after printing.

EOL_CLASSES 1 - 64 characters [default none] For incoming report transfers only. A character string containing print classes for which an ASCII newline character will be appended to each record.

ETOA_FILENAME 1 - 256 characters [default none] The name of the file containing the EBCDIC-to-ASCII character conversion set.

EXPIRATION_TIME 0 - 5 characters [default 600] The maximum time in seconds that a transfer can be held in the queue.

FCB 0 - 4 characters [default none] For report transfers only. Indicates the FCB JCL parameter when sending the report file to an IBM mainframe, defining print density, lines per page, etc.

FILE_OPTION CREATE, APPEND, or REPLACE [default CREATE] For file transfers only. Indicates how the transferred data is to be processed by the receiving system. CREATE - Create a new file on the receiving system. APPEND - Append this data to an existing file on the receiving system. REPLACE - Replace an existing file on the receiving system. When creating a file on an IBM mainframe system, some additional information may be necessary. See explanations for RECORD_FORMAT, BLKSIZE, VOLUME, and UNIT parameters for more information.

FORM 0 - 10 characters [default none] For report transfers only. Indicates to the remote operator or user the type of form on which to print the report. Since XCOM 6.2 places the print job in the remote system's print queue, the print control functions will depend on the remote system. You must verify that the type of form you are requesting is available at the remote site.

HOLDFLAG YES or NO [default NO] For report transfers only. Indicates whether this transferred report file is to be placed on HOLD status (HOLDFLAG=YES) on the remote system or is to be printed immediately.

INTERNAL_CONVERSION_TABLES YES or NO [default YES] Indicates to xcomtp whether or not to use the internal code conversion tables. If NO is specified the tables contained in ETOA_FILENAME and ATOE_FILENAME will be used.

LOCAL_FILE 0 -256 characters [default none] For send file transfers only. The name of the file on the local system that is being transferred. If this variable is null or unset, standard input is read. In this manner, XCOM 6.2 commands can be used in a pipeline or with redirection. All UNIX file naming conventions apply. NOTE: User must specify full path name if QUEUE=YES.

LOCAL_FILE_RF 0 - 256 characters [default none] For retrieve file transfers only. The filename which is created, appended, or replaced on the local system when it retrieves a file. All UNIX file naming conventions apply. If this value is null or unset, then XCOM writes to the standard OUT file. NOTE: User must specify full path name if QUEUE=YES.

LOCAL_FILE_SJ 0 - 256 characters [default none] For job transfers only. Indicates the name of the file on the local system to be sent. All the UNIX file name conventions apply. If this value is null or unset, then XCOM reads the standard input file. NOTE: User must specify full path name if QUEUE=YES.

LOCAL_FILE_SR 0 - 256 characters [default none] For report transfers only. Local filename to be sent as a report to the remote system. If this value is null or unset, then XCOM read the standard input file. NOTE: User must specify full path name if QUEUE=YES.

LOCAL_NOTIFY 0 - 64 characters [default none] The users on the local system who are to be notified that XCOM has completed this transfer. XCOM uses the NOTIFYL parameter to determine how to notify them. The system default is the string contained in the LOGNAME environment variable.

LRECL 0 - 32767 [default 160] Specifies the actual or maximum length, in bytes, of a logical record. this parameter corresponds to the JCL LRECL subparameter. For variable blocked record format, this value should equal the maximum record length + 4. for fixed or fixed block record formats, this value should equal the constant record length.

MAX_QUEUE_ENTRIES 0 - n n=max allowed shared memory segment divided by 512 [default 500] Specifies the maximum number of entries allowed in the queue. NOTE: If you exceed the maximum number of queue entries, subsequent transfer attempts will fail.

MAX_SESSIONS_ENTRIES 0 - 999 [default 15] Specifies the maximum number of lines in the xcom.ses file.

MAXRECLEN 1 - 32767 [default 1024] When transferring a text file, this parameter designates the length in bytes of the largest record which can be transferred. If a record length is longer than this value, then the TRUNCATION parameter is read. When transferring binary files, this value indicates the length of the records that are transferred. This value must be equal to or lower than the LRECL on the remote system.

METACODE_CLASSES 1 - 64 characters [default none] For incoming report transfers only. Classes of print jobs that will be saved in the metacode format, a variable length record format.

NOTIFY_NAME 0 - 12 characters [default none] The user on the remote system who is to be notified when XCOM has completed the transfer. If the remote system is an IBM mainframe, XCOM uses the NOTIFYR parameter to determine the type of notification to deliver.

NOTIFYL WRITE, MAIL, TSO, WTO, CICS, LU, VM, or NONE [default NONE] Local user notification flag. The options are as follows: WRITE - A message will appear on the screen. MAIL - A mail message will be sent to the user. TSO - Notify specified TSO user. WTO - Write to log only (WTO). CICS - Notify CICS user. LU - Notify Logical Unit. VM - Notify VM/CMS user. NONE - No notification will be sent.

 

NOTIFYR WRITE, MAIL, TSO, WTO, CICS, LU, VM, or NONE [default NONE] Remote user notification type, when sending data to an IBM mainframe. WRITE - A write message will appear on the screen. MAIL - A mail message will be sent to the user. TSO - Notify specified TSO user. WTO - Write to log only. CICS - Notify CICS user. LU - Notify Logical Unit.

VM - Notify VM/CMS user. NONE - No notification to be sent.

NOTIFY_TERM 0 - 256 characters [default none] Specifies terminals to write if NOTIFYL=WRITE. If NOTIFY_TERM is not set, all users specified in the LOCAL_NOTIFY will be notified at the first terminal where they are logged in as found in the system table.

NUMBER_OF_RETRIES 0 - 99,999,999 [default 1] the maximum number of retries before a failed transfer is logged as failed and taken out of the transfer queue.

NUM_OF_DIR_BLOCKS 0 - 32767 [default 0] Specifies the number of directory blocks to allocate.

PASSWORD 0 - 31 characters [default none] The password associated with the USERID on the remote system.

PRIMARY_ALLOC 0 - 99999 [default 1] Identifies primary Cyls/Trks/Blks allocation.

QUEUE YES or NO [default NO] Indicates whether to execute the transmission requests immediately (specify NO) or to allow the request to be queued (specify YES). Note that if NO is specified, the request will abort if the remote system is unavailable. If YES is specified, the START_TIME and START_DATE parameters are read. See these parameters for more information.

QUEUE_PATH 0 - 256 characters [default /usr/spool/xcom/Q] The directory containing the queue data files.

RECORD_FORMAT F, FB, FBA, FS, FBS, FA, FM, V, VB, VBA, VS, VBS, VA, VM, VBM, or U [default VB] Specifices the record format of the dataset. This corresponds to the JCL RECFM subparameter. F - Fixed Unblocked indicates that all records have the same length. FB - Fixed Blocked indicates a fixed record length with multiple records per block. FBA - Fixed Block ANSI indicates a fixed record length with multiple records per block; these records contain ISO/ANSI/FIPS control characters. FS - Fixed Unblocked Spanned indicates that all records have the same length, and are to be written as standard blocks; these records do not contain any truncated blocks or unfilled tracks. FBS - Fixed Blocked Spanned indicates a fixed record length with multiple records per block, written as standard blocks. FA - Fixed Unblocked ANSI indicates that all records have the same length and contain ISO/ANSI/FIPS control characters. FM - Fixed Unblocked Machine indicates that all records have the same length and contain machine code control characters. V - Variable Unblocked indicates that records have variable lengths. VB - Variable Blocked indicates variable length records with multiple records per block. VBA - Variable Blocked ANSI indicates variable length records with multiple records per block; these records contain ISO/ANSI/FIPS control characters. VS - Variable Unblocked Spanned indicates that records have variable lengths, but a record can span more than one block. VBS - Variable Blocked Spanned indicates variable length records which may or may not have multiple records per block; these records can span more than one per block. VA - Variable Unblocked ANSI indicates that records have variable lengths and contain ISO/ANSI/FIPS control characters. VM - Variable Unblocked Machine indicates that records have variable lengths and contain machine code control characters. VBM - Variable Blocked Machine indicates variable length records with multiple records per block; these records contain machine code control characters. U - Undefined indicates the records are of undefined length.

REMOTE_FILE 1 - 256 characters [default none] For send file transfers only. The file on the remote computer to which the transferred data is being written. If you are creating the file (FILE_OPTION=CREATE), make sure your designated filename is consistent with the file naming conversions of the remote system. The local XCOM does not validate this name. The remote I/O system will determine whether or not the filename is valid.

REMOTE_FILE_RF 1 - 256 characters [default none] For retrieve file transfers only. Indicates the name of the file on the remote system to be retrieved.

REMOTE_SYSTEM 1 - 17 characters [default LU name] The name of the remote system with which you will communicating.

REMOTE_TRAIL_BLANKS YES or NO [default NO] Indicates whether or not to remove the blanks at the end of each record when receiving a text file.

REPORT_TITLE 0 - 21 characters [default none] For report transfers only. this field is used by different remote systems in the following ways: AS/400 S/38 - will assume this to be a printer filename. MVS - will interpret a non-blank value in this field as specifying the generation of a separator (banner) page for this value. UNIX - will pass this field to the LP spooler as a title field. Other systems - This field is generally used as a descriptive comment and is not printed as part of the report.

RESTART_SUPPORTED YES or NO [default YES]

RETRY_TIME 0 - 99,999,999 [default 1] The number of seconds between retries of unseccessful transfers. If it is zero, then no retries will be attempted.

SECONDARY_ALLOC 0 - 99999 [default 0] Identifies secondary Cyls/Trks/Blks allocation.

SESSIONS_FILE 0 - 256 characters [default /usr/spool/xcom/config/xcom.ses] The pathname of the xcom.ses file. This file tells XCOM how many sessions can be used by a single LU.

SHELL_CMD 1 - 256 characters [default /bin/sh] Shell commands to run scripts (jobs, reports, and notifications).

SPOOL_FLAG YES or NO [default NO] For report transfers only. Specifies whether to spool the report to disk or to print it directly. If the remote system is an IBM mainframe, this parameter will have no effect on the transfer.

START_DATE mm/dd/yy [default none] Specifies the date the transfer becomes eligible for execution. the format is mm/dd/yy. For example, 02.28.92 indicate a February 28, 1992, start date. If this field is blank, then the current date is used.

START_TIME hh:mm:ss [default none] Specifes the time the transfer becomes eligible for execution. the format is hh:mm:ss. For example, 14:00:00 indicates a 2 p.m. start time. If this field is blank, then the current time is used.

TMPDIR 1 - 256 characters [default /tmp] Directory in which to create temporary files invoked as a separate UNIX process.

TP_CMD 0 - 256 characters [default /usr/lib/xcom/xcomtp] Pathname of the transaction program.

TRACE_PATH 0 - 256 characters [default /usr/spool/xcom/trace] The pathname of the directory containing the trace data files.

TRUNCATION YES or NO [default NO] Indicates whether or not XCOM will truncate any excess characters on the source file which exceeds the maximum record length permitted on the target computer. If NO is selected, XCOM will abort the transfer is the maximum record length is exceeded. NOTE: Truncation is not valid for BINARY data and for data received on the UNIX platform.

UNIT 0 - 6 characters [default none] Used when (FILE_OPTION=CREATE). Specifies the unit on which the file is to be created. See the FILE_OPTION parameter for more information.

USERID 0 - 12 characters [default none] The user ID which the security system on the remote system checks before granting access for the file transfer.

VOLUME 0 - 6 characters [default none] Used when (FILE_OPTION=CREATE). Specifies the volume on which the file is to be created. See the FILE_OPTION parameter for more information.

XLOGFILE 0 - 256 characters [default /usr/spool/xcom/xcom.log] Name of the file where XCOM logs its activities. NOTE: User must specify full path name if QUEUE=YES.

XLPCMD 1 - 256 characters [default none] Configured for incoming reports only. Name of the command used to send print jobs to the spooler.

XLUNAME 1 - 17 characters [default LU name] The name of the local LU. This name must match the remote LU name on the remote system.

XMODE 1 - 8 characters [default XCOMMODE] Mode name associated with the SNA/APPC configuration for the XLUNAME-Remote LU name pair. This name must match the mode name defined on the remote system.

XTRACE 0 - 9 [default 0] Used only during testing and tracing procedures. Level of execution tracing desired. 0 indicates no tracing. Level 9 is the maximum level and shows the contents of data buffers.

 

Computer Manager Install

 

REQUIREMENTS:

1) UUCP running,

2) CRYPT installed on all participating systems,

3) The IP Name of the Manager System in all of the Managed

System’s /etc/hosts file,

4) The IP Name of all Managed Systems in the Manager System’s

/etc/hosts file.

 

INSTALLATION:

1) UUCP Setup

Run the tcpcfg script out of /home/z9061j on each

participating system. (Note: You must be in root to execute

this script.)

If you are installing a system, move /home/z9061j/tcpcfg from

USATLUAZ.

Responses:

- For the "TYPE MACHINE" message, enter 2 ....NCR

- Take the defaults for the next three messages concerning

paths.

- Enter the IP Name (Note: If you are on a Managed System,

enter the Manager IP Name. Then reply "N". If you are on

the Manager System, enter each of the Managed IP Name(s).

Then reply "Y", until all Managed System names are

entered.)

***You only need a new Managed System name if this is an addition

to current configuration.

You should receive a success message at the end of the script processing. To verify, enter the following:

sacadm -l (to show active for TCP)

nlsadmin -x

pmadm -l

 

Set-up Managed System IP Name(s) on the Manager System

Add IP Name to /etc/hosts on Manager System (USATLUAZ)

Add Manager System IP Name (USATLUAZ) to /etc/hosts on

each Managed System.

cd /usr/lib/uucp

./Uutry -r xxxxxxxx where xxxxxxxx is the IP Name of the system you are attempting to contact. You should do this on the Manager System and the Managed system to make sure contact works both ways.

AGENT SOFTWARE:

1) INSTALLED ON MANAGER MACHINE AS SPOOLED, sscma

2) Set-up to share for all managed boxes:

share -F nfs /usr/spool/pkg

3) TELNET to the Managed Systems.

SU to root,

Then mount the file as follows:

mount -F nfs usatluaz:/usr/spool/pkg /mnt

cd /mnt

pkgadd -d /mnt sscma

Responses:

- Reply "n" to the exception report message,

- Enter IP Name of the Manager Machine usatluaz,

- Reply "y" for permissions,

- Take the default for the nuucp default,

- Reply "n" for manager installed message,

- Reply "n" for the process accounting file message.

After completion of the install, issue the following unmount command:

umount /mnt

To get out of mount, issue the cd / command.

 

CONFIGURATION:

Logon to the Manager System as root and enter the following:

/home/compmgr/bin/compmgr

This should bring up MENU. Get the manuals "Using Computer Manager. There are two (2) volumes. Through the panels, add the new machine. Then set-up the default alarms, scheduling defaults, file status, etc.

- Go to the main menu,

- Select Node Administration,

- Select Node Inventory,

- Select Machine Inventory,

- Select Add,

- Fill in fields marked with an asterisk (*), then submit,

- Select Submit again to select Default Schdeules,

- Select Node Administration,

- Select Alarm Thresholds,

- Select General Thresholds,

- Select choices and select any machine to fill in data then

modify any entries if necessary, then Submit.

Go back to File System Thresholds and Processes and Alarm emptions and fill in as necessary.

- Go to the main menu,

- Select Utilities,

- Select On Demand Event Requests,

- Fill in data via choices & submit keys and submit,

- Go to main menu and select Critical Indicator panel.

Eventually messages concerning the new machine will appear

--------------------------------------------------------------

 

O/A ADMIN Installation

 

Get the Operations/(Ease)/Advantage Administration Guide and go to installation section.

Make sure each of the below is located on USATLUAZ file /usr/spool/pkg:

eaadm..... Administration

eafw...... Framework

easp...... Spool

eats...... Task Interface

eatsbase.. Task Base

Install eafw (install this package first).

Install eatsbase.

Install easp.

Install eats (prereq. for MENU and ADMIN).

Install eaadm.

Enter on the Manager System USATLUAZ:

share -F nfs /usr/spool/pkg

Telnet to the new system and su to root.

mount -F nfs usatluaz:/usr/spool/pkg /mnt

pkgadd -d /mnt eafw

pkgadd -d /mnt eatsbase

pkgadd -d /mnt eats

pkgadd -d /mnt easp

pkgadd -d /mnt eaadm

On the Manager System do following:

cd /etc/conf/cf.d

enter eafwconfig See O/A Admin manuals.

Change table to add new machine.

UAZ select all columns

UU2 select cs & nf only (select no columns)

UU3 select cs & nf only

UA0 select cs & nf only

UA1 select cs & nf only

UA2 select cs, nf & bs only

UA3 select cs & nf only

UA4 select cs & nf only

UA5 select cs & nf only

ASERV select cs & nf only

Note: Unless told otherwise, all new machines should be "cs & nf" only. After you exit config file a new config file will be distributed to all defined systems.

Enter eamenu & to bring up Admin menu.

Get the menu manuals and set-up an ICON for the new system.

Bring down Menu.

If you get config messages, follow the directions given.

enter eafwconfig and configure the system again.

Secure all new hosts to do commands against that host.

Enter commands:

easecure list to list hosts status

easecure + nodename to secure new host (replace

nodename with new system name)

easecure - nodename to unsecure a host

Bring up eamenu & again and test access to the new system.

--------------------------------------------------------------

 

Installation Verification

 

The final step in the staging process is to verify the install. The execution of the command: pkginfo -p gives a list of the packages that were installed incorrectly and currently show as in error. It does not show what the error is however.

This simple test should be performed before signing off on the staging of the machine.

--------------------------------------------------------------

 

Removing WIN-TCP

 

If WIN-TCP was installed before LLC, you will need to remove it. Go to state 1 and enter the following command: pkgrm WIN-TCP

This process will take approximately 15 minutes to complete.

You should remove and install WIN-TCP under level 1. You can get into level 1 by entering the command:

"init 1".

The system will partually re-boot. You will need to logon again.

Start the pkgrm WIN-TCP procedure.

Reply "y" to message asking if you want to remove WIN-TCP.

You will get a dependency check message for rpc and csf, reply

"y" to continue.

Reply "y" to rebuild kernel message also.

Follow shutdown instructions when they appear.

-----------------------------------------------------------------

Kernel Link Failure

 

Follow these steps to view the trace log if you get a kernel link failure.

From the boot diskettes, select System Maintenance.

Insert the maintenance diskette when prompted and type:

fsck -F ufs /dev/dsk/device number /mnt

This will do a file check on the /var directory on the root drive.

Type mount -F ufs /dev/dsk/device number /mnt

This will mount the /var directory

Then do a cat on /mnt/spool/install/log.install and page through the install messages. Just before the link edit failure messages, you should see an unresolved symbol message. From this point, you should be able to determine the package that failed.

 

Additional Commands

 

pkginfo | pg This will list all packages in page format.

pkginfo -p This will list all packages installed

incorrectly or not completely.

pkginfo -x | pg This will list all currently installed

packages with release level information in

page format.

pkginfo -xd /dev/dsk/f03ht | pg This will list packages with

release level information in page format on a

diskette.

pkginfo -xd /dev/rmt/device number | lp -d tcp1u

This will list packages with release level

information on a tape (where device number is

the tape drive device number which will vary

on each system, ex. c100t0d0s0) and route the

output to the printer.

who -r lists info. on run-level of init process.

who -s lists current users’ name, line, and time

logged in.

ALT, PRT-SC, and H At the same time, will get you out of

the editor.

CNTL D Is like an ESC key on 3270 terminals.

CNTL Z Halt command.

DEL Key Halt command.

ALT, PRT-SC, and P At almost the same time will get you out of

the vterm mode.

wall Broadcast to all users (see Install Patches

section on how to use this command).

To rebuild the kernel, issue the following command:

/etc/conf/bin/idbuild

and then issue the command: shutdown -y -i6 -g0

In sysadm, the Function keys are as follows:

F2 = CHOICES

F3 = SAVE (or execute)

F6 = CANCEL (like PF3)

F7 = and choose EXIT to leave main menu

To find a connection between modules and packages, issue the following commands:

cd /var/sadm/install

grep module_name contents

To find the CPUID, issue the following command:

/etc/ildconfig

The information under the column NODE is the CPUID.

Patch Install Instructions

 

These instructions assume that one or more patches will be installed on a weekday, and that they will take affect on a normal Sunday reboot. As much testing of the install is done during the week as possible.

Preparation.

You should have a patch tape or diskette provided by AT&T. You should have instructions on which patches to install, and any order of install that is required. These instructions will be provided by someone else at The Coca-Cola Company based on research of Read-me files, discussion with Dave Tennison, and experience on a test system.

Adding the patch.

Logon to the machine console. If the tape drive is not labeled, look in /dev/rmt. Tape drive names are in the format c#t#d#s0, ignore the entries ending in y’s and n’s. These are rewind and norewind options. Enter "ls | pg" to get the list. You will have to find all of the tape drives to find the correct name. If the drives are labeled, skip this research. Many patches need to be installed on the system in "stand-alone" mode. The research on the patch done before-hand should reveal this. If it does, you must set the system to init state 1. This cannot be done during the week unless scheduled with the users/clients. See the section "Additional Information" for commands to use if this is required.

enter

pkgadd -d /dev/rmt/c#t#d#s0 [package_name] (for cartridges)

pkgadd -d /dev/rdsk/f03ht [package_name] (for diskettes)

there will be a prompt to insert the tape. The package_name is optional. If you know the patch/package to install, go ahead and enter it after the pkgadd command as shown above. If not, the tape/diskette will be read to determine which patches are on it. Then a list will be presented. Select the patch that you wish to add by number. It is recommended to select them one at a time, and to install each one before continuing. If the tape is already inserted in the drive, type go, at the prompt to get a list of the patches.

If you are prompted to rebuild the kernel, reply No.

For any other prompts, you should reference your install instructions.

Test build of the kernel

If the patch you are installing DOES NOT require a kernel build, then skip this step.

Since the actual kernel build will be done on Sunday, it is prudent to do a test build of the kernel during the week while it is easy to call ATT support.

The kernel exists in /stand/unix. The kernel build process automatically puts a copy in /stand/unix.old.

Make a copy of the old backup.

cd /stand

cp /stand/unix.old /stand/unix.old2

Now build the kernel.

/etc/conf/bin/idbuild

 

During the kernel build process often there are 2 messages referring to illegal {} signs on lines 33 and 35 of /etc/conf/pack.d. Ignore these messages.

Any other messages should be investigated, unless they are in your instructions.

After the kernel is successfully built, restore the kernel.

cp /stand/unix.old /stand/unix

mv /stand/unix.old2 /stand/unix.old

The mv is faster than the cp because a move is just a rename.

Open a change record and indicate Operator action, so that it is discussed at the Thursday change meeting. Be sure that the Sunday reboot person knows to do a kernel rebuild.

The final step in the installation of a patch is to verify the install. The execution of the command: pkginfo -p gives a list of the patches/packages that are installed incorrectly and currently show as in error. It does not show what the error is however. Execute this command before signing off on the patch install.

Additional Information

To set the system to "stand-alone" mode, enter the following command: init 1. The network functions of the system will be shutdown as well as many other products. Check that all users are off the system because this command will log them off. To check if users are currently on the system, enter the who command: who -s lists current users’ name, line and time logged in. The command: who -r lists informations on the current run-level (use this to see if you are in init state 1).

If users are on the system and you need them to logoff, the wall command will send a broadcast message to all users. Type wall and press enter. This will put you into edit mode. Type the message you want to send. When complete, hit enter and then

"CNTL D" to escape edit mode.

 

To remove a package enter the command: pkgrm [package_name]

 

There is an upgrade option for Install Tapes when going from one point release to another. Files like mtune and stune should be manually backed-up before starting the upgrade. These files may be overlayed or updated with undesirable parameters. Do compares (diff command) after the install to see what has changed and make updates accordingly. Also, you can check the files timestamp to see what was updated.

One method to derive the impact of an upgrade is to put all the patches into a spool area and see which directories/files the upgrade updates.

Be sure to check the README files for the patches to see the dependencies and to check the areas the patches effect.

Generally the patch process is this, install patch, rebuild the kernel, and then re-boot the system.

PKERNxxx and Networking patches should be installed by themselves and the kernel should be rebuilt immediately after the last patch has been installed.

Tunable parameters do not take effect until the kernel has been rebuilt and the system rebooted.

PBASEAxxx - PBASEGxxx are patches to the BASE. The BASE is divided into functional areas which are numbered A-G. To upgrade the entire BASE, all A-G patches at a certain level would have to be installed.

There are seven layers of UNIX software. The top layer is Application software (ex. nfs, rpc, ftp, etc.), the second layer is Presentation, the third is Session (packages built into the kernel), the fourth is Transport (like netbios), the fifth is Network (ex. TCP/IP, LLC, etc.), the sixth is Data Link (like dlt), and the seventh and final layer is Token Ring.