Firmware upgrades

From a local CDROM

1. Enable the firmware update jumper. On DECstation 3000/300s it is called the W1 jumper. The write-enabled position of the jumper is on pins 2 and 3. You can write-protect the flash ROM by moving the jumper to pins 1 and 2. On DECstation 3000/500, open the lock, remove the top cover and the left side panel. The jumper is located at the bottom of the memory banks. Putting the jumper on both pins will enable it. Leaving on the bottom pin will disable it. On the DECstation 3000/600 machines there is a switch installed on the turbochannel slot that can be conveniently flipped on and off. We have left the jumpers in the DECstation 3000/700, 3000/900, Alphastation 250s and 500s enabled. On the Alphastation 250, the jumper is J4 (1 to 2 is Write enabled, 2 to 3 is disabled)

2. Find out the current firmware revision level look at the power up CPU line. It is part of a long string. You need to upgrade the firmware with most Operating System upgrades on the DEC Workstations. Boot the firmware Update Utility using

   boot dkB400
   

   If the CDROM is internal to the machine the device name could be DKA400. The Firmware Update Utility will then give some instructions and pick a file for you. When prompted with Show, Update, Verify and Quit use show to see the effects of what will happen and then do the update. You may be asked are you ready to proceed. Type yes. Most of our DECstation 3000 series machines are currently at version 7.0 The Alphastation 250 4/266 is [alpha200]as200_v5_5.exe. For Alphastation 500 5/333 the filename is [alpha500]as500_v5_0.exe

3. Power Cycle the machine and look for the new firmware string.

From the network using bootp

1. Mount the Firmware Update Utility CD-ROMi using the following command.

   #  mount -rt cdfs -o noversion /dev/rzXc /mnt
   

   Or better yet, grab the latest firmware from the DEC web-site http://www.workstation.digital.com/firmware.html

2. Modify or create the client database in the /etc/bootptab file.

   hep02:ht=ethernet:ha=08002b236423:bf=/fullpath/filename:
   ip=128.227.156.2
   

   where filename is the name of the default bootfile image which is the same as for the local CDROM images for the DECstation 3000 series machines excepting for the extension being btp For example, if you are updating a DEC 3000 Model 300/300L/300X/300LX AXP system the file name would be m300_v7_0.btp. The filenames are identical in the case of Alphastations 250 and 500 and they are called as200_v5_9.exe. The filename for the DEC 3000 Model 700 is m400_v7_0.btp and for Model 900 is m500_v7_0.btp

3. Enable the bootpd and tftpd daemons if you have not already done so for X-terminals or printers:
   Modify the /etc/inetd.conf file. Uncommon the tftp and bootps process. It should look like the following:

   
   tftp   dgram   udp    wait root /usr/sbin/tftpd tftpd -d -r /tftpboot /tftpboot
   bootps dgram   udp     wait    root     /usr/sbin/bootpd  bootpd
   
   

   • Find the process id of /usr/sbin/inetd daemon.

     #  ps aux | grep inetd
     #  kill -HUP 
     

4. Enter the following commands from the boot prompt of the system to boot from the BOOTP server:

   For DECstations
   >>>  boot ez0
   
   For Alphastations
   >>> set ewb0_protocols bootp
   >>> boot ewb0
   

5. If you are having difficulty getting bootp to respond, you may be able to change the ethernet connection from different subnets in the computer room to get one machine on one subnet respond to the other subnet. If you are in a different building, then you need to get another machine configured as bootp server in the same building before you can get this done.
6. Turn the jumper switches to off position on the machines that have them. Completely power cycle the computer. Check on the console for the new firmware string.

Installing Digital UNIX on a Non-DEC disk

The following steps are done to install a new DEC Alpha with a non-DEC system disk. We assume that we are going to be completely installing the software from a CD and repartition the drives. Several of these steps are not necessary if the Alpha is already delivered with a DEC system disk and is ready to be configured.

1. Install the disk drive inside the machine and connect it up to the network, power and a console terminal. We usually attach a local CDROM and proceed with the installation.

2. At the >>> prompt, type boot dkB400 and choose the advanced installation option. Let the machine install a default partition table and transfer the root partition from the CDROM. We then set the bootdef_dev, boot_osflags, server, ethernet (TENBT), auto_action variables and boot from the harddrive. When you are setting up the 100 Base TX Fast Ethernet cards you need to set the prom variable set ewb0_mode fastfd as the Asante hub supports full duplex networking.

3. Stop the boot midway by using ctrl-C and confirm that you do want to interrupt the installation. You should get the # sign as the prompt.

   cd dev
   ./MAKEDEV rz0
   disklabel -rw /dev/rrz0c SEAGATE
   disklabel -r /dev/rrz0c
   If you want to change you can use
   disklabel -r -e /dev/rrz0c
   

   If you are changing the partion table, it is done using ed. You will most probably need to adjust the sizes of /, swap. /STRING will look for the STRING in ed. s/OLD/NEW will substitute the OLD with NEW strnig. You need to make sure the partitions that we plan to use do not have overlaps and have enough free space. Swap partition should never be placed on partition A. It overwrites the bootblock. Save the new partition table (w) and exit ed (q). It will ask you whether you want to write the label out and you reply Y. Type disklabel -r /dev/rrz0c to verify the new partition table.

   Starting from Digital Unix 4.0B, it looks like DEC is getting picky about overlapping partitions when they are defined as 4.2BSD. One way to get around this is to do the following

   disklabel -w rz0 RZ28D
   disklabel -r -e rz0
   

   and change the partitions to your liking. Note that this is pretty disastrous for the data on that disk. So make sure you backup everything on the disk including the partitions you were not going to touch before you do this.
4. When you are ready to proceed hit ctrl-D. Proceed with the installation and confirm the partition information for swap and usr partitions. You also select the ufs as the filesystem type for / and /usr. The system shutsdown. You are asked to set the boot prom variables like boot_osflags, bootdef_dev and then type boot. On DEC 3000/500/600/700/900 it looks like we also need to define the variable server to be on. Otherwise machine will halt at the boottime with an error message indicating that the pointing device is absent.

5. The system will continue the installation and will proceed to the subset selection process. The optional subsets we chose for Digital Unix 3.2 are 8 15 18-23 27-29 32-33 43 47-61 65-66. These numbers vary with the release. The names of the corresponding subsets are:

   
    - General Applications:
   
        8) UNIX(tm) SVID2 Compatibility
   
    - Network-Server/Communications:
       15) Additional Networking Services
   
    - Printing Environment:
       18) Adobe Font Metric Files
   
    - Reference Pages:
       19) Ref Pages: Admin/User
       20) Ref Pages: Programming
       21) Ref Pages: Realtime
       22) Ref Pages: Windows Admin/User
       23) Ref Pages: Windows Programming
   
    - Software Development:
       27) Compiler Extensions, Profilers and Libraries
       28) DEC C (cc -migrate)
       29) GNU Revision Control System
       32) Software Development Environment
       33) Source Code Control System
   
    - System Administration:
       43) System Accounting Utilities
   
    - Text Processing:
       47) Doc. Preparation Tools
       48) Doc. Preparation Tools Extensions
   
    - Windowing Environment:
       49) Adobe Fonts
       50) Basic X Environment
       51) DECwindows 100dpi Fonts
       52) DECwindows 75dpi Fonts
       53) X Customizations for OEM
       54) X Fonts
       55) X Servers
   
    - Windowing Hardware Support:
       56) LK201 Keyboard Support
       57) LK401 Keyboard Support
       58) LK411 Keyboard Support
       59) LK421 Keyboard Support
       60) LK444 Keyboard Support
       61) PCXAL Keyboard Support
       62) X Servers for PCI
       63) X Servers for QVision
       64) X Servers for TurboChannel
   
    - Windows Applications:
       65) Additional DECwindows Applications
       66) Additional X Applications
   
   

6. After this the system starts loading software. This takes some time (30 mts) and a question will appear on the screen Configure the system ?. We answer y at which point we will be required to enter the system name, root password, time zone (it is U.S, Eastern) and the kernel options. The kernel parameters chosen are are system V devices, ISO 9660 cdfs and Audit subsystems.

7. The system will finish building the kernel and reboot with the new kernel.

Installing Digital UNIX on a DEC Disk

The following steps are done to install a new DEC Alpha with a DEC system disk which has factory installed software and is ready to be configured.

1. Attach the ethernet cable to the machine. Attach the vt300 terminal to the port marked COM1.

2. We attach any external disk drives that will be installed on this machine. We need to make sure that the SCSI device chain has unique SCSI ids on all the drives and the drive chain is terminated. Usually the internal system disks already come configured with SCSI ID 0 or 3. The CDROM is usually set to 4. Any tapedevice is usally set to 5. The machine itself is set to ID 7. We can pick any of the remaining IDs for the external drives.

3. The machines usually come with the proper version of firmware that is needed to run the OS pre-installed. You may want to check to verify that this is indeed the case. This is done by looking at the release notes.

4. Check and set the bootdef_dev, boot_osflags, server, ethernet (TENBT), auto_action variables and type boot to jumpstart the installation from the harddrive.

5. The system starts configuring the pre-loaded software. This takes about 10 minutes. You need to enter the system name, root password, time zone (it is U.S, Eastern).

6. The system will finish building the kernel and reboot with the new kernel.

Installing Digital Unix using RIS.

The following steps are done to install Digital Unix using Remote Installation Services over the network.

1. Configure one machine to be a RIS server. This is done during the DU installation by selecting RIS subsets. We pick HEP01 for this purpose.

2. Move /var/adm/ris area to a disk that has lot of free space. We use /charm01/tmp3/ris. Make the link from the new area to /var/adm/ris using the command

   ln -s /charm01/tmp3/ris /var/adm/ris
   

3. Run ris and add software. First you create a new software area and then you add software to it, typically from the CDROMs and any subsets stored on the disk (C++ runtime, Fortran runtime, Sort runtime library). It seems the best strategy is to install all possible subsets into the disk and each client will then have a chance to pick and choose what to install.

4. Edit the /etc/hosts temporarily to get rid of the phys.ufl.edu names. Just name the hosts hep02 etc.

5. Run ris and add clients. You need the ethernet addresses of the clients you will be installing the software to. This can be found by typing show device on the console or uerf -R | more if the machine is already running. Comment out any firmware bootptab entries you already have in the /etc/bootptab file as ris will delete any entries it finds with that hostname.

6. Edit /etc/inetd.conf and put in the actual directory where ris subsets are being stored in the tftpd line.

7. Get to the client console and type boot ez0 and answer the installation questions.

8. When the machine has configured itself and rebooted with the root password and kernel options issue the command

   # setld -l hep01:
   

   to finish the other subset installation.

Adding additional disks

If the machine has additional temporary or data disks that need to be configured, it is best to have all the disks connected at the time the operating system is configured for the first time. The genvmunix kernel recognizes the harddrives and makes the appropriate devices in the /dev directory. You can also add the drives later on and do all the necessary steps.

To find out the disk drives and Tape Drives configured
uerf -R | more
and look for rz and tz strings.

1. Physically attach the other drive to the machine. Take care to see that its rz number is set in such a way that it does not clash with that of any other device's on the SCSI bus.

2. Go to the directory /sys/conf.

3. We need to make a back-up of the file which is machine-name (in capitals).

4. Open the file using vi and come to the section where the bus is listed. We add an additional entry, a sample of which is shown below: device disk rzX at scsi0 drive 8*X The rz number X must be decided beforehand.

5. Build the kernel by typing doconfig -c MACHINENAME

6. Copy kernel in the directory /vmunix and reboot.

7. Once the machine is up again check if the devices is recognized or not by searching the directory /dev. Files with that rz number should be found there.

8. Now make the device using the command ./MAKEDEV rzX number

You then add the information to /etc/fstab so that they get mounted at boot time. The commands to do this assuming you have a scsi disk set at address 1 without any partition table information are

To read the partition table:
disklabel -r rz1

To write a partition table if one doesn't exist (Use with CARE!):
disklabel -rw rz1 ELITE9

To edit the partition table and adjust the sizes
disklabel -re /dev/rrz1c

To make a filesystem on a partition (say h with 1% free space)
newfs -m 1 /dev/rrz1h

To mount the filesystem
mkdir /charm01
mkdir /charm01/tmp1
vi /etc/fstab
mount -a -t ufs

Networking the machine

• Login as root and run setup. The very first thing we do is to configure the networking. This usually involves configuring the ethernet interface (ln0 or tu0), providing an IP address (128.227.XX.XXX) and subnet mask (255.255.255.0). On most machines we choose not run the rwhod, gated or routed. In stead we provide a default route. Depending on the subnet (XXX) the machine is located in this is 128.227.XXX.1.

  In the case of Fast Ethernet (100-Base TX) we provide additional ifconfig flags speed 200 to enable the full duplex 100 Mbps connections on the tu1 interface. You then let the machine start the networking services and we should be able to telnet out of this machine to remote machines using the IP address. If you are going to be doing 100-Base TX networking, make sure that the cables you use are truely Category 5 and the connections are done with great care to ensure that all leads are to the tips of the connectors. Any small variations will result in Packet Loss, CRC Framing errors and Link LEDs not lighting up when the cables are inserted. Do not use the Pink cables purchased at SKIPPER Electronics!

• Ftp to 128.227.17.70 and get /etc/securettys, /etc/hosts. Make sure our new host name is added in the /etc/hosts file with full domain name and the short name. This lets you log back into the new machine over the network as root.

• At this time we are now ready to leave the machine room and the console and go the office where we can work in windowing environment with access to CUT and Paste. You may want shutdown the machine, put the machine where it is going to located and make the network, console and power connections.

Unusual Problems

Setup Filesystems

# newfs  /dev/rrz0e SEAGATE
# newfs  /dev/rrz0g SEAGATE
# newfs  /dev/rrz1a SEAGATE
# newfs  /dev/rrz1d SEAGATE
# mkdir /newton
# mkdir /newton/apps
# mkdir /newton/tmp
# mkdir /newton/users

and add to the /etc/fstab

/dev/rz0e       /usr/local     ufs rw 1 3
/dev/rz0g       /newton/apps   ufs rw 1 4
/dev/rz1a       /newton/users  ufs rw 1 4
/dev/rz1d       /newton/tmp    ufs rw 1 4

and then do a

# mount -a -t ufs

Setting up BIND

• A default bind domain name needs to be specified. The domain name is phys.ufl.edu.

• The hostname of the bind server should be entered next. The hostname of the bindserver is neptune-gw, quark..

• If the script gives you an option of not changing the hostname take that option.

• In the menu about the service order, we choose the option of consulting local files first and then DNS next.

• You need to edit /etc/rc.config and take out the phys.ufl.edufrom the HOSTNAME line. The files /etc/resolv.conf which now contains the one server we entered could be updated to include the campus nameservers by copying it from another client machine.

• You can quit the setup program and try to log into any distant machine (e.g. lns101.lns.cornell.edu). The name service should resolve the number and connect you to that machine.

Setting up the NIS

Server 1:hep01
Server 2:neptune-gw
Server 3:quark

The next option to disallow ypset and ypsetme requests. We choose not to allow these requests by selecting the DISALLOW ALL ypset requests option.

We choose the option of all services using NIS and then start the NIS daemons if the machine is already on the network and exit from the setup. We can type ypwhich and ypcat -k passwd to see which server we are bound to and a NIS map that was sent to us.

Setting Up NFS

• First make sure that the new machine is in some netgroup to which filesystems are exported. This is done on neptune by editiing the /etc/yp/netgroup file.

• Run the setup and select the NFS option in the menu.

• Normally we do not allow nonroot mounts and automounts. An exception is made to the machines that will serve user disks to be mounted by PCs (quark, hep01).

• The number of NFS Daemons and NFS I/O Daemons can be set to the default values of 8 and 7. This may need to be increased for the server machines whose main purpose is to server NFS clients (quark, hep01, proton, neutron).

• The NFS LOCKING option should be enabled and the PC NFS support is normally disabled.

• Provide the directories this machine is going to export and the name of the netgroup to which these directories are exported. These names are created and maintained on neptune in the /etc/yp/netgroup area. We do not enter the machine names for the disks we mount from remote machines. We just edit /etc/fstab manually and put in the remote directories so that they are mounted again at the power-up time.

• We then edit /etc/exports file and add the option -root=0 so that the roots on the clients can create and move files on these filesystems.

• Edit /etc/fstab and add the information about the filesystems we mount from remote machines. We create the new directories. This can be done using a script. See /local/adm/newmachine for HEE machines and /local/adm/newmachine.cmt for CMT machines. The scripts also contain some swap strategy change to Lazy and setting up of the print spool areas. You will need to setup the /etc/printcap as a copy from some other machine. If it is a brand new machine it may be necessary to add the new hostname to the /etc/hosts.equiv on the printserver machines (quark and neptune) You can then test the printers by sending dummy printouts. It is important to check non-root printing to verify that spool area protections are set correctly.

Setting NNTP

NNTP is used to synchronize times on different machines in the local area network. We have NNTP servers which have the right time and date by querying other time servers on the internet and all our machines act as clients and get their time from one or more of our local servers.

• Run setup and select the NTPconfiguration option.

• The system will ask you the name of a NTP server now, The name of the server is neptune for the 24 subnet and hep01 for the 156 subnet. We can choose either for the 17 subnet.

• Select the correct large time differences (-g) option and Select the limit syslog messages option (-l) to record only error messages.

• Exit the set up program and edit /etc/ntp.conf file. The script puts in the line peer nameofserver. We change the word peer to server so that we are a pure client. We can then stop and restart the NTP daemons using the /sbin/init.d/xntpd stop and /sbin/init.d/xntpd start commands. We can then verify the correct time and date by typing date or ntpq followed by peers.

Setting Up SendMail

• Run setup, select Mail and decline the quick setup of mail

• The unqualified hostname is the machinename alone (without the domain extension)

• The BIND domain name is phys.ufl.edu

• Top level domain name is alsophys.ufl.edu

• We add a general purpose relay which is phys.ufl.edu.

• Select the policy for forwarding all mail not local to this machine to relay (2).

• Machine is not a mail hub. neptune servers that function for the department.

• The tcp address format is user@phys.ufl.edu(2).

• The user/alias lists are not modified and the aliases are considered local. The UUCP style, UMC and DECnet style addressing are not supported. We do not add any nicknames. We complete the mail setup and exit setup.

• We need to add new clients to our mailserver (neptune) database so that it knows to accept mail addressed to those clients. This is done by editing the /etc/mail/sendmail.clientsfile and adding the new client name to it. We also need to stop and restart the sendmail process on neptune. This is done on neptune

  /etc/init.d/sendmail stop
  /etc/init.d/sendmail start
  

• Send mail to some other "distant" account of yours and check if you have recieved the mail.

  
  # Mail -v chandra@ufl.edu
  Subject: Testing
  Hi
  This is a test.
  .
  EOT
  chandra@ufl.edu... Connecting to neptune-gw.phys.ufl.edu (smtpr)...
  220-neptune.phys.ufl.edu Sendmail 8.6.10/8.6.10 ready at Thu, 11 Apr 1996 17:02:44 -0400
  220 ESMTP spoken here
  >>> HELO charm01.phys.ufl.edu
  250 neptune.phys.ufl.edu Hello charm01.phys.ufl.edu [128.227.156.31], pleased to meet you
  >>> MAIL From:
  250 ... Sender ok
  >>> RCPT To:
  250 ... Recipient ok
  >>> DATA
  354 Enter mail, end with "." on a line by itself
  >>> .
  250 RAA28704 Message accepted for delivery
  >>> QUIT
  221 neptune.phys.ufl.edu closing connection
  chandra@ufl.edu... Sent
  
  

Finishing Setup

mount /usr/local@hep01 /mnt
cd /mnt/adm
ln -s /usr/local /local
ln -s /XXXXX/apps /usr/local/apps

lmf list | more

cd /
umount /mnt

Setting up Software

To find out what software subsets are currently installed on the machine you can type setld -i | more. The subsets that were once installed but now removed will not have the installed words in the second column. You can delete an individual or several subsets using setld -d SUBSETNAMEs. Some subset deletions require you to rebuild the kernel. They almost always require you to edit the configuration file explained above to remove the options that are no longer valid. Charm01 currently has a stripped down subsets and the kernel configuration file as examples for other machines.

• DEC_C++_RTL cxx500
• DECLADEBUG
• DEC_Fortran_RTLdfa380
• DEC_Sort_RTL
• If the system is required to provide X-term support install CDE.

If there is no local cdrom, we need to mount it from a remote machine. We usually mount cdroms on a directory called /cdrom If that directory does not exist, create it. Now type the following:

mount /(directory name)@(machine) /cdrom
cd /cdrom
cd softwaredirectory
setld -l .

mount -r -t cdfs -o rrip /dev/rz4c /cdrom
mount -r -t cdfs -o noversion /dev/rz4c /cdrom

DEC server 90M: DSRVH-M Rev A01
wm218a 08-00-2b-a2-35-c8     AS32604081
wm218b 08-00-2b-a2-37-6f     AS32604082

wm218a
Port 1 		Console
Port 2		hep02
Port 3		hep03
Port 4		hep04
Port 5		hep05
Port 6		hep06
Port 7		hep07
Port 8		hep08

wm218b
Port    Access    Status        Services Offered

  1     Remote    Idle          WOMBAT
  2     Remote    Idle          NILE02
  3     Remote    Idle          NILE01
  4     Remote    Idle          HEP01
  5     Remote    Idle          HEP12
  6     Remote    Connected     HEP11
  7     Remote    Idle          HEP10
  8     Remote    Idle          HEP09

show port all

change server announcement disabled
change port 2 name hep02 username hep02

change service hep02 port 2 password " "
change port 2-8 access remote autobaud disabled autoprompt disabled
dedicated none dialup disabled dsrlogout disabled dtrwait disabled
signal control disabled modem control disabled inactivity logout disabled

LOCAL> C MACHINE

ccr -h XX-XX-XX-XX-XX-XX -c ln0

/dev/rz0c       /quark/users    ufs rw,userquota 1 2