Scyld Beowulf Installation Guide

Scyld Beowulf Scalable Computing

Version 0.13

Scyld Computing Corporation


Introduction

This guide contains installation instructions for the Scyld Beowulf Scalable Computing Distribution. There are two methods available: The Quick Start Installation (x86) from CD and Installation (x86) from RPMs. Advanced users, may prefer the second method.


Quick Start Installation:
To launch the "Quick Start" installation, boot the cluster's front-end machine from the Scyld CD-ROM, see section Scyld Beowulf "Quick Start" Installation (x86) from CD.


Installation from RPM packages:
To install Scyld Beowulf from RPM packages, see section Scyld Beowulf Installation (x86) from RPMs.


Scyld Beowulf Reference Information
See section Scyld Beowulf System Overview, for an overview of the main portions of the Scyld Beowulf Scalable Computing Distribution and hardware recommendations for building a Scyld Beowulf cluster. For access to detailed technical information regarding Scyld Beowulf, see section `Introduction' in Scyld Beowulf Reference Manual.

Scyld Beowulf System Overview

The Scyld Beowulf Scalable Computing Distribution streamlines the processes of configuring, administering, running and maintaining a Beowulf cluster computer. This software may be used with a group of off-the-shelf computers connected through a private network to build a Beowulf cluster computer. The front-end "master" computer in the cluster distributes computing tasks to the other machines, the slave nodes, in this parallel architecture supercomputer.

For any Beowulf system, hardware selection is based upon the price/performance. Scyld recommends the following hardware for use with this release of the Scyld Beowulf Scalable Computing Distribution:

processors
Intel Pentium III TM, Intel Celeron TM, AMD Atholon TM or Alpha AXP* TM
architecture
one, two or four processors per motherboard
physical memory
128 MBytes or more (min. 32 MBytes)
operating system
Scyld Beowulf Scalable Computing Distribution
network cards
Fast or Gigabit Ethernet PCI Adapters with existing Linux driver support (see http://www.scyld.com/network for list)
network hub
Fast or Gigabit Ethernet switch
drives
IDE (up to ATA/66) hard drive and floppy drive for each node


* Requires the Scyld Beowulf Scalable Computing Distribution for AXP.

A Beowulf cluster is connected by a private internal network. The front end machine must have two network interfaces, one connected to the cluster internal network and one connecting to the outside network. Each of the slave nodes requires a network interface connected to the cluster network.
Assemble the cluster as shown:


                       ----------
  External    -------> | Switch | <--------------------
  Network     |        ---------- <----               |
  <--------   |           ^           |               |
          |   |           |           |               |
       -----------  ----------- -----------     -----------
       | Master  |	 | Slave   | | Slave   |     | Slave   |
       | Node    |	 | Node 0  | | Node 1  |     | Node N-1|
       |(front   |	 |         | |         | ... |         |
       |   end)  |	 |         | |         |     |         |
       |         |	 |         | |         |     |         |
       -----------	 ----------- -----------     -----------

The full Scyld Beowulf Scalable Computing Distribution is installed only onto the front end machine. A graphical user interface is available and included to further streamline the cluster configuration processes.

The Beowulf configuration file is created on the front end machine which specifies the master's cluster interface and a range of IP number assignments for the nodes. The slave configuration file is also created on the front end machine detailing the file system configuration for the slave nodes. A slave node boot image and complete operational kernel image for the slave nodes are created and stored on the front end machine. This completes the installation on the front end machine.

For the slave nodes, N identical slave node boot floppy disks (one for each of the N nodes) must be created containing the slave node boot image. These floppy disks must be inserted into each of the nodes before the nodes are powered up, one at a time. As a node becomes active on the network, the assigned IP address appears in the master configuration file as an "unknown" computer on the network. You may then select which entries to include in the cluster. You may also choose to partition the slave node hard disks over the network. Finally, the complete operational kernel image is downloaded to the slave node over the cluster network.

Software to implement the Beowulf Distributed Process Space (BProc) is an integral part of the Scyld Beowulf Scalable Computing Distribution. This allows you to start processes on slave nodes in the cluster and track them in the process table on the front end machine. It also provides process migration mechanisms to help in the creation of remote processes. This removes the need for most binaries on the remote nodes.

Turn to the next page for a brief description of the major portions of the Scyld Beowulf Scalable Computing Distribution.

A brief description of the major portions of the Scyld Beowulf Scalable Computing Distribution is given below:

BeoSetup
A GUI interface for configuring the cluster
beofdisk
A utility for remote partitioning of the slave node hard disks
BeoBoot
A set of utilities for booting the slave nodes
beoserv
The BeoBoot boot server: it responds to slave nodes and serves the boot image
BProc
The Beowulf distributed process space: the integral part of Scyld Beowulf
bpmaster
The BProc master daemon: it runs on the front end machine
bpslave
The BProc slave daemon: it runs on the slave nodes
bpstat
A part of BProc: it displays various pieces of status information
bpctl
A part of BProc: it is used to apply commands to referenced slave nodes
bpsh
A part of BProc: it is an rsh replacement
bpcp
A part of BProc: it copies files between machines
BeoMPI
The Message Passing Interface, optimized for Scyld Beowulf
perf
Support for Intel Pentium Pro, Pentium II, Celeron and Pentium III TM hardware performance counters
BeoStatus
A GUI interface for monitoring the status of the Scyld Beowulf cluster

Refer to the Table of Contents and Index of this manual to obtain detailed information for each of these components.

Scyld Beowulf "Quick Start" Installation (x86) from CD

In a Scyld Beowulf system there is a single front-end machine (master node) and one or more slave nodes. The front-end machine is configured as a normal Linux installation and the slave nodes each contain a sparse amount of code. Each machine in the cluster is installed with a Network Interface Controller (NIC) and communicates across the cluster network. See section `Scyld Beowulf System Overview' in Scyld Beowulf Reference Manual.

System Requirements

Front-End Machine

The front-end machine requires at least two network interfaces. One is connected to the private interface of the cluster network and one is connected to the outside network.

Slave Nodes

Slave nodes each require one network interface. Additionally, one hard drive, one floppy drive and one blank floppy disk per slave node are recommended. (The slave nodes are normally booted from floppy disk, but the Scyld Beowulf CD-ROM may be also be used to boot slave nodes installed with a CD-ROM drive.)

Installation on the Front-End Machine

Installation is run from the Scyld Beowulf Scalable Computing Distribution CD-ROM. If necessary, modify the front-end machine's BIOS to boot from the CD-ROM (Access the BIOS Setup Utility by hitting the BIOS control key immediately after re-booting the computer. The control key varies by manufacturer; it may be F1, F2, F10 or delete. Select the Startup and Boot submenus, following on-screen instructions; move/add CD-ROM to the list of boot devices, noting the device order).

Scyld Beowulf on-screen instructions will guide you through the installation. You may choose between graphics mode or text mode. Graphics mode presents more details than does the text mode. For further helpful information, see the Scyld website, www.scyld.com.

  1. Boot from the Scyld Beowulf Distribution CD-ROM on the front-end machine, typing "graphics" or "text" at the prompt.
  2. Follow the on-screen instructions to execute the installation.

Installation on the Slave Nodes

Each slave node requires a `BeoBoot' initial image to boot and operate as a member of the cluster. This BeoBoot image may be created using the `BeoSetup' application. At that time, users will copy this image onto floppy disk(s), one for each slave node. For subsequent boots, the user may choose to store this image in a `BeoBoot partition' on the slave node hard disk(s). For the `floppy boot', most Beowulf installations use a separate floppy disk for each slave node, though a single floppy disk may be used if manually, and sequentially booting the slaves. It is also possible to run slave nodes with a CD-ROM drive and bootable CDs, as well (The Scyld Beowulf CD-ROM will boot the slave nodes). See the mkisofs manual page for information on creating bootable CD images. For the `hard disk boot', the BeoBoot image must reside in the BeoBoot partition on the hard drive of each slave node that will use the hard disk boot. After the initial installation, each slave node may use whichever boot method is appropriate -- not all nodes must use the same boot method.

Assignment of Slave Nodes to the Cluster

  1. As root user on the front-end machine, start the beosetup gui.
    # /usr/bin/beosetup
    
  2. If you plan to boot the slave nodes from floppy disk, create node floppy disks (recommended 1/slave node) in beosetup.
  3. Boot the slave nodes from the node floppies or use the CD-ROM to sequentially boot the nodes. Slave nodes will be listed in beosetup by Ethernet Station Addresses in the order they connect to the cluster. You may change this order, but it is easiest to power them up in order.
  4. Drag slave node Ethernet Addresses to the center column; click Apply. This assigns the nodes to the cluster, using numbers 0 through N-1.
  5. If your slave node hard disks are already partitioned, edit the file, `/etc/beowulf/fstab', on the front-end machine, to record the mapping of the partitions on your slave node disks to your filesystems (see Note, below). Note: If the slave node hard disks have not been previously partitioned, you may use beofdisk to generate default partition tables for the slave node hard disks. See optional subsection, next.
  6. Reboot all of the slave nodes, from the front-end machine.
    # bpctl -S all -s reboot
    
    Status of the slave nodes will be listed in the beosetup window. All slave nodes should show status of up when ready for use. (Nodes without a disk partition table will remain in the unavailable state.)

    Optional Slave Node Disk Partitioning

    When slave nodes are in the unavailable state, their hard disks may be remotely partitioned from the front-end machine. For those nodes with no partition table, they should appear as unavailable after the previous procedures have been executed. For more details and options for the following steps, see section Slave Node Hard Disk Partitioning.

  7. On the front-end machine, in the directory, `/usr/lib/beoboot/bin', capture (or create default) partition tables for the nodes.
    # beofdisk -q
    
  8. Write the appropriate partition table to every drive on every node.
    # beofdisk -w
    
    After writing the default partition tables, running beoboot-install -a device (where device = hda or sda) will write the BeoBoot floppy image intox the BeoBoot partitions on the hard disks. See section `beoboot-install' in Scyld Beowulf Reference Manual. This will enable the slave nodes to be booted from their BeoBoot partitions, thus eliminating the need for a floppy disk to subsequently boot the nodes.
  9. Optionally, write the BeoBoot floppy image to existing BeoBoot partitions on all nodes with disk named device.
    # beoboot-install -a device
    
  10. If needed, update the file, `/etc/beowulf/fstab', on the front-end machine to record the mapping of the partitions on your slave node disks to your filesystems.
  11. Reboot all of the slave nodes.
    # bpctl -S all -s reboot
    
This completes the installation.

Scyld Beowulf Installation (x86) from RPMs

In a Scyld Beowulf system there is a single front-end machine (master node) and one or more slave nodes. The front-end machine is configured as a normal Linux installation and the slave nodes each contain a sparse amount of code. Each machine in the cluster is installed with a Network Interface Controller (NIC) and communicates across the cluster network.

See section `Scyld Beowulf System Overview' in Scyld Beowulf Reference Manual.

System Requirements

Front-End Machine

The front-end machine requires at least two network interfaces. One is connected to the private interface of the cluster network and one is connected to the outside network.

Slave Nodes

Slave nodes each require one network interface. Additionally, one hard drive, one floppy drive and one blank floppy disk per slave node are recommended. (The slave nodes are normally booted from floppy disk, but the Scyld Beowulf CD-ROM may be also be used to boot slave nodes installed with a CD-ROM drive.) Each slave node requires a `BeoBoot' initial image to boot and operate as a member of the cluster. This BeoBoot image may be created using the `BeoSetup' application. At that time, users will copy this image onto floppy disk(s), one for each slave node. For subsequent boots, the user may choose to store this image in a `BeoBoot partition' on the slave node hard disk(s). For the `floppy boot', most Beowulf installations use a separate floppy disk for each slave node, though a single floppy disk may be used if manually, and sequentially booting the slaves. It is also possible to run slave nodes with a CD-ROM drive and bootable CDs, as well (The Scyld Beowulf CD will boot the slave nodes). See the mkisofs manual page for information on creating bootable CD images. For the `hard disk boot', the BeoBoot image must reside in the BeoBoot partition on the hard drive of each slave node that will use the hard disk boot. After the initial installation, each slave node may use whichever boot method is appropriate -- not all nodes must use the same boot method.

Getting Scyld Beowulf Scalable Computing Distribution

The Scyld Beowulf Scalable Computing Distribution for x86 is available on the web from: Scyld Computing Corporation, http://www.scyld.com and via FTP from: ftp://ftp.scyld.com/pub.

The most recent version of this documentation is also available on the web from: http://www.scyld.com/support_documentation.html

Redistribution

All the software above is redistributable under the following terms:

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.

Front-End Installation

  1. Install the Scyld Beowulf packages (requires root privilege). The packages are available on the Scyld Computing Corporation website and ftpsite. The packages may be installed as follows:
    # rpm -U package_name
    
  2. Use the gui-based tool from Red Hat 6.2, linuxconf, to update the LInux LOader configuration file, `lilo.conf' (requires root privilege).
    # linuxconf Select `boot mode/Lilo/Add to LILO a kernel you
    have compiled'
    # Select <package_name>
    # Click OK 
    
  3. Re-run lilo (requires root privilege).
    # /sbin/lilo
    
  4. Reboot the front-end machine. Reboot your machine to start running the Beowulf kernel. Note: You may safely ignore error messages regarding Beowulf servers at this time - they are expected, since the servers have yet to be configured.

Cluster Configuration File

Specify the cluster's internal network addresses in the Beowulf master configuration file, `/etc/beowulf/config'. Choose from recommended network addresses, those of reserved networks (i.e. 192.168.x.x or 10.x.x.x). Sample configuration file is included that designates nodes as 192.168.1.100 - 192.168.1.131. If you intend to share an internal network with another cluster at some time in the future, be careful to choose addresses which don't conflict.

Record the following in `/etc/beowulf/config' (see example file below).

  1. The front-end's internal interface.
  2. Arange of IP addresses for nodes in the cluster. The number of addresses in this range determines the maximum size of the cluster.

Example `/etc/beowulf/config' from an 8-node cluster. The internal interface is eth1. The IP range is 192.168.1.100 - 192.168.1.107.

# Interface (the internal cluster network interface)
interface eth1

# IP Range
# range of IP addresses to be given to nodes
# These MUST be ip numbers and NOT hostnames.
iprange 192.168.1.100 192.168.1.107 

Boot Images

Initially, each slave node will be booted using an initial (floppy) image stored on a floppy disk (or CD-ROM). Then, the slave node will download a full boot image over the network from the front-end machine. Thus, there are two boot images to be created. The first is the initial image, that is copied onto a floppy disk or bootable CD-ROM and optionally, into a BeoBoot hard disk partition. The second is the network boot image, which is served to the slave nodes by the beoserv daemon. `/var/beowulf' is the conventional place to create these images.

  1. Create the slave node initial (floppy) boot image. To boot the nodes initially, you will need to create a slave node boot floppy disk for every node. These boot images will all be identical. Generate this image by running @command{beoboot -1}. If you have network hardware that isn't supported by the initial boot image as shipped, you can add new network drivers or PCI ID's to the boot image. See section `Adding A New Network Driver' in Scyld Beowulf Reference Manual, for details. Note: For information on creating bootable CD images, see the mkisofs man page.
    > cd /var/beowulf
    > beoboot -1
    Building phase 1 file system image in /tmp/beoboot.14151...
    ram disk image size: 1428K
    compressing...done
    Floppy boot image in nodeboot.img
    
  2. Copy the initial boot (floppy) image to a floppy disk. You can copy the initial boot (floppy) image onto a formatted floppy with the following command:
    > cat nodeboot.img > /dev/fd0
    
    Write-protecting the newly-created slave node boot floppy disk is recommended, as it will be used every time the node boots.
  3. Create the network boot image. Create the boot image that the slave nodes will download over the cluster network and run. This is done by running @command{beoboot} with the @command{-2} option. If you want to add any kernel command line parameters for the slave nodes, you should add them here using @option{-c}. (For example, perhaps you need to add "noapic" to the command line for your nodes to function correctly. Some machines may require other options like "apm=off".)
    > cd /var/beowulf
    > beoboot -2 [-c noapic] -k /boot/vmlinuz-2.2.16-21.beosmp 
    phase 2: kernel: /boot/vmlinuz-2.2.16-21.beosmp (2.2.16-21.beosmp)
    phase 2: command line: apm=power-off mem=128M
    phase 2: creating image: /var/beowulf/boot.img
    Building phase 2 file system image in /tmp/beoboot.24781...
    Building phase 3 rd in /tmp/beoboot.24781.3...
    ram disk image size (uncompressed): 12936K
    
  4. Update the Beowulf configuration file. In the Beowulf configuration file, `/etc/beowulf/config', the bootfile tag should point to the network boot image, `boot.img'.
    bootfile=/var/beowulf/boot.img
    
  5. As super-user with root privilege, start the Beowulf services on the front-end machine.
    > /etc/rc.d/init.d/beowulf start 
    Loading modules: [ OK ] 
    Setting up libraries: [ OK ] 
    Starting beoserv: [ OK ] 
    Starting bpmaster: [ OK ]
    

Slave File System Configuration File

The slave file system configuration is stored on the front end machine and specifies the file system configuration for the slave nodes. The format of this information is the same as the standard UNIX `/etc/fstab' file format. The default fstab for the nodes resides in `/etc/beowulf/fstab'.

In addition to the standard fstab format, Scyld Beowulf allows for variable expansion. `$MASTER' will be expanded to the IP address of the front end machine. This is useful for NFS mounts. There is no normal name service on the nodes, so server addresses must be IP addresses. A ramdisk may be automatically specified using `$RAMDISK'. Specify the mount point for the ramdisk in the second column (/ is allowed). Set the size of the ramdisk using the mount option, fs_size=65536 (for example) in the fourth column of the fstab file.

The fifth and sixth fields (dump frequency and fsck pass number)in the fstab file are ignored. Also, the `user' and `noauto' optional fstab flags are not meaningful here and should not be used.

  1. Create the `/etc/beowulf/fstab' file. Example fstab file:
    # This file is the fstab for nodes.
    # Note: Allows for variable expansions; variables to be expanded:
    # $MASTER = IP address of the front end machine (good for doing NFS mounts).
    # $RAMDISK = /dev/ramN, where N is automagically assigned.
    # The "noauto" and "user" optional flags and cols 5 and 6 are meaningless.
    #
    #/dev/hda1        beoboot     ext2    defaults        0 0
    #$RAMDISK         /foo        ext2    fs_size=65536   0 0
    /dev/hda2         swap        swap    defaults        0 0
    /dev/hda3         /           ext2    defaults        1 1
    none              /proc       proc    defaults        0 0
    none              /dev/pts    devpts  gid=5,mode=620  0 0
    $MASTER:/home     /home       nfs     defaults        1 2
    
    If one particular node needs to have a different file system layout (or just different device names) than the rest of the nodes, you may create a file system table for it in `/etc/beowulf/fstab.N' where N is the node number. If that file exists then the default fstab will be ignored for node N.
  2. Don't forget to have an `/etc/exports' entry for each NFS file system you'd like to add to this list.

Slave Node Installation

Once the cluster and slave file system configuration files are setup on the front-end machine, the slave nodes may be automatically installed from the front-end machine.

  1. Insert the slave node boot floppies (CDs) into the slave node floppy (CD) drives.
  2. Power on the slave nodes one at a time. As new nodes appear on the network and start making RARP requests, their Ethernet Station Addresses will appear in the `/etc/beowulf/config' file, marked as unknown.
  3. Select the slave nodes. It is easiest to wait until all nodes have appeared, labelled as unknown, before changing any of them to node entries. To select nodes, appropriately change unknown to node in `/etc/beowulf/config' on the front-end machine. The N-selected slave nodes are referenced by number, 0 through N-1, assigned by the order listed in the Beowulf config file.
  4. Send a SIGHUP to the beoserv daemon.
    # killall -HUP beoserv
    
    The beoserv daemon will now respond to RARP requests from the selected slave nodes. Nodes will go from down to unavailable to up in the bpstat output (run bpstat on the front-end machine). See section `Program Usage' in Scyld Beowulf Reference Manual. If you have not yet partitioned the nodes, they will remain in the unavailable state. You may partition the disks or do other manual setup tasks while in this state. Nodes in the up state are available to users.

    Optional Slave Node Disk Partitioning

    When slave nodes are in the unavailable state, their disks may be remotely partitioned from the front-end machine. For those nodes with no partition table, they should appear as unavailable after the previous procedures have been executed. For more details regarding the following steps, see section Slave Node Hard Disk Partitioning.

    Instructions here apply default partitioning to all slave node hard disks with no existing partition tables. This default scheme allocates three partitions: a 2 MB BeoBoot partition, swap space equal to two times the physical memory and the remainder of the disk as root partition.

  5. In the directory, `/usr/lib/beoboot/bin', capture (or create default) partition tables for the nodes.
    # beofdisk -q
    
  6. Write the appropriate partition table to every drive on every node.
    # beofdisk -w
    

    Optionally, Copy Initial Boot Image to Slave Node Hard Drive

  7. Copy the initial boot image into the BeoBoot Partition of the Slave Node(s) hard disk. To install on all the nodes' hda hard drives:
    #  beoboot-install -a hda
    
    To install on node 1, hda hard drive:
    #beoboot-install 1 hda
    
  8. If needed, update the file, `/etc/beowulf/fstab', on the front-end machine to record the mapping of the partitions on your slave node disks to your filesystems.
  9. Reboot all of the slave nodes.
    # bpctl -S all -s reboot
    

Slave Node Hard Disk Partitioning

While the slave nodes are in the unavailable state, their hard disks may be remotely partitioned using the beofdisk script. Note that beofdisk can handle more than one hard disks per node. The beofdisk shell script is designed to automate the process for hard disks by type, position and geometry (cylinders, heads, sectors). beofdisk uses one partition table for all disks of a particular type, position and geometry and may be used to generate recommended, default partition tables.

beofdisk may also be used to read an existing partition table on a slave node hard disk, as long as that disk is properly positioned in the cluster. It must be the first hard disk of its type and geometry (cylinder, heads, sectors) in the position identified, as the script sequentially queries the slave nodes numbered, 0 through N-1.

The default partition table allocates three partitions: a beoboot partition equal to 2 MB, a swap partition equal to two times the node's physical memory, and a single root partition equal to the remainder of the disk. The partition table for each disk geometry is stored in the directory, `/etc/beowulf/fdisk' on the front end machine in the format using nomenclature which reflects the disk type, position and geometry (ex: hda2495:255:63, hdb3322:255:63, sda2495:255:63).

Given this background regarding the design of beofdisk, noting the required configuration of disks to preserve any existing partition tables, consider the following scenarios:

All new disks; apply recommended default partitioning.
  1. Create default partitioning scheme for all disk types.
    # beofdisk -d
    
Follow steps 2-4, below.
New or old disks; generate generalized, user-specified partitions.
On the front-end machine, remotely run fdisk on the first disk of a particular type, position and geometry to create a unique partition table using bpsh N fdisk, where N is slave node number, 0 through N-1. Repeat this procedure for each type/position/geometry triplet. Finally, follow steps 1-4, below to read and write the newly created partition tables to the appropriate disks.
New or old disks; generate a unique partition for a particular disk.
Follow one of the scenarios, above, to partition the disk with a group. Then, on the front-end machine remotely run fdisk on any node to re-create a unique partition table using bpsh N fdisk, where N is slave node number, 0 through N-1. Or, simply run fdisk on each node, using bpsh N fdisk, appropriately.

Program Usage:

@command{beofdisk -h}
@command{beofdisk -v}
beofdisk [-d]
beofdisk [-q]
beofdisk [-w]

beofdisk is designed to allow one partition table for all slave node hard disks of each type/position/geometry (ex: hda2495:255:63, hdb3322:255:63, sda2495:255:63).

Options:

@option{-h}
Display a help message and exit.
@option{-v}
Display version information and exit.
@option{-d}
Create the default partitioning scheme file on the front-end machine.
@option{-q}
Sequentially query the nodes, to identify all hard disks on a node, by type/position/geometry (triplet). Upon encountering a new triplet, beofdisk captures the existing partition table, if there is one, or generates a default partition table for use with that triplet.
@option{-w}
Write the appropriate partition table to every drive on every node.

  1. Capture (or create default) partition tables for the nodes.
    # beofdisk -q
    
  2. Write the appropriate partition table to every drive on every node.
    # beofdisk -w
    
    After writing the default partition tables, running beoboot-install -a device (where device = hda or sda) will write the BeoBoot floppy image into the BeoBoot partitions on the hard disks. See section `beoboot-install' in Scyld Beowulf Reference Manual. This will enable the slave nodes to be booted from their BeoBoot partitions, thus eliminating the need for a floppy disk to subsequently boot the nodes.
  3. Optionally, write the BeoBoot floppy image to all existing BeoBoot partitions on nodes with hard disk named, device.
    # beoboot-install -a device
    
  4. Reboot all of the slave nodes.
    # bpctl -S all -s reboot
    

Troubleshooting

If any nodes fail to make it to the up state, check for error messages in the node boot log files, `/var/log/beowulf/node.node', where node is the assigned node number, 0 through N-1.

GNU General Public License

Version 2, June 1991

Copyright (C) 1989, 1991 Free Software Foundation, Inc. 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA

Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed.

Preamble

The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. This General Public License applies to most of the Free Software Foundation's software and to any other program whose authors commit to using it. (Some other Free Software Foundation software is covered by the GNU Library General Public License instead.) You can apply it to your programs, too.

When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things.

To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it.

For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights.

We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software.

Also, for each author's protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modified by someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reflect on the original authors' reputations.

Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all.

The precise terms and conditions for copying, distribution and modification follow.

TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION

  1. This License applies to any program or other work which contains a notice placed by the copyright holder saying it may be distributed under the terms of this General Public License. The "Program", below, refers to any such program or work, and a "work based on the Program" means either the Program or any derivative work under copyright law: that is to say, a work containing the Program or a portion of it, either verbatim or with modifications and/or translated into another language. (Hereinafter, translation is included without limitation in the term "modification".) Each licensee is addressed as "you". Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program). Whether that is true depends on what the Program does.
  2. You may copy and distribute verbatim copies of the Program's source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and give any other recipients of the Program a copy of this License along with the Program. You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee.
  3. You may modify your copy or copies of the Program or any portion of it, thus forming a work based on the Program, and copy and distribute such modifications or work under the terms of Section 1 above, provided that you also meet all of these conditions:
    1. You must cause the modified files to carry prominent notices stating that you changed the files and the date of any change.
    2. You must cause any work that you distribute or publish, that in whole or in part contains or is derived from the Program or any part thereof, to be licensed as a whole at no charge to all third parties under the terms of this License.
    3. If the modified program normally reads commands interactively when run, you must cause it, when started running for such interactive use in the most ordinary way, to print or display an announcement including an appropriate copyright notice and a notice that there is no warranty (or else, saying that you provide a warranty) and that users may redistribute the program under these conditions, and telling the user how to view a copy of this License. (Exception: if the Program itself is interactive but does not normally print such an announcement, your work based on the Program is not required to print an announcement.)
    These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it. Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Program. In addition, mere aggregation of another work not based on the Program with the Program (or with a work based on the Program) on a volume of a storage or distribution medium does not bring the other work under the scope of this License.
  4. You may copy and distribute the Program (or a work based on it, under Section 2) in object code or executable form under the terms of Sections 1 and 2 above provided that you also do one of the following:
    1. Accompany it with the complete corresponding machine-readable source code, which must be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or,
    2. Accompany it with a written offer, valid for at least three years, to give any third party, for a charge no more than your cost of physically performing source distribution, a complete machine-readable copy of the corresponding source code, to be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or,
    3. Accompany it with the information you received as to the offer to distribute corresponding source code. (This alternative is allowed only for noncommercial distribution and only if you received the program in object code or executable form with such an offer, in accord with Subsection b above.)
    The source code for a work means the preferred form of the work for making modifications to it. For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable. However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable. If distribution of executable or object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place counts as distribution of the source code, even though third parties are not compelled to copy the source along with the object code.
  5. You may not copy, modify, sublicense, or distribute the Program except as expressly provided under this License. Any attempt otherwise to copy, modify, sublicense or distribute the Program is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance.
  6. You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Program or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Program (or any work based on the Program), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Program or works based on it.
  7. Each time you redistribute the Program (or any work based on the Program), the recipient automatically receives a license from the original licensor to copy, distribute or modify the Program subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein. You are not responsible for enforcing compliance by third parties to this License.
  8. If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Program at all. For example, if a patent license would not permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Program. If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply and the section as a whole is intended to apply in other circumstances. It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system, which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice. This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License.
  9. If the distribution and/or use of the Program is restricted in certain countries either by patents or by copyrighted interfaces, the original copyright holder who places the Program under this License may add an explicit geographical distribution limitation excluding those countries, so that distribution is permitted only in or among countries not thus excluded. In such case, this License incorporates the limitation as if written in the body of this License.
  10. The Free Software Foundation may publish revised and/or new versions of the General Public License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns. Each version is given a distinguishing version number. If the Program specifies a version number of this License which applies to it and "any later version", you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation.
  11. If you wish to incorporate parts of the Program into other free programs whose distribution conditions are different, write to the author to ask for permission. For software which is copyrighted by the Free Software Foundation, write to the Free Software Foundation; we sometimes make exceptions for this. Our decision will be guided by the two goals of preserving the free status of all derivatives of our free software and of promoting the sharing and reuse of software generally.

NO WARRANTY

  1. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
  2. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.

END OF TERMS AND CONDITIONS

How to Apply These Terms to Your New Programs

If you develop a new program, and you want it to be of the greatest possible use to the public, the best way to achieve this is to make it free software which everyone can redistribute and change under these terms.

To do so, attach the following notices to the program. It is safest to attach them to the start of each source file to most effectively convey the exclusion of warranty; and each file should have at least the "copyright" line and a pointer to where the full notice is found.

<one line to give the program's name and a brief idea of what it does.> Copyright (C) 19yy <name of author>

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA

Also add information on how to contact you by electronic and paper mail.

If the program is interactive, make it output a short notice like this when it starts in an interactive mode:

Gnomovision version 69, Copyright (C) 19yy name of author Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'. This is free software, and you are welcome to redistribute it under certain conditions; type `show c' for details.

The hypothetical commands `show w' and `show c' should show the appropriate parts of the General Public License. Of course, the commands you use may be called something other than `show w' and `show c'; they could even be mouse-clicks or menu items--whatever suits your program.

You should also get your employer (if you work as a programmer) or your school, if any, to sign a "copyright disclaimer" for the program, if necessary. Here is a sample; alter the names:

Yoyodyne, Inc., hereby disclaims all copyright interest in the program `Gnomovision' (which makes passes at compilers) written by James Hacker.

<signature of Ty Coon>, 1 April 1989 Ty Coon, President of Vice

This General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Library General Public License instead of this License.

Index

Jump to: b - c - d - f - i - m - n - p - q - s

b

  • beofdisk
  • boot.img
  • c

  • cluster configuration
  • config file
  • d

  • disk partitioning
  • downloading
  • f

  • fdisk
  • front-end installation
  • front-end machine
  • fstab file
  • i

  • Index
  • m

  • master node
  • n

  • nodeboot.img
  • p

  • partitioning
  • q

  • Quick Start
  • s

  • slave file system config file
  • slave node installation
  • slave nodes
  • system requirements

  • This document was generated on 9 October 2000 using texi2html 1.56k.