Design and Deployment Guide for Windows HPC Server 2008 Microsoft Corporation Published: September 2008 Abstract This guide provides detailed information and step-by-step procedures for designing and installing a high performance computing cluster using Windows® HPC Server 2008. You can use this guide to first plan the deployment of your HPC cluster, and when you are ready to deploy you can follow the procedures to configure the head node, add compute nodes to the cluster, and verify that your cluster deployment has been successful.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Design and Deployment Guide for Windows HPC Server 2008
Microsoft Corporation
Published: September 2008
AbstractThis guide provides detailed information and step-by-step procedures for designing and installing
a high performance computing cluster using Windows® HPC Server 2008. You can use this guide
to first plan the deployment of your HPC cluster, and when you are ready to deploy you can follow
the procedures to configure the head node, add compute nodes to the cluster, and verify that your
cluster deployment has been successful.
Copyright Information
Information in this document, including URL and other Internet Web site references, is subject to
change without notice. Unless otherwise noted, the companies, organizations, products, domain
names, e-mail addresses, logos, people, places, and events depicted in examples herein are
fictitious. No association with any real company, organization, product, domain name, e-mail
address, logo, person, place, or event is intended or should be inferred. Complying with all
applicable copyright laws is the responsibility of the user. Without limiting the rights under
copyright, no part of this document may be reproduced, stored in or introduced into a retrieval
system, or transmitted in any form or by any means (electronic, mechanical, photocopying,
recording, or otherwise), or for any purpose, without the express written permission of Microsoft
Corporation.
Microsoft may have patents, patent applications, trademarks, copyrights, or other intellectual
property rights covering subject matter in this document. Except as expressly provided in any
written license agreement from Microsoft, the furnishing of this document does not give you any
license to these patents, trademarks, copyrights, or other intellectual property.
Windows Vista® Enterprise, Windows Vista Business, Windows Vista Home, or Windows
Vista Ultimate
Windows Server 2003 Standard Edition or Windows Server 2003 Enterprise Edition with
Service Pack 2 or later (x86- or x64-based)
Windows Server 2003, Compute Cluster Edition
Windows Server 2003 R2 Standard Edition or Windows Server 2003 R2 Enterprise Edition
(x86- or x64-based)
1.2. Decide how to add compute nodes to your clusterThere are three ways to add compute nodes to your cluster:
From bare metal. The operating system and all the necessary HPC cluster components are
automatically installed on each compute node as it is added to the cluster. No manual
installation of the operating system or other software is required.
Add preconfigured compute nodes. The compute nodes are already running Windows
Server 2008 HPC Edition, or another 64-bit edition of Windows Server 2008, and Microsoft
HPC Pack 2008 is manually installed on each node.
Import a node XML file. An XML file that contains a list of all the nodes that will be deployed
is used. This XML file can be used to add nodes from bare metal or from preconfigured
nodes. For more information about node XML files, see Appendix 2: Creating a Node XML
File.
The following is a list of details to take into consideration when choosing how to add nodes to
your HPC cluster:
When deploying nodes from bare metal, Windows HPC Server 2008 automatically generates
computer names for your compute nodes. During the configuration process, you will be
required to specify the naming convention to use when automatically generating computer
names for the new nodes.
Compute nodes are assigned their computer name in the order that they are deployed.
If you want to add compute nodes from bare metal and assign computer names in a different
way, you can use a node XML file. For more information about node XML files, see Appendix
2: Creating a Node XML File.
If you want to add preconfigured nodes to your cluster, you will need to install Windows
Server 2008 HPC Edition, or another 64-bit edition of Windows Server 2008 on each node (if
not already installed), as well as Microsoft HPC Pack 2008.
9
1.3. Choose the Active Directory domain for your clusterThe head node and the compute nodes in your HPC cluster must be members of an Active
Directory domain. Before deploying your cluster, you must choose the Active Directory domain
that you will use for your HPC cluster.
If you do not have an Active Directory domain to which you can join your cluster, or if you prefer
not to join an existing domain, you can install the Active Directory Domain Services role on the
head node and then configure a domain controller on that node. For more information about
installing the Active Directory Domain Services role on a computer that is running Windows
Server 2008, see the AD DS Installation and Removal Step-by-Step Guide
(http://go.microsoft.com/fwlink/?LinkID=119580).
If you choose to install and configure an Active Directory domain controller on the head
node, consult with your network administrator about the correct way to isolate the new
Active Directory domain from the enterprise network, or how to join the new domain to an
existing Active Directory forest.
1.4. Choose a user account for installation and diagnosticsDuring the configuration process of your HPC cluster, you must provide credentials for a domain
user account that will be used for installation and diagnostics. You must choose an existing
account or create a new account, before starting your cluster deployment.
The following is a list of details to take into consideration when choosing the user account:
The user account that you choose must be a domain account with enough privileges to create
Active Directory computer accounts for the compute nodes. Alternatively, you can create the
computer accounts manually or ask your domain administrator to create them for you.
If part of your deployment requires access to resources on the enterprise network, the user
account must have the necessary permissions to access those resources—for example,
installation files that are available on a network server.
If you want to restart nodes remotely from the cluster administration console (HPC Cluster
Manager), the account must be a member of the local Administrators group on the head
node. This requirement is only necessary if you do not have scriptable power control tools
that you can use to remotely restart the compute nodes.
1.5. Choose a network topology for your clusterWindows HPC Server 2008 supports five cluster topologies. These topologies are distinguished
by how the compute nodes in the cluster are connected to each other and to the enterprise
network. The five supported cluster topologies are:
Topology 1: Compute Nodes Isolated on a Private Network
Topology 2: All Nodes on Enterprise and Private Networks
Topology 3: Compute Nodes Isolated on Private and Application Networks
Topology 4: All Nodes on Enterprise, Private, and Application Networks
Topology 5: All Nodes on an Enterprise Network
For more information about each network topology, see Appendix 1: HPC Cluster Networking.
When you are choosing a network topology, you must take into consideration your existing
network infrastructure:
Decide which network in the topology that you have chosen will serve as the enterprise
network, the private network, and the application network.
Do not have the network adapter that is connected to the enterprise network on the head
node in an automatic configuration (that is, the IP address for that adapter does not start with:
169.254). That adapter must have a valid IP address, dynamically or manually assigned
(static).
If you choose a topology that includes a private network, and you are planning to add nodes
to your cluster from bare metal:
Ensure that there are no Pre-Boot Execution Environment (PXE) servers on the private
network.
If you want to use an existing DHCP server for your private network, ensure that it is
configured to recognize the head node as the PXE server in the network.
If you want to enable DHCP server on your head node for the private or application networks
and there are other DHCP servers connected to those networks, you must disable those
DHCP servers.
If you have an existing Domain Name System (DNS) server connected to the same network
as the compute nodes, no action is necessary, but the compute nodes will be automatically
deregistered from that DNS server.
Contact your system administrator to determine if Internet Protocol security (IPsec) is
enforced on your domain through Group Policy. If IPsec is enforced on your domain through
Group Policy, you may experience issues during deployment. A workaround is to make your
head node an IPsec boundary server so that compute nodes can communicate with the head
node during PXE boot.
1.6. Prepare for multicast (optional)If you will be deploying nodes from bare metal and want to multicast the operating system image
that you will be using during deployment, we recommend that you prepare for multicast by:
Enabling Internet Group Management Protocol (IGMP) snooping on your network switches, if
this feature is available. This will help to reduce multicast traffic.
Disabling Spanning Tree Protocol (STP) on your network switches, if this feature is enabled.
For more information about these settings, contact your network administrator or your
networking hardware vendor.
Note
11
1.7. Prepare for the integration of scriptable power control tools (optional)The cluster administration console (HPC Cluster Manager) includes actions to start, shut down,
and reboot compute nodes remotely. These actions are linked to a script file (CcpPower.cmd) that
performs these power control operations using operating system commands. You can replace the
default operating system commands in that script file with your own power control scripts, such as
Intelligent Platform Management Interface (IPMI) scripts that are provided by your vendor of
cluster solutions.
In preparation for this integration, you must obtain all the necessary scripts, dynamically linked
library (DLL) files, and all other components of your power control tools. After you have obtained
all the necessary components, test them independently and ensure that they work as intended on
the computers that you will be deploying as compute nodes in your cluster.
For information about modifying CcpPower.cmd to integrate your own scriptable power control
tools, see Appendix 5: Scriptable Power Control Tools.
Step 2: Deploy the Head Node
The next step in the deployment of your HPC cluster is to deploy the head node. The following
checklist describes the steps involved in deploying the head node.
Checklist: Deploy the head node
Task Description
2.1. Install Windows Server 2008 on the head
node computer
Install Windows Server 2008 HPC Edition, or
another 64-bit edition of Windows Server 2008
on the computer that will act as the head node.
2.2. Join the head node computer to a domain Join the computer that will act as the head node
to a Microsoft Active Directory Domain.
2.3. Install Microsoft HPC Pack 2008 on the
head node computer
Install Microsoft HPC Pack 2008 on the
computer that will act as the head node, using
the installation media or from a network
location.
12
2.1. Install Windows Server 2008 on the head node computerTo deploy the head node of your HPC cluster, you must start by installing Windows Server 2008
HPC Edition, or another 64-bit edition of Windows Server 2008 on the computer that will act as
the head node. For more information about installing Windows Server 2008, including system
requirements, see Installing Windows Server 2008 (http://go.microsoft.com/fwlink/?
LinkID=119578).
We strongly recommend that you perform a clean installation of Windows Server 2008
before installing Microsoft HPC Pack 2008. If you want to install Microsoft HPC
Pack 2008 on an existing installation of Windows Server 2008, remove all server roles
first and then follow the procedures in this guide.
It is recommended that you obtain the latest device drivers for your head node computer
from the Web site of your hardware vendors.
2.2. Join the head node computer to a domainAs described in the Step 1: Prepare for Your Deployment section, the head node must be a
member of an Active Directory domain. After you have installed Windows Server 2008 on the
head node, manually join the head node to an existing Active Directory domain.
2.3. Install Microsoft HPC Pack 2008 on the head node computerAfter Windows Server 2008 is installed on the head node computer, and the head node is joined
to an Active Directory domain, you can install Microsoft® HPC Pack 2008 on the head node.
1. To start the Microsoft HPC Pack 2008 installation wizard on the computer that will act as
the head node, run setup.exe from the HPC Pack 2008 installation media or from a
network location.
2. On the Getting Started page, click Next.
3. On the Microsoft Software License Terms page, read or print the software license
terms in the license agreement, and accept or reject the terms of that agreement. If you
accept the terms, click Next.
4. On the Select Installation Type page, click Create a new HPC cluster by creating a
head node, and then click Next.
5. Continue to follow the steps in the installation wizard.
Important Note To install Microsoft HPC Pack 2008 on the head node computer
After you have deployed the head node of your HPC cluster, you must configure the head node
by following the configuration to-do list in HPC Cluster Manager.
Checklist: Configure the head nodeThe following checklist includes the items in the configuration to-do list in HPC Cluster Manager
that you need to complete in order to configure your head node.
Task Description
3.1. Configure the HPC cluster network Configure the cluster network by using the
Network Configuration Wizard.
3.2. Provide installation credentials Specify which credentials to use for system
configuration and when adding new nodes to
the cluster.
3.3. Configure the naming of new nodes Specify the naming convention to use when
generating names automatically for new
compute nodes.
3.4. Create a node template Create a template that defines the steps to
follow when configuring a compute node.
3.5. Add drivers for the operating system
images (optional)
If you will be deploying compute nodes from
bare metal and those nodes require special
device drivers, add drivers for the operating
system images that you created for your node
template on the previous task.
3.6. Add or remove users (optional) If you will be giving access to the cluster to
other members of your organization, add or
remove users or administrators for your cluster.
3.1. Configure the HPC cluster networkThe HPC cluster network configuration is the first step in the configuration process of your head
node. The HPC cluster network is configured by following the Network Configuration Wizard in
HPC Cluster Manager. When configuring the HPC cluster network, you must choose the network
topology that you have selected for your cluster, as described in “1.5. Choose a network topology
for your cluster”, in Step 1: Prepare for Your Deployment.
Before you start configuring the HPC cluster network in HPC Cluster Manager, ensure
that the head node and the computers that you will add as compute nodes to the cluster
Important
14
are physically connected according to the network topology that you have chosen for
your cluster. Also, ensure that you are able to identify to which network each one of the
network adapters in the head node is connected.
1. If HPC Cluster Manager is not already open on the head node, open it. Click Start, point
to All Programs, click Microsoft HPC Pack, and then click HPC Cluster Manager.
2. In the To-do List, click Configure your network. The Network Configuration Wizard
appears.
3. On the Network Topology Selection page, click the topology that you have chosen for
your cluster, and then click Next.
4. On the Enterprise Network Adapter Selection page, in the Network adapter list, click
the name of the network adapter that is physically connected to your enterprise network,
and then click Next.
Important
To ensure that you are selecting the correct network adapter, use the
information displayed on this wizard page after you select a network adapter
from the list. Use the IP address, domain information, and Media Access
Control (MAC) address of each adapter as a reference.
5. If you chose topology number 5 for your cluster, jump to step 9 in this procedure.
Otherwise, repeat step 4 for the private network adapter.
6. On the Private Network Configuration page, type a static IP address and a subnet
mask for the head node. Optionally, select network services for that network:
a. To give access to resources on the enterprise network to compute nodes that are
connected to this network, select the Enable network address translation (NAT) on
the head node check box.
b. To enable DHCP services for the nodes connected to this network, select the Enable
DHCP and define a scope check box, and then type the starting and ending IP
addresses for the DHCP scope. If the Gateway and DNS server IP addresses have
not been automatically detected, type each of these addresses.
Note
For more information about enabling NAT and DHCP on your cluster
network, see “HPC network services” in Appendix 1: HPC Cluster
Networking.
7. Click Next after you are done configuring the private network.
8. Repeat steps 4, 6, and 7 for the application network adapter. Click Next after you are
done configuring the application network.
9. On the Firewall Setup page, select the firewall setting for the cluster:
a. To apply firewall settings automatically to head nodes and compute nodes on each
network, click ON for that network.
To configure the HPC cluster network
15
b. To disable the firewall on a network, click OFF.
c. If you do not want to change any firewall settings, click Do not manage firewall
settings.
Note
For more information about firewall settings for your cluster, see “Windows
Firewall configuration” in Appendix 1: HPC Cluster Networking.
10. On the Review page, verify your settings and click Configure. If you want to change any
of the settings, navigate to the appropriate wizard page by clicking it on the navigation
pane or by clicking Previous.
11. After the network configuration process is completed, on the Configuration Summary
page, review the list of configuration items. If you want to save a report of the network
configuration, click Save the configuration report.
12. To close the wizard, click Finish.
3.2. Provide installation credentialsInstallation credentials must be provided in order to configure new compute nodes. These
credentials will be used when installing the operating system, applications, and when adding
nodes to the Active Directory domain. Also, these same credentials will be used when running
diagnostic tests on the cluster nodes. For more information, see “1.4. Choose a user account for
installation and diagnostics” in Step 1: Prepare for Your Deployment.
1. In the To-do List, click Provide installation credentials. The Installation Credentials
dialog box appears.
2. Type the user name, including the domain (DOMAIN\User), and then the password for
the domain user account you will use to deploy compute nodes and to run diagnostic
tests.
Important
The account must be a domain account with enough privileges to create
Active Directory computer accounts for the compute nodes. Alternatively, you
can create the computer accounts manually or ask your domain administrator
to create them for you.
Important
If part of your deployment requires access to resources on the enterprise
network, the account should have the necessary permissions to access
those resources.
Important
If you want to restart nodes remotely from the cluster administration console
(HPC Cluster Manager), the account must be added as an HPC cluster
To provide installation credentials
16
administrator on the head node. This requirement is only necessary if you do
not have scripted power control tools that you can use to remotely restart the
compute nodes.
3. To save the specified credentials, click OK.
3.3. Configure the naming of new nodesIf you deploy compute nodes from bare metal, and you are not using a node XML file to import
nodes to the cluster, Windows HPC Server 2008 will automatically generate computer names for
the new nodes that are being deployed. You need to specify how those names will be generated,
by defining a naming series.
The naming series is defined by selecting a root name and the starting number that will
accompany that name. The starting number is enclosed in percentage signs (%). For example:
ClusterNode%1000%.
When you deploy compute nodes from bare metal, nodes will be named in sequence, as they
become available. For example, if you deploy three nodes after specifying the following naming
series: ClusterNode-%100%, those nodes will be assigned these names:
ClusterNode-100
ClusterNode-101
ClusterNode-102
Compute node names are limited to 15 characters. When specifying the compute node
naming series, take into account the number of compute nodes in your deployment and
ensure that the series that you specify will not generate names that exceed 15
characters. For example, if your deployment will consist of 1,000 compute nodes, and
your starting number is 1, your root name cannot have more than 12 characters;
otherwise, your node number 1,000 will need a name that consists of 16 characters.
1. In the To-do List, click Configure the naming of new nodes. The Specify Compute
Node Naming Series dialog box appears.
2. Type the naming series that you want to use. The preview helps you to see an example
of how the naming series will be applied to the names of the compute nodes.
Note
You cannot specify a compute node naming series that consists only of
numbers.
3. To save the compute node naming series that you have specified, click OK.
3.4. Create a node templateNode templates are new in Windows HPC Server 2008. They define the necessary tasks for
configuring and adding compute nodes to your cluster. With a node template, you can deploy an
Important To specify the compute node naming series
17
operating system image, add specific drivers and software to compute nodes, or simply add a
preconfigured node to your cluster. Because you might have more than one type of compute
node, or you may be adding compute nodes to your cluster in different ways, you can create
different templates that apply to different nodes or situations.
You can create two types of node templates:
With an operating system image. This type of template includes a step to deploy an
operating system on the compute nodes. Use this type of template when adding compute
nodes from bare metal.
Without an operating system image. This type of template is used to add preconfigured
compute nodes to the cluster, or to update existing nodes.
The type of template that you create for the initial deployment of your HPC cluster depends on
how you decided to add compute nodes to your cluster. For more information, see “1.2. Decide
how to add compute nodes to your cluster” in Step 1: Prepare for Your Deployment.
If you will create a node template with an operating system image, you will need the
installation media for Windows Server 2008 HPC Edition or another 64-bit edition of
Windows Server 2008, or you must have the installation files available on a network
location that is accessible from the head node computer.
1. In the To-do List, click Create a node template. The Create Node Template Wizard
appears.
2. On the Specify Template Name page, type a descriptive name for the template, and
then click Next.
3. If you will be adding compute nodes to your cluster from bare metal:
a. On the Select Deployment Type page, click With operating system, and then click
Next.
b. If the operating system image that you want to use for your deployment is already
listed in the Image Name list, click that image and then jump to step 3.f. If you want
to use a different operating system image, on the Select Operating System Image
page, click Add Image.
c. On the Add Operating System Image window, click Create a new operating
system image, and then type or browse to the location of the Windows setup file for
one of the 64-bit editions of Windows Server 2008.
d. Type a descriptive name for the new operating system image, and then click OK.
e. After the image is added, in the Image Name list, click the image that you want to
use with the template.
f. Optionally, specify if you want to multicast the operating system image during
deployment. For more information, see “1.6. Prepare for multicast (optional)” in Step
1: Prepare for Your Deployment.
g. Optionally, specify if you want to include a product key to activate the operating
system on the compute nodes, and then type the product key that should be used.
Important To create a node template
18
h. Click Next to continue.
i. On the Specify Local Administrator Password for Compute Node page, click Use
a specific password, and then type and confirm the password that you want to use.
j. Click Next to continue, and then jump to step 5 in this procedure.
4. If you will be adding preconfigured compute nodes to your cluster, on the Select
Deployment Type page, click Without operating system, and then click Next.
5. On the Specify Windows Updates page, specify if you want to add a step in the
template to download and install updates using Microsoft Update or the enterprise
Windows Server Update Services (WSUS). Also, you can specify specific updates to be
added to the template. Click Next to continue.
6. On the Review page, click Create.
Note
The node templates that you create with the Create Node Template Wizard include
the most common deployment and configuration tasks. You can add more tasks to
the node templates that you create by using the Node Template Editor. For more
information, see Appendix 3: Node Template Tasks and Properties.
3.5. Add drivers for the operating system images (optional)If you will be deploying compute nodes from bare metal and those nodes require special device
drivers, you will need to add those drivers during the configuration process of your head node.
Drivers must be in the .inf format, and must be accessible from the head node.
The device drivers that you add will be available to all operating system images in the
image store.
It is recommended that you obtain the latest device drivers for your compute nodes from
the Web site of your hardware vendors.
1. In the To-do List, click Manage drivers. The Manage Drivers dialog box appears.
2. To add a driver, click Add.
3. Type or browse to the location of the setup information file for the driver that you want to
add (.inf format), and then click Open.
4. Repeat the two previous steps for all the drivers that you want to add.
5. After you are done adding drivers, click Close.
Note Note To add drivers for the operating system images
19
3.6. Add or remove users (optional)If you will be giving access to the cluster to other members of your organization, you need to add
them as HPC cluster users or HPC cluster administrators. Also, you can remove users or
administrators that were added by default during installation.
The Domain Users group is added as an HPC cluster user during installation. If you do
not want all users in the domain to have access to your cluster, you can remove the
Domain Users group from the list of HPC cluster users, and add a different domain
group specifically created for users of your HPC cluster, or you can add individual domain
users.
1. In the To-do List, click Add or remove users.
2. To add a user to the cluster:
a. In the Actions pane, click Add User. The Select Users or Groups dialog box
appears.
b. Type the user name of the user that you want to add, and then click Check Names.
For more information, on the Select Users or Groups window, click examples.
c. Repeat the previous step for all users that you want to add.
d. After you are done adding users, click OK.
3. To add an administrator to the cluster:
a. In the Actions pane, click Add Administrator. The Select Users or Groups dialog
box appears.
b. Type the user name of the administrator that you want to add, and then click Check
Names. For more information, on the Select Users or Groups window, click
examples.
c. Repeat the previous step for all administrators that you want to add.
d. After you are done adding administrators, click OK.
4. To remove a user or administrator, select it on the Users list, and then click Remove.
Note
You cannot remove the domain Administrator account from the list of cluster
administrators.
Step 4: Add Compute Nodes to the Cluster
Windows HPC Server 2008 simplifies the deployment process of compute nodes by providing
automatic node imaging, automatic naming of nodes, and other capabilities to streamline
Important To add or remove users for the cluster
20
deployment tasks. Also, it provides tools that you can use to monitor the progress of your
deployment.
Unlike previous versions of Windows HPC Server 2008, the default in Windows HPC
Server 2008 is to respond only to Pre-Boot Execution (PXE) requests that come from
existing compute nodes. This default setting is automatically changed when you use the
Add Node Wizard to add nodes from bare metal. Also, you can manually change this
setting in the Options menu, under Deployment Settings.
After creating a node template, you can use the Add Node Wizard to add compute nodes to your
HPC cluster. There are three ways by which you can add compute nodes to your cluster:
Deploy compute nodes from bare metal
Add compute nodes by importing a node XML file
Add preconfigured compute nodes
For more information about each of these three node deployment options, see “1.2. Decide how
to add compute nodes to your cluster” in Step 1: Prepare for Your Deployment.
In this section:
4.1. Deploy Compute Nodes from Bare Metal
4.2. Add Compute Nodes by Importing a Node XML File
4.3. Add Preconfigured Compute Nodes
4.4. Monitor deployment progress
4.5 Cancel the deployment of a node
4.1. Deploy compute nodes from bare metalThe following procedure describes how to add compute nodes to your HPC cluster from bare
metal, by using a node template that includes a step to deploy an operating system image.
To complete this procedure, you must have a template that includes a step to deploy an
operating system image. If you do not have a template that includes a step to deploy an
operating system image, create one by following the steps in “3.4. Create a node
template”, in Step 3: Configure the Head Node.
Before turning on a compute node for this procedure, verify in the configuration of the
BIOS of that computer that the compute node will boot from the network adapter that is
connected to the private network, instead of booting from the local hard drive or another
device, and that Pre-boot Execution Environment (PXE) boot is enabled for that network
adapter.
1. If HPC Cluster Manager is not already open on the head node, open it. Click Start, point
to All Programs, click Microsoft HPC Pack, and then click HPC Cluster Manager.
2. In Node Management, in the Actions pane, click Add Node. The Add Node Wizard
appears.
Important Important Important To deploy compute nodes from bare metal
21
3. On the Select Deployment Method page, click Deploy compute nodes from bare
metal using an operating system image, and then click Next.
4. On the Select New Nodes page, in the Node template list, click the name of a node
template that includes a step to deploy an operating system image.
5. Turn on the computers that you want to add as compute nodes to your cluster.
Computers will be listed on the Add Node Wizard as they contact the head node during
PXE boot. They will be named using the naming series that you specified when you
configured the head node. For more information, see “3.3. Configure the naming of new
nodes” in Step 3: Configure the Head Node.
6. When all computers that you have turned on are listed, click Select all, and then click
Deploy. If you see a node that you do not want to deploy at this time, you can unselect it.
7. On the Completing the Add Node Wizard page, if you will be deploying more nodes,
click Continue responding to all PXE requests. If you will not be deploying more
nodes, click Respond only to PXE requests that come from existing compute nodes.
8. To monitor deployment progress, select the Go to Node Management to track
progress check box, and then click Finish. For more information, see 4.4. Monitor
deployment progress.
4.2. Add compute nodes by importing a node XML fileThe following procedure describes how to add compute nodes by importing a node XML file.
To complete this procedure, you must have a valid node XML file. For more information,
see Appendix 2: Creating a Node XML File.
1. If HPC Cluster Manager is not already open on the head node, open it. Click Start, point
to All Programs, click Microsoft HPC Pack, and then click HPC Cluster Manager.
2. In Node Management, in the Actions pane, click Add Node. The Add Node Wizard
appears.
3. On the Select Deployment Method page, click Import compute nodes from a node
XML file, and then click Next.
4. On the Select Node XML File page, type or browse to the location of the node XML file,
and then click Import.
5. To monitor deployment progress, on the Completing the Add Node Wizard page, select
the Go to Node Management to track progress check box, and then click Finish. For
more information, see 4.4. Monitor deployment progress.
Important To add compute nodes by importing a node XML file
22
4.3. Add preconfigured compute nodesA preconfigured compute node is a computer that has HPC Pack 2008 already installed and that
is connected to the HPC cluster network according to the network topology that you have chosen
for your cluster. After HPC Pack 2008 is installed on all the compute nodes that you want to add
to your cluster, you can use the Add Node Wizard on the head node to add the preconfigured
nodes to your cluster.
The following procedures describe how to add preconfigured compute nodes to your HPC cluster.
The first procedure describes how to install HPC Pack 2008 on the computers that will act as
compute nodes, and the second procedure describes how to add the preconfigured compute
nodes to the cluster.
The computers that you will add to your cluster as preconfigured compute nodes must
already be running Windows Server® 2008 HPC Edition, or another 64-bit edition of the
Windows Server 2008 operating system.
We strongly recommend that you perform a clean installation of Windows Server 2008
before installing HPC Pack 2008. If you want to install HPC Pack 2008 on an existing
installation of Windows Server 2008, remove all server roles first and then follow the
procedures in this guide.
To complete this procedure, you must have a node template that does not include a step
to deploy an operating system image. If you do not have a node template that does not
include a step to deploy an operating system image, create one by following the steps in
“3.4. Create a node template”, in Step 3: Configure the Head Node.
1. To start the HPC Pack 2008 installation wizard on the computer that will act as a compute
node, run setup.exe from the HPC Pack 2008 installation media or from a network
location.
2. On the Getting Started page, click Next.
3. On the Microsoft Software License Terms page, read or print the software license
terms in the license agreement, and accept or reject the terms of that agreement. If you
accept the terms, click Next.
4. On the Select Installation Type page, click Join an existing HPC cluster by creating a
new compute node, and then click Next.
5. On the Join Cluster page, type the computer name of the head node on your cluster,
and then click Next.
6. On the Select Installation Location page, click Next.
7. On the Install Required Components page, click Install.
8. On the Installation Complete page, click Close.
After HPC Pack 2008 is installed on all the compute nodes that you want to add to your cluster,
follow the steps in the Add Node Wizard on the head node to add the preconfigured nodes to
your cluster.
Important Important Important To install HPC Pack 2008 on a compute node computer
23
1. If HPC Cluster Manager is not already open on the head node, open it. Click Start, point
to All Programs, click Microsoft HPC Pack, and then click HPC Cluster Manager.
2. In Node Management, in the Actions pane, click Add Node. The Add Node Wizard
appears.
3. On the Select Deployment Method page, click Add compute nodes that have already
been configured, and then click Next.
4. Turn on all the preconfigured nodes that you want to add to your cluster.
5. After all the preconfigured nodes are turned on, on the Before Deploying page, click
Next.
6. On the Select New Nodes page, in the Node template list, click the name of a node
template that does not include a step to deploy an operating system image.
7. Select the preconfigured compute nodes that you want to add to your cluster. To select all
the preconfigured compute nodes, click Select all.
8. To add the selected compute nodes to your cluster, click Add.
9. On the Completing the Add Node Wizard page, if you will be deploying more nodes,
click Continue responding to all PXE requests. If you will not be deploying more
nodes, click Respond only to PXE requests that come from existing compute nodes.
10. To monitor deployment progress, select the Go to Node Management to track
progress check box, and then click Finish. For more information, see 4.4. Monitor
deployment progress.
4.4. Monitor deployment progressDuring the deployment process of a compute node, its state is set to Provisioning. After the
deployment process is complete, the state changes to Offline. You must bring compute nodes
online before they can process jobs.
You can monitor the progress of the deployment process of compute nodes in Node
Management, and bring online nodes that have finished deploying. You can also see detailed
information for each deployment operation, and any errors that may have occurred.
1. If HPC Cluster Manager is not already open on the head node, open it. Click Start, point
to All Programs, click Microsoft HPC Pack, and then click HPC Cluster Manager.
2. To view information about the deployment operations:
a. In Node Management, in the Navigation Pane, click Operations.
b. To view more information about a specific operation, click that operation. The Detail
Pane will list the log entries for that operation.
3. To view the list of compute nodes that are currently being deployed:
a. In Node Management, in the Navigation Pane, under Nodes, under By State, click
Provisioning.
To add preconfigured compute nodes to your cluster To monitor deployment progress
24
b. To view the list of operations related to the deployment of a specific node, double-
click that node, and then click the Operations tab.
4. To bring online the nodes that have finished deploying:
a. In Node Management, in the Navigation Pane, under Nodes, under By State, click
Offline.
b. Select all the nodes that you want to bring online. To select all nodes that are
currently offline, on the list of offline nodes, click any node and then press CTRL+A.
c. In the Actions pane, click Bring Online.
5. If the deployment process of a compute node fails, the state of that node is set to
Unknown and the health is set to Provisioning Failed. To determine the reason of the
failure, review the provisioning log for that node and the list of operations that were
performed:
a. In Node Management, in the Navigation Pane, under Nodes, under By Health,
click Provisioning Failed.
b. To review the provisioning log for a node, in the views pane, click the node, and then
in the Detail Pane, click the Provisioning Log tab
c. To view the list of operations related to the deployment failure, on the Properties tab,
click View operations. The pivoted view in Node Management will list all the
operations related to that node.
d. To view more information about a specific operation, click that operation. The Detail
Pane will list the log entries for that operation.
4.5 Cancel the deployment of a nodeYou can stop the deployment process of a compute node from HPC Cluster Manager by
canceling the provisioning operations.
1. To view only compute nodes that are currently being deployed, in Node Management, in
the Navigation Pane, under Nodes, under By State, click Provisioning.
2. In the views pane, click the node that you want to stop deploying.
3. To cancel the provisioning operations, in the Detail Pane, on the Properties tab, click
Cancel Operations. The deployment process will stop, the node will be moved to the
Unknown state, and the health for that node will be changed to Provisioning Failed.
To cancel the deployment of a node
25
Step 5: Run Diagnostic Tests on the Cluster
After you have configured your head node and added all compute nodes to the cluster, you
should run diagnostic tests to validate cluster functionality and troubleshoot any configuration
issues.
1. If HPC Cluster Manager is not already open on the head node, open it. Click Start, point
to All Programs, click Microsoft HPC Pack, and then click HPC Cluster Manager.
2. In Configuration, in the Navigation Pane, click To-do List.
3. In the To-do List, click Validate your cluster (under Diagnostics).
4. In the Run Diagnostics dialog box, ensure that the Run all functional tests and All
nodes options are selected, and then click Run.
5. To view the progress of the diagnostic tests and the test results, in Diagnostics, in the
Navigation Pane, click Test Results.
6. To view detailed information about a test, double-click the test. To expand the information
in a section of the test results, click the down arrow for that section.
Step 6: Run a Test Job on the Cluster
After you have finished deploying your cluster, you can run a simple test job to verify that your
cluster is fully functional. The following checklist describes the steps involved in running a simple
test job on your cluster.
Checklist: Run a test job on the HPC cluster
Task Description
6.1. Create a Job Template Create a job template by running the Generate
Job Template Wizard in HPC Cluster
Manager.
6.2. Create and Submit a Job Create and submit a basic job in HPC Cluster
Manager.
6.3. Create and Submit a Job Using the
Command-Line Interface (Optional)
Create and submit a basic job by using the
HPC command-line tools.
6.4. Create and Submit a Job Using HPC
PowerShell (Optional)
Create and submit a basic job by using the
cmdlets in HPC PowerShell.
To run diagnostic tests on the cluster
26
6.1. Create a job templateJob templates simplify the job management of your HPC cluster by helping you to limit the kinds
of jobs that can be submitted to your cluster, what resources are assigned to jobs, and which
users can submit jobs. HPC Cluster Manager includes the Generate Node Template Wizard to
help you create basic job templates
1. If HPC Cluster Manager is not already open on the head node, open it. Click Start, point
to All Programs, click Microsoft HPC Pack, and then click HPC Cluster Manager.
2. In Configuration, in the Navigation Pane, click Job Templates.
3. In the Actions pane, click New. The Generate Node Template Wizard appears.
4. On the Enter Template Name page, type Test Template for the name of the new job
template, and optionally a description. Click Next to continue.
5. On the Limit Run Time page, select the Run jobs no longer than check box, and then
click Next without changing any settings. This will limit all jobs that are submitted using
this template to run for no longer than one minute.
6. On the Set Priorities page, click Next without changing any settings. This will run jobs
that are submitted using this template with Normal priority.
7. On the Set Project Names page, click Next without changing any settings. This will allow
jobs from any project to be submitted using this template.
8. On the Limit Node Groups page, click Next without changing any settings. This will
allow jobs that are submitted using this template to run on any node group.
9. On the Finish page, click Finish.
6.2. Create and submit a jobThis section describes how to submit a job in HPC Cluster Manager that:
Displays a directory list of the files in the C:\Program Files folder of a compute node in your
cluster.
Uses the job template that you created in the previous section, which limits to 1 minute the
maximum duration of time that a job can run.
Runs at low priority.
1. In Job Management, in the Actions pane, click New Job.
2. In Job Details, specify the following job parameters:
a. In the Job template list, click Test Template (the template that you created in
section “6.1. Create a job template”).
b. When you are prompted if you want to change the job template for the job, click Yes.
c. In the Job name box, type Folder Contents.
To create a simple job template To create and submit a job
27
d. In the Priority list, click Lowest.
3. To add a task, click Task List, and then specify the following task parameters:
a. To add a new basic task to the job, click Add.
b. In the Task name box, type a name for the new task.
c. In the Command line box, type dir.
d. In the Work directory box, type c:\Program Files.
e. To add this task, click Save.
4. To limit the job so that it only runs on a specific compute node in your HPC cluster, click
Resource Selection, and then specify the following resource parameters:
a. Select the Run this job only on nodes in the following list check box.
b. Select the check box for one of the nodes in your HPC cluster.
5. To submit the job, click Submit.
6. If you are prompted to enter your credentials, type your user name and password, and
then click OK.
7. To view the progress and the results of the job that you submitted:
a. In Job Management, in the Navigation Pane, click All Jobs.
b. In the views pane, click the job that you submitted.
c. When the state of the job is Finished, in the Details Pane, double-click the task that
you created in step 3.
d. In the Task Properties window, in the Results tab, the Output box will display the
directory list of c:\Program Files for the compute node that you selected in step 4.
e. If you want to copy the results to the clipboard, click Copy output to clipboard.
6.3. Create and submit a job using the command-line interface (optional)You can create and submit a job similar to the job that you created and submitted in the previous
section, using the command-line interface tools that are included with Windows HPC
Server 2008.
1. Open a Command Prompt window. Click Start, point to All Programs, click
Accessories, and then click Command Prompt.
2. To create a new job, type the following command:
job new /jobname:"Folder Contents" /priority:"Lowest" /RunTime:0:0:1
/requestednodes:”<ComputeNodeName>”
Where <ComputeNodeName> is the name of a compute node in your HPC cluster.
3. To add a task to the job, type the following command:
To create and submit a job using the command-line interface
28
job add <JobID> /workdir:"C:\Program Files" dir
Where <JobID> is the identification number for the job, as displayed on the command-
line interface after typing the command in step 2.
4. To submit the job, type the following command:
job submit /id:<JobID>
Where <JobID> is the identification number for the job, as displayed on the command-
line interface after typing the command in step 2.
5. If you are prompted to enter your credentials, type your password, and then type ENTER.
6.4. Create and submit a job using HPC PowerShell (optional)You can also create and submit the same job that you created and submitted in the previous
section, using HPC PowerShell.
For more information about HPC PowerShell, see Appendix 6: Using HPC PowerShell.
1. On the head node, click Start, point to All Programs, and then click Microsoft HPC
Pack.
2. Right-click HPC PowerShell, and then click Run as administrator.
3. If you are prompted by Windows PowerShell if you want to run the ccppsh.format.ps1xml
script, type A, and then press ENTER.
4. To create a new job, type the following cmdlet:
You can also add the HPC PowerShell snap-in from Windows PowerShell.
1. Click Start, point to All Programs, click Windows PowerShell 1.0, and then click
Windows PowerShell.
2. In Windows PowerShell, type the following cmdlet:
Add-PsSnapin Microsoft.HPC
Add the HPC PowerShell snap-in to your Windows PowerShell profileIf you have a Windows PowerShell profile, you can add the HPC PowerShell snap-in to it so that
it is available in every PowerShell session under your user name. For more information about
Windows PowerShell profiles, see Windows PowerShell Profiles (http://go.microsoft.com/fwlink/?
LinkID=119587).
1. Open Windows PowerShell. Click Start, point to All Programs, click Windows
PowerShell 1.0, and then click Windows PowerShell.
2. To edit your profile in Notepad, type:
notepad $profile
3. Type the following cmdlet as a new line in the profile:
Add-PsSnapin Microsoft.HPC
4. To save the profile, in the File menu, click Save.
5. To close Notepad, in the File menu, click Exit.
View Help in HPC PowerShell In-context help is available for HPC PowerShell cmdlets:
To view a list of the cmdlets that are available in HPC PowerShell, type the following cmdlet:
Get-Command –PSSnapin Microsoft.HPC
To view basic help information for a specific cmdlet, type:
Get-Help <cmdlet>
Where <cmdlet> is an HPC PowerShell cmdlet.
To view detailed information for a specific cmdlet, type:
Get-Help <cmdlet> -Detailed
Where <cmdlet> is an HPC PowerShell cmdlet.
To view the help information on the screen one page at a time, type | More at the end. For
example, if you type:
Get-Help New-HpcJob –Detailed | More
To add HPC PowerShell from Windows PowerShellTo add the HPC PowerShell snap-in to your Windows PowerShell profile