GE XVB601

Company site: GE Intelligent Platforms

Module manufacturer information

Board page on company site: XVB601

Documentation: Datasheet Hardware Reference Transition Module Installation Guide Product Configurations Ruggedization

List of boards

• [01] 00:20:38:04:23:5E / 129.57.68.88 / croctest1

• [02] 00:20:38:04:23:60 / 129.57.68.89 / croctest2

• [03] 00:20:38:04:29:2E / 129.57.68.68 / tage

• [04] 00:20:38:04:23:64 / 129.57.68.86 / tage2

Initial settings

Install new board into VME crate. Connect ethernet to the upper LAN port, rs232 9600N1 serial line (optional), VGA monitor and USB keyboard. Turn on power, setup page should appear (for future reboots hold DEL key to enter setup). Goto Advanced->Onboard Device and enable Onboard Gigabit LAN 1. Goto Advanced->Serial Port Console Redirection, enable COM1 Console Redirection and in Console Redirection Settings set Bits per second=9600. Do Save & Exit.

Sergey Boyarinov only: open web browser from clonweb and goto jnet.jlab.org. Login using CUE login and password. Click on Machine Registration button. Click Add Device. Fill in a form, for example:

Connection Type: wired
MAC address: 00:20:38:04:23:64
VLAN: VLAN 568 : Hall B Data Acquisition <choose from the list>
Hostname: tage2.jlab.org
IP address: 129.57.68.86 <unselect 'Automatically Assign Statis IP>
JLAB Username: boiarino
Has PII: no
Machine Type: other
Property Tag: F

Click Submit. Following message must shows up:

The device (00:20:38:04:23:64) has been successfully added to the registration table.
-----------------
The device (00:20:38:04:23:64) has been successfully added to the assignment table.
-----------------
(DHCP) This entry will be added to DHCP the next time the server restarts (about 10 seconds).
-----------------
hallb-rtr.jlab.org ethernet8/9 is not an access port, blip failed...

Recycle power. Messages on monitor must show successful DHCP connection. Go to the following section to setup TFTP etc.

Software structure description

• following is installed in /tftpboot area on clon10 server:

linux-install/Centos5PAE_new/initrd.img
                            /initrd.img.backup
                            /vmlinuz
             /msgs/boot.msg
                  /expert.msg
                  /general.msg
                  /param.msg
                  /rescue.msg
                  /snake.msg
             /pxelinux.0
             /pxelinux.cfg/81394458 -> croctest1
                          /81394459 -> croctest2
                          /croctest1
                          /croctest2
                          /default
                          /pxeos.xml
         

Every client must has a file in /pxelinux.cfg/ area with the name containing IP address in HEX format, for example croctest1's IP is 129.57.68.88, so file name must be 81394458. It can be a symbolic link. The contents of the file for croctest1:

default Centos5PAE_new
label Centos5PAE_new
    kernel Centos5PAE_new/vmlinuz
    append  initrd=Centos5PAE_new/initrd.img root=/dev/ram0 init=disklessrc NFSROOT=129.57.167.16:/vol/diskless/Centos5 ramdisk_size=31809 ETHERNET=eth0 SNAPSHOT=croctest1.jlab.org NISDOMAIN=CCCHP acpi=force vga=0x305 vmalloc=256MB console=ttyS0,9600

All settings in that file are more or less generic. To customize it for another client only SNAPSHOT=croctest1.jlab.org must be changed. Statement NFSROOT=129.57.167.16:/vol/diskless/Centos5 defines the fileserver IP and directory where diskless Linux is located, in our case it is clonfs1. Statement console=ttyS0,9600 force Centos5 to send boot messages to the com port so they can be monitored using tsconnect.

• diskless Linux is installed on clonfs1 fileserver in /vol/diskless/Centos5 volume. That volume is mounted on the clonpc3 RHEL machine as well, and all Centos5 installation was performed on clonpc3, as it described in EXPERT PART section below. Directory /diskless/Centos5/root will be mounted as / level on all diskless clients, so that part is generic for all clients. Custom areas in /diskless/Centos5/snapshot will be client-specific, for every new client new directory with appropriate name must be creates there using templete tar ball, for example:

tar xvf templete.jlab.org.tar
mv templete.jlab.org croctest1.jlab.org

• File files contains the list of all files/directories to be used from snapshot area rather then from root area, that file was created during Centos5 installation. File files.custom can be used to specify additional files/directories, not in use right now.

Customizing kernel installed in /diskless/Centos5/root

After expert installation is done, following changes must be applied. They are similar to the RHEL installation on clon cluster PCs and servers:

• /usr/bin/system-config-network

• fix /etc/passwd, /etc/shadow, /etc/group

• fix /etc/nsswitch.conf

• fix /etc/yp.conf (add lines domain CCCHP server clon00.jlab.org and ypserver clon10.jlab.org)

• fix /etc/sysconfig/network (add line NISDOMAIN=CCCHP)

• /etc/init.d/ypbind start

• chkconfig ypbind on

• set RHEL-style automounts

• /etc/init.d/autofs restart

• fix symbolic links to /apps, /home, /usr/local etc