Compaq 112726-001 - ProLiant - 6500 Quick start guide

,QWHJUDWLRQ1RWH
September 1999
0096-0699-A
Prepared byECG Technology
Communications Group
Compaq Computer Corporation
Contents
Clustering Overview....................3
Clusters Defined........................3
Causes of Downtime..................3
Cost ofDowntime.......................4
Benefits of Clustering.................4
Summary....................................4
ProLiant Cluster for NetWare
4.2 Overview.................................5
Features and Benefits of the
ProLiant Cluster for NetWare
4.2..............................................5
Novell’s High-Reliability
Solution......................................5
The Failover Process.................6
Hardware Requirements............8
CompaqProLiant Servers..........8
CompaqStorageWorks
RAID Array 4000 Storage
System.......................................9
ClusterInterconnects...............12
Software Components...............13
Novell Software........................13
CompaqSoftware....................13
Planning NovellHigh
AvailabilityServer
Installation..................................15
Verification of NHAS
Installation................................15
ProLiant Cluster for NetWare
4.2 Administration.....................18
Increasing Availability...............18
IP and Applications Failover.....18
Management............................26
Technical FAQs.........................28
'HSOR\LQJWKH3UR/LDQW&OXVWHU
IRU1HW:DUH
Abstract:Compaq is renowned in the Enterprise market with its industry-
leading, highly available, fault-tolerant servers. Compaq servers and storage
deliver on this reputation byoffering the highest level of reliability and
uptime with features suchas:
•Industrystandard PCI Hot Plug and internal hot-plugstorage
•Redundanthot-plug fans, power supplies, and NICs
•Automatic Server Recovery-2 (ASR-2)
•ECC protected memory
•Integrated Management Display (IMD)
•Integrated Remote Console (IRC)
•Compaq SmartStart and Compaq Insight Manager
The Compaq ProLiant Cluster for NetWare 4.2 is designed to help
customers reduce the risk and costs of downtime due to hardware or
softwarefailures. Benefits include:
•Automatic failover protection - In the event of a failure on one server,
the remaining server maintains availabilityof network data and
resources.
•Active servers - Users can nowcluster two active NetWare 4.11 or
NetWare 4.2 servers onthe network, providingfailover support and
significant cost savings, since purchasing a dedicated server for failover
is no longer necessary.
•Shared storage -Support for shared storage devices allows customers to
reduce overall management costs bycentralizing administration and
back-up. Shared storage also allowsmore efficient use of storage space
byeliminating the need to maintain duplicate files on different servers.
•No single point offailure – The newCompaq StorageWorks RAID
Array4000 storage systemprovides redundant fibre channel loops for
No Single Point of Failure configurations.
The ProLiant Cluster for NetWare 4.2 provides the highest levels ofsystem
and application availabilityfor customers with high systemuptime
requirements for NetWare 4.x environments. These levels are necessaryto
meet the requirements of demanding business-critical environments. This
Integration Note details the configuration options and server requirements,
provides insight into the administration, and describes some ofthe features
available with the ProLiant Cluster for NetWare 4.2.

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
Notice
The information in this publication is subject to change without notice and is provided “AS IS” WITHOUT
WARRANTY OF ANY KIND. THE ENTIRE RISK ARISING OUT OF THE USE OF THIS
INFORMATION REMAINS WITH RECIPIENT. IN NO EVENT SHALL COMPAQ BE LIABLE FOR
ANY DIRECT, CONSEQUENTIAL, INCIDENTAL, SPECIAL, PUNITIVE OR OTHER DAMAGES
WHATSOEVER (INCLUDING WITHOUT LIMITATION, DAMAGES FOR LOSS OF BUSINESS
PROFITS, BUSINESS INTERRUPTION OR LOSS OF BUSINESS INFORMATION), EVEN IF
COMPAQ HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
The limited warranties for Compaq products are exclusively set forth in the documentation accompanying
such products. Nothing herein should be construed as constituting a further or additional warranty.
This publication does not constitute an endorsement of the product or products that were tested. The
configuration or configurations tested or described may or may not be the only available solution. This test
is not a determination or product quality or correctness, nor does it ensure compliance with any federal
state or local requirements.
Product names mentioned herein may be trademarks and/or registered trademarks of their respective
companies.
Compaq, Contura, Deskpro, Fastart, Compaq Insight Manager, LTE, PageMarq, Systempro, Systempro/LT,
ProLiant, TwinTray, ROMPaq, LicensePaq, QVision, SLT, ProLinea, SmartStart, NetFlex, DirectPlus,
QuickFind, RemotePaq, BackPaq, TechPaq, SpeedPaq, QuickBack, PaqFax, Presario, SilentCool,
CompaqCare (design), Aero, SmartStation, MiniStation, and PaqRap, registered United States Patent and
Trademark Office.
Netelligent, Armada, Cruiser, Concerto, QuickChoice, ProSignia, Systempro/XL, Net1, LTE Elite,
Vocalyst, PageMate, SoftPaq, FirstPaq, SolutionPaq, EasyPoint, EZ Help, MaxLight, MultiLock,
QuickBlank, QuickLock, UltraView, Innovate logo, Wonder Tools logo in black/white and color, and
Compaq PC Card Solution logo are trademarks and/or service marks of Compaq Computer Corporation.
Microsoft, Windows, Windows NT, Windows NT Server and Workstation, Microsoft SQL Server for
Windows NT are trademarks and/or registered trademarks of Microsoft Corporation.
NetWare and Novell are registered trademarks and intraNetWare, NDS, Novell Directory Services, and
High Availability Server (NHAS) are trademarks of Novell, Inc.
Pentium is a registered trademark of Intel Corporation.
Copyright ©1999 Compaq Computer Corporation. All rights reserved. Printed in the U.S.A.
Deploying the ProLiant Cluster for NetWare 4.2
Integration Note prepared by ECG Technology Communications Group
Second Edition (June 1999)
Document Number 0096-0699-A

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
Clustering Overview
The computer industry has been using a wide range of solutions to reduce system downtime for
years. Many of these solutions have been both expensive and difficult to setup and maintain.
Historically, only mission-critical applications, such as those controlling stock exchange trading
floors and aerospace missions, were deemed important enough to justify proprietary clustering
solutions.
However, as the presence of computer systems in day-to-day business operations has increased,
the amount of acceptable downtime has decreased. Today, a new class of applications exists;
business-critical applications key to business success but not significant enough to justify the high
price tag of a proprietary clustering solution. More and more applications are becoming business-
critical. Computer applications are becoming more widely deployed and their failure causes
immediate lost revenue, decreased productivity, and, potentially, dissatisfied customers.
Poised to meet the increased demands, clustering technology is entering mainstream industry-
standard computing. As a result, new clustering solutions, using industry-standard hardware and
software, provide key clustering features at a lower price than proprietary clustering systems.
Clusters Defined
Traditionally, clustering is the integration of software and hardware that enables a set of loosely
coupled servers and storage to operate as a single system, presenting a single entity to clients. As
a cluster, the group of servers and storage offers a level of availability greatly exceeding the
reliability of a standalone server. This translates into increased performance and greater data
availability for end users.
The ProLiant Cluster for NetWare 4.2 extends the definition by providing the same level of
reliability while maintaining distinct network entities for each server. By defining distinct server
identities, tasks and users may be distributed easily between the distinct servers. With the focus
for clustered systems being reliability, the ProLiant Cluster for NetWare 4.2 delivers again with
automatic failover of all network resources and responsibilities in the event of a failure. Should
one server in the cluster lose access to the shared storage and network, the other server assumes
the responsibilities of the downed server.
Causes of Downtime
Computer downtime is the period of time that a computer system cannot meet the requests of its
users. The following are the leading causes of downtime:
•Planned service - In a cluster, a single server can be taken offline, while the other server takes
on the workload of the downed server. This allows planned service to occur with only
minimal interruption to the overall system operation.
•Software failures - Because clustering provides a mechanism to automatically fail over
processes when a discernible software failure occurs, the overall system operation can
continue with only minimal interruption.
•Hardware failures - If it is determined that a hardware failure will result in system downtime,
the cluster will fail over processes from one node to another, allowing the overall system
operation to continue with little interruption.

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
Cost of Downtime
The driving force behind the high-availability systems has been the elimination of costs
associated with downtime. The cost of downtime formula must include several factors. Among
these are:
•Productivity loss
•IM costs
•Lost transactions
•Customer or end user dissatisfaction
Each factor is weighted differently, depending on the critical nature of each factor as it pertains to
your business and to your specific application systems. For example, downtime during peak hours
of a point-of-sale operation would have a much greater impact on customer satisfaction issues
than downtime in an end-of-day email server backup operation. To understand the true cost of
computer downtime in your business environment, you must examine the impact and contribution
to increased costs of each factor as it applies to your business-critical applications.
Benefits of Clustering
With the causes enumerated and the costs understood, it is clear that the Novell High Availability
Server can eliminate your concerns and ease your tightening networking budget through
•Availability - Simply defined, availability is the measure of how well a computer system can
continuously deliver services to clients. This is dependent upon the system’s ability to
prevent or recover from failures or faults.
•Scalability - Clustering for scalability means increasing performance beyond that of a single
computer node by adding more nodes to the cluster. Performance scalability across cluster
nodes is difficult to achieve and requires not only scalable hardware, but also scalable
software, for example, a parallel database.
•Lower Total Cost of Ownership - By reducing downtime, you can eliminate both the high
costs of administrative support and, at the same time, the lost revenue caused by downtime
and customer dissatisfaction.
Summary
In general, clusters can provide both high availability and scalability for business-critical
applications. In today’s market, experts estimate that approximately 80-90% of clusters are used
to take advantage of the increase in server availability created by clusters.
The use of clusters to reduce computer system downtime has a direct impact on a company’s
revenue and the MIS department budget. The impact depends on your calculated costs of
downtime. In most cases, the cost of installing and maintaining a cluster are likely to be offset by
the reduction in downtime costs, and, as businesses’ reliance on computer systems intensifies, the
cost of downtime is likely to increase.

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
ProLiant Cluster for NetWare 4.2 Overview
Novell products are designed to support incremental network growth, and the ProLiant Cluster for
NetWare 4.2 continues this philosophy with its solid NetWare environment designed to provide
both industrial-strength availability and scalability at the hardware, operating system, and
application levels.
By deploying the ProLiant Cluster for NetWare 4.2, your company can reduce downtime and
operating costs. At the same time, your network can benefit from the latest Internet/intranet and
file and print services in a robust network environment for deploying general-purpose
applications.
Features and Benefits of the ProLiant Cluster for
NetWare 4.2
The ProLiant Cluster for NetWare 4.2 provides these features and benefits:
•Two node active-active cluster configuration
•Automatic failover and failback
•Protection against both hardware and software failures
•Failure detection over dedicated and/or non-dedicated links
•Support for any ODI™, IP compliant network interface card
•Support for Symmetric Multi-Processing (SMP)
•Industry-standard, non-proprietary hardware requirements
•High availability and lower total cost of ownership through:
−Support for Fiber Channel Arbitrated Loop (FCAL)
−Novell Directory Services (NDS) support
−NetWare Loadable Module™ (NLM™)-based installation
−Dynamic user licensing
−Application fallback via IP
Novell’s High-Reliability Solution
The ProLiant Cluster for NetWare 4.2, featuring Novell High Availability Server (NHAS),
increases network reliability through the use of two NetWare 4.11 or NetWare 4.2 servers with a
shared disk subsystem. By pairing two independent servers in an active-active configuration, the
NHAS allows each server to access and mount the other server's data if there is a failure on a
server in the cluster.
Each server monitors the status of the other server over three redundant lines of communication:
•NetWare Link – each server can monitor the presence of the other across the local network.
•Disk Link – NHAS uses quorum arbitration to allow each server to monitor the other server’s
shared storage connection and prevent concurrent access to the same data.
•Optional Dedicated Link – by creating a dedicated link between the two servers, each server
can monitor the presence of the other more reliably, eliminating false failures caused by
network traffic or prolonged disk inactivity.

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
The Failover Process
Costandby State
During normal operation, shown in Figure 1, each server node in the ProLiant Cluster for
NetWare 4.2 operates as if it were a completely independent entity on the network. Server A and
Server B are allocated their own dedicated volumes on the Compaq RAID Array 4000 shared
storage. NHAS controls access for the shared storage and prevents one server from accessing the
other server’s volume. This controlling mechanism is vital to the operation of the cluster, since it
prevents a split-brain scenario, wherein two servers may try to access the same data
simultaneously; obviously, this could result in data corruption. In the given configuration, each
server is operating in the costandby state, that is, independently but waiting for the other server to
fail.
Compaq RAID Array 4000
Volume 1
Accessed by
Node A only
Volume 2
Accessed by
Node B only
Server
Node A
Server
Node B
Local Area Network
Fibre Channel Storage Hub
Data
Inter-server Monitoring
Data and Monitoring
Network Clients
Dedicated Interserver Direct Link
Fibre Channel Arbitrated Loop
NetWare
Interconnects
Server A
Volume A
Logical Connections
Server B
Volume B
Figure 1. ProLiant Cluster for NetWare 4.2 Costandby State

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
Failover and Recovery States
Figure 2 illustrates the failover and recovery process. If no communication between the servers
can be established across any of the three lines of communication, the server that has lost
connections to the network and storage will enter failover state. At the same time, the other node,
in this case, Server B, will enter the recovery state (Figure 2A). Upon entering the recovery state,
Server B assumes responsibility for Server A volumes and clients (Figure 2B). Security and
licensing are maintained by transferring NDS rights to the surviving server. Total failover time
corresponds to the time necessary to detect the failure and complete the failover process. During a
failover, the remaining server must mount the new volumes, launch any active server-based
applications that were running on the failed device, and transfer security and license information.
NHAS also allows manual failover of shared volumes at any time by pressing F at the NHAS
console. By forcing a failover, you can remove a server from the network for scheduled
maintenance without interrupting access to that server’s volumes.
Disabled Server
Server
Node B
Local Area Network
Fibre Channel Storage Hub
Data
Inter-server Monitoring
Data and Monitoring
Network Clients
Dedicated Interserver Direct Link
Fibre Channel Arbitrated Loop
NetWare
Interconnects
Compaq RAID Array 4000
Volume 1
Accessed by
Node B only
Volume 2
Accessed by
Node B only
Server A
Volume A
Logical Connections
Server B
Volume B
A. Node B Entering Recovery State B. Node B Assuming
Responsibility for
Node A’s Volumes
Figure 2. Failover and Recovery

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
Hardware Requirements
As shown in Figure 3, a typical ProLiant Cluster for NetWare 4.2 hardware configuration consists
of two ProLiant servers with Fibre Channel Host Controllers, a Fibre Channel Storage Hub, and
a Compaq RAID Array 4000 storage system with a Fibre Channel Array Controller.
Fibre Channel Array
Fibre Channel
Storage Hub 7
ProLiant Server
Fibre Cable
Fibre Channel Array Controller
Fibre Channel Host Controller
Network Interface Controller
(Dedicated Interconnect)
Gigabit Interface Converter
Interconnect Cable
Figure 3. Typical ProLiant Cluster for NetWare 4.2 Configuration
Whether your intention is to upgrade an existing server or purchase a new Compaq server, the
following minimum hardware requirements must be met for a ProLiant Cluster for NetWare 4.2
configuration.
Compaq ProLiant Servers
The primary components of a cluster are the servers. The initial release of the ProLiant Cluster
for NetWare 4.2 supports a two-node cluster, where each node is a server. The key software
component that imparts high availability to the cluster servers is Novell High Availability Server
(NHAS). Throughout the development of the NHAS software, Compaq has been a partner with
Novell to ensure that Compaq ProLiant servers are the ideal platforms for NHAS. Compaq has
logged countless hours testing ProLiant clusters, which have successfully passed Novell’s NHAS
Certification test suite. This rigorous suite of tests ensures that the cluster works as an integrated
unit, not just as a collection of individual components.
CompaqRAIDArray 4000

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
The following Compaq ProLiant servers have been certified for use with both NHAS and Fibre
Channel:
•Compaq ProLiant 1850R – Intel Pentium II
•Compaq ProLiant 3000 – Intel Pentium II
•Compaq ProLiant 5500R – Intel Pentium II Xeon
•Compaq ProLiant 6400R – Intel Pentium III Xeon
•Compaq ProLiant 6500 – Intel Pentium II Xeon
•Compaq ProLiant 7000 – Intel Pentium II Xeon
NOTE: Check the Compaq website at http://www.compaq.com to obtain an up-to-date list of cluster-
certified servers. Each server in the cluster must have one local hard drive. In addition, Compaq
recommends 256MB of RAM in each server.
In addition to the increased availability of clustering, Compaq ProLiant servers include many
reliability features that provide a solid foundation for effective clustered server solutions. They
offer excellent reliability through redundant systems, hot-pluggable components, and server
health and monitoring. The high-availability features of ProLiant servers are a critical foundation
of an effective ProLiant Cluster for NetWare 4.2 deployment and the cornerstone of mission-
critical computing.
Compaq StorageWorks RAID Array 4000 Storage
System
The ProLiant Cluster for NetWare 4.2 is based on a cluster architecture known as Shared Storage
Clustering, in which clustered servers share access to a common set of hard drives. Novell High
Availability Server requires all shared data to be stored in an external storage system to
implement the necessary quorum arbitration.
The Compaq RAID Array 4000 storage system brings significant advantages to clustering,
including:
•100MB/s throughput
•Server-to-hub and hub-to-storage distances of up to 500 meters for short-wave and 10
kilometers for long-wave (per Fibre Channel link, 25 kilometers cumulative for all
links)
•Increased connectivity and ease of use
•Reliable transmission of data in shared configurations
•Hot-pluggable drives
•Hot-pluggable power supply (in redundant power supply configurations)
•Manageability through Compaq Insight Manager
By using standard short-wave Fibre Channel cabling, the Compaq RAID Array 4000 can be
placed up to 500 meters from the Fibre Channel Storage Hub, and the Fibre Channel Storage Hub
can be placed up to 500 meters from the server. By using long-wave Fibre Channel cabling,

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
these distances can be increased to 10 kilometers (and a total of 25 kilometers on the Fibre
Channel loop).
NOTE: The maximum distance between a single server and the Fibre Channel Storage Hub is 10
kilometers. A NetWare connection or dedicated direct link will need to connect each cluster node.
Connecting the nodes over long distances will require a combined use of Ethernet network devices, such as
hubs, switches, or repeaters.
The Compaq RAID Array 4000 storage system consists of four primary components:
•Fibre Channel Array
•Fibre Channel Storage Hub
•Fibre Channel Array Controller
•Fibre Channel Host Controller
When planning your Compaq RAID Array 4000 storage system, remember to include any
necessary cabling. Also, the Fibre Channel Array Controller is not a separate component; rather,
the array controller is an integral part of the Fibre Channel Array.
Compaq Fibre Channel Array
The Compaq Fibre Channel Array is the storage chassis that contains the disk drives, power
supply, and Fibre Channel Array Controller. The Fibre Channel Array can hold twelve 1-inch or
eight 1.6-inch Wide-Ultra SCSI-3 drives. The Compaq Fibre Channel Array comes in both a rack
and a tower model.
Each ProLiant Cluster for NetWare 4.2 must have at least one Compaq RAID Array 4000 storage
system as external shared storage. As many as five Compaq RA4000s can be attached to the Fibre
Channel Storage Hub 7 and, with the Fibre Channel Storage Hub 12, up to ten Compaq RA4000s
are supported. In both cases, two ports on the hubs are used by the server connections. With the
Compaq RA4000 connected to both servers through the Fibre Channel Storage Hub, you can
independently manage volumes on the servers in the cluster as needed by your application and
environment.
Compaq Fibre Channel Storage Hub
The Fibre Channel Storage Hub creates the data path that allows one or more Compaq RAID
Array 4000 storage systems to communicate with one or more ProLiant servers. The Compaq
ProLiant Cluster for NetWare 4.2 requires at least one Fibre Channel Storage Hub; both servers
and all shared storage for the cluster must connect to this hub.
As the names imply, the Fibre Channel Storage Hub 7 and the Fibre Channel Storage Hub 12
contain seven and twelve ports, respectively. As mentioned, using the hubs in a cluster
environment will consume two ports for server connections, allowing five Compaq RAID Array
4000 storage systems on the Fibre Channel Storage Hub 7 and 10 arrays on the Fibre Channel
Storage Hub 12.
For higher availability, two Fibre Channel Storage Hubs can be configured as a redundant pair.
When used in combination with redundant Fibre Channel Array Controllers and redundant Fibre
Channel Host Controllers, a complete backup data path can be configured between the cluster
servers and the Compaq RAID Array 4000, as described below under “Redundant Fibre Channel
Loop for Increased Availability.”

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
NOTE: Currently, Fibre Channel Storage Hubs cannot be daisy chained, that is, hubs cannot be
connected in series.
Compaq Fibre Channel Array Controller
One or two Fibre Channel Array Controllers, which reside in the Compaq RAID Array 4000
storage system, provide RAID and caching capabilities and provide the Fibre Channel interface to
connect to the server. An important aspect of a high-availability system is fault tolerance,
traditionally accomplished by implementing RAID technology. Hardware RAID is an integrated
part of the Compaq Fibre Channel Array Controller, located within the Fibre Channel Array
chassis.
The Fibre Channel Array Controller supports RAID 0, 1, 4, and 5, as well as online capacity
expansion and pre-failure notification for hard drives. In addition, the Fibre Channel Array
Controller provides ECC (Error Checking and Correcting) memory for user-configurable read
and write cache. The cache on the Array Accelerator is equipped with onboard rechargeable
batteries, ensuring that the cached data is safe even with equipment failure or power outage. For a
complete list of features and accompanying descriptions, refer to the Fibre Channel information
on the Compaq website, http://www.compaq.com.
Typically, a Compaq RAID Array 4000 storage system is configured with a single controller.
However, for a higher level of availability, two array controllers can be configured as a redundant
pair, such that one controller is a backup for the other, as described under “Redundant Fibre
Channel Loop for Increased Availability,” below.
Compaq Fibre Channel Host Controller
The host controller, available as a PCI adapter, interfaces between the server and the Fibre
Channel Storage System through the fiber optic cable. At least two PCI Fibre Channel Host
Controllers, one for each cluster node, are required in your Compaq ProLiant cluster. For higher
availability, Fibre Channel Host Controllers can be configured in redundant pairs, such that one
controller is a backup for the other, as described under “Redundant Fibre Channel Loop for
Increased Availability,” below.
Redundant Fibre Channel Loop for Increased Availability
For increased availability, a redundant Fibre Channel loop can be configured, which provides two
redundant data paths between each cluster server and the Compaq RAID Array 4000 storage
system. One data path is configured as the primary path and is initially active. The other data
path is configured as a standby path and is initially inactive. The standby data path serves as a
backup for the primary path. The redundant data path configuration requires the following
additional components:
•A second, standby Fibre Channel Host Controller installed in each server
•A second, standby Fibre Channel Array Controller installed in the Compaq RA4000
•Recommended but not required: A second, standby storage hub with connections to the
standby host controllers and to the standby array controller.
Figure 4 shows an example of this configuration.

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
Primary Array
Controller Standby Array
Controller
Standby
Storage Hub
Primary
Storage Hub
Primary
Host Controller
Standby
Host Controller
Primary Path
Backup Path
Figure 4. Redundant Fibre Channel Loop
As the figure shows, a redundant data path is configured between the cluster servers and the
Compaq RAID Array 4000. The redundant components operate initially in standby mode. If any
component (host controller, storage hub, array controller, or cable) in the primary path fails, the
primary path becomes inactive, the standby path becomes active, and data transmission continues
without interruption over the standby path.
Cluster Interconnects
The cluster interconnect is a data path over which nodes of a cluster communicate. This type of
communication is termed intra-cluster communication. The ProLiant Cluster for NetWare 4.2
uses the standard public network controller card (NetWare Link), an optional Dedicated Link, and
the Fibre Channel Disk Link to implement intra-cluster communication. At a minimum, the
interconnect consists of two network adapters (one in each server) and a cable connecting the
adapters to a public network hub to be utilized as the NetWare Link.
The cluster nodes use the interconnect data path to:
•Communicate cluster status
•Send and receive heartbeat
Since Novell High Availability Server does not mirror data between servers, the transfer of the
heartbeat is very small (just the heartbeat or information about the health of the other server in the
cluster is sent over the links).
IMPORTANT: NHAS 1.0 requires IPX as the cluster communication protocol for the Direct
Link connection.

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
Network Adapters
Standard Compaq Ethernet adapters are the network adapters of choice for Compaq ProLiant
clusters. Either 10Mbit/sec or 100Mbit/sec Ethernet may be used.
Software Components
Novell Software
Novell offers Novell High Availability Server to complement Novell NetWare 4.11 or NetWare
4.2. As of June 1999, NHAS will only run with Novell NetWare 4.11 and NetWare 4.2. Previous
versions of NetWare are not supported.
Compaq Software
Compaq offers an extensive set of features and optional tools to support effective configuration
and management of your clustered servers:
•Compaq SmartStart and Support Software CD
•Compaq System Configuration Utility
•Compaq Insight Manager
•Compaq Support Software Diskette for Novell NetWare (Novell SSD)
Compaq SmartStart and Support Software CD
SmartStart is the intelligent way to configure your Compaq ProLiant Server. SmartStart uses a
step-by-step process to configure your server, load the system software, and integrate Compaq
utilities, thereby ensuring that your server delivers maximum dependability and supportability.
For information concerning SmartStart, refer to the Compaq Server Setup and Management pack.
Compaq System Configuration Utility
The SmartStart and Support Software CD also contains the Compaq System Configuration
Utility. This utility is the primary means to configure hardware devices in your server, such as I/O
addresses, boot order of disk controllers, etc. For information concerning the Compaq System
Configuration Utility, see the Compaq Server Setup and Management pack.
Compaq Insight Manager
Compaq Insight Manager, loaded from the Compaq Management CD, is an easy-to-use software
utility for collecting server information. Insight Manager performs the following functions:
•Monitors fault conditions and system status
•Monitors shared storage and interconnect adapters
•Forwards server alert fault conditions
•Allows remote server administration
In Compaq servers, each hardware subsystem, such as disk storage, system memory, and system
processor, has a robust set of management capabilities. Compaq Full Spectrum Fault

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
Management notifies of impending fault conditions and keeps the server up and running in the
unlikely event of a hardware failure. For information concerning Compaq Insight Manager, see
the Compaq Server Setup and Management pack.
Compaq Support Software Diskettes for Novell NetWare (Novell SSD)
The Compaq SSD for Novell NetWare contains device drivers and utilities that enable you to take
advantage of specific capabilities offered on Compaq products. The drivers and utilities are
provided for use only with Compaq hardware.
For additional details on the features contained in the Novell SSD, visit
http://www.compaq.com/support/files/server/softpaqs/Netware/NSSD.html
The Compaq SSD for Novell NetWare is included in the Compaq Server Setup and Management
pack.

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
Planning Novell High Availability Server
Installation
Listed below are some guidelines to keep in mind before installing NHAS on the ProLiant Cluster
for NetWare 4.2:
•It is imperative that the volumes created in the shared storage are not mounted on both
servers. Depending on your requirements, you may want to allocate a volume to each server,
keeping in mind that NHAS will later prohibit the other server from accessing that volume
until a failover occurs.
To check the status of each server’s allocated storage device, at the command prompt issue
the following commands in the order listed:
CONFIG
SCAN FOR NEW DEVICES
LIST DEVICES
VOLUME
•Be sure to install the latest NetWare Support Pack, which is available from Novell, and the
Compaq Support Software for Novell NetWare (Novell SSD).
•The installation of NHAS requires administrative privileges.
•The servers in the cluster must reside in the same NDS tree to fully utilize the NDS
capabilities. Note that NHAS does not require that the servers in the cluster reside in the same
NDS container, but it is recommended for manageability sake.
•Each server in the cluster must have an internal hard drive that is not part of the shared
subsystem. If NetWare is installed with the shared drive subsystem connected, the installer
should make sure that SYS: volumes of both clustered servers are installed on the internal,
non-shared drives. Having the SYS: volumes on the shared subsystem would be confusing
and potentially damaging.
•NetWare volumes can span multiple drives, but if a volume spans more than one drive, all of
the drives that it spans must be shared, or all of them must be not shared. It is also possible to
have more than one volume on a single shared NetWare partition. Some hardware
configurations will allow volumes to be mounted on multiple servers at the same time but
once NHAS is installed this is not allowed. If a single volume is mounted on multiple servers,
at the same time, data loss will occur.
Verification of NHAS Installation
These procedures will help you validate a proper NHAS installation on both NHAS servers.
•After the NHAS installation, restart your Compaq ProLiant server. NHAS should load, and
you should have the capability to switch to the HA Server screen by pressing Ctrl+Esc and
selecting the HAServer Screen from the menu
•At the HA Server screen, there will be generalized information about the status of your
NHAS cluster. Make sure that the SHARED VOLUME is MOUNTED, and the SERVER is
in COSTANDBY mode on both servers. Also, verify that NETWARE LINK, DISK LINK,
and DIRECT LINK (if utilized) are all labeled CONNECTED.

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
•To perform a simple verification of operation, press Fon the keyboard. Compaq recommends
that you pause the NHAS software during these tests, by typing P, so an automatic failback
cannot take place. Pressing F will force a server to fail. The failed server should report the
status as FAILOVER. Soon after the induced failure, both servers should recognize that they
are both operational. The surviving server will return to COSTANDBY state when the
previously failed server enters the RECOVERY state. If successful, both servers in the NHAS
cluster will fallback to COSTANDBY state and all quorum arbitration links will be
connected.
•Some other commands to note:
C - Changes the first server into the COSTANDBY state.
ESC - Used to exit the program. When the Escape button is pressed, you are presented with a
confirmation prompt. Choose Yes to exit, or choose No to remain in the application.
F - Initiates a manual failover on the first server. When you press f, you are prompted to
confirm that you want to manually failover the first server and acquire the disk devices from
the second server.
H - Displays the Help screen.
P - Pauses the AutoSwitch feature. When you press P, the AutoSwitch feature is paused and
automatic failover or rollback will not take place. When you pause NHAS, the status turns
red to indicate that the server cannot change states. To unpause the NHAS Console, press p
again. Note that the Pause feature does not affect a transitional state. For example, once the
failover process has reached the ENTERING FAILOVER state, the Pause feature will not
stop the failover.
R - Changes the server into the RECOVERY state without changing the second server to the
FAILOVER state. Any shared volumes are dismounted and any shared drives are released. If
you want to failover the system, press F on the second server, which will force first server
into the RECOVERY state.
•If the failover was unsuccessful, check the network configuration and verify that it is properly
set up.
•If further problems persist, check the AUTOEXEC.NCF file for the following entries:
REM HA Server Installation
Search Add Sys:\HASERVER
Set Auto Restart After Abend=0
Load VincaCom
Load VincIPX (master/slave)=address, card SLUNBIND
Load HASERVER
LOAD HASTRUST
If any of these entries are incorrect, you should either edit the file to reflect the correct
information or reinstall NHAS.
•If problems continue, check SYS:\HASERVER\HASERVER.LOG for errors. Contact your
local support representative if necessary.

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
•If you wish to cause a manual failover on the cluster containing Server A and Server B, with
Server B being the surviving server, initiate the failure from the console of Server A. Server
B would then be the surviving server and Server A would enter recovery.
•If the Disk Link fails, then there is a more serious problem somewhere along the shared drive
channel. If the problem is in the channel between one server in the cluster and the shared
drive subsystem, then that server will be disarmed and its shared drives will have already
become inaccessible. The volumes will not be mounted on the other server unless all other
links fail or a manual failover is invoked. If the problem is in the shared subsystem, then
some or all of the shared drives will become inaccessible and one or both servers will be
disarmed. A Disk Link failure alone will not cause a server failover.

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
ProLiant Cluster for NetWare 4.2
Administration
Outlined below are some helpful guidelines to assist you in administering and supporting the
ProLiant Cluster for NetWare 4.2.
Increasing Availability
Intra-cluster Communication
Should the servers experience a disruption in intra-cluster communication, they will initiate the
failover process. Obviously, this condition is only acceptable in the event of an actual failure that
disrupts the server’s ability to service client requests. Therefore, the reliability of the connections
between the servers in the cluster is a direct measure of the clustered system’s ability to maintain
costandby operation, wherein both servers are fully operational.
There are several ways to reduce the possibility of a disruption of intra-cluster communication.
Each method shares an underlying objective: to create redundant paths over which monitoring
will continue if the another path is disrupted. The ProLiant Cluster for NetWare 4.2 provides this
redundancy through the three lines of communication between servers: the NetWare link, the
optional dedicated link, and the disk link.
Compaq extends this redundancy with a feature that configures two Compaq Ethernet adapters or
two ports on a single adapter, such that one is a hot backup for the other. This feature, called
Advanced Network Fault Detection and Correction, is available on all Compaq 10/100 Fast
Ethernet products. This feature reduces the possibility of downtime due to network failure by
allowing the administrator to configure a redundant interconnect for each communication path.
In a redundant interconnect configuration, each communication path is configured with a primary
link and a backup link. If a component in the primary link fails, communication continues over
the backup link. The redundant interconnect feature significantly enhances the availability of
clustered systems.
Quorum Arbitration
Each server uses quorum arbitration to monitor and verify that the other server is operational. By
monitoring the other server’s usage of a reserved region on the shared volumes, both servers can
be sure that the other is still accessing the storage array. In the event that a server does not update
the information in this shared region, the cluster becomes aware of a potential failure. In order to
prevent false failovers, NHAS will verify the disk link status 4 times. If after the fourth
verification, or roughly 20 seconds, Server B has not accessed the reserved region of the disk,
Server A will set the disk link status to disconnected. If the NetWare link (and dedicated link, if
implemented) status is also disconnected, NHAS will initiate a failover to Server A.
IP and Applications Failover
Configuration
If you want your IP-based applications to failover to the other server when their host server fails,
you must set up your ProLiant Cluster for NetWare 4.2 to redirect all IP traffic from the failed

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
address to the surviving server. The following example provides basic information on the
necessary steps to setup a simple IP failover scenario.
There are three key commands to add a failed IP address to a surviving server. They are ADD,
DEL,and DISPLAY. Their formats and syntax are shown below:
add secondary ipaddress <failed server IP address>
del secondary ipaddress <failed server IP address>
display secondary ipaddress
In the examples above, <failed server IP address> is the IP address of the IP-based application to
be failed-over if the host server fails. The ADD command assigns the IP address to the surviving
server in the event of a failover. The DEL command removes the IP address from the server when
recovery occurs. The DISPLAY command is for informational purposes only and can be used
from the command line to view all secondary IP addresses.
The following lines below show the usage of the commands in a .NCF batch file one can
configure. Comments are remarked with a “#” sign:
# enables duplicate IP addresses support under NetWare
set allow ip address duplicates=on
# adds all secondary IP addresses
add secondary ipaddress 123.45.67.8
add secondary ipaddress 123.45.67.9
# launches your application
run-now.nlm
# deletes all secondary IP addresses except for the one you want to launch
del secondary ipaddress 123.45.67.8
These .NCF batch files should be created in the search directory of each server node, such as
SYS:\SYSTEM. Each server will have varying IP addresses input in their .NCF files to discern
which secondary IP addresses are to be deleted and which are to be kept during a failover and
recovery. IP handling during a failover should add the secondary IP to the surviving server so that
it may take over client requests for the failed node. At the same time, NHAS IP handling should
remove the same secondary IP from the failed node. Two separate batch files will be made for
each server: one for the failover to a node and the other to failback to the recovered node. In
addition, the HASERVER.INI file must contain the FailOverCommand= and
RecoveryCommand= entries to automatically reconnect IP-based applications after failover. The
names of the respective failover and recovery .NCF files must be added to the commands for a
successful failover and recovery of the IP-based application.
Note: The sample HASERVER.INI on page 43 of the NHAS User’s Guide is incorrect. See
Novell’s Errata sheet for the correct layout. This is only a document change. Your
HASERVER.INI file should be in the correct state.
HASERVER.INI Information
To enable IP application failover, it is vital you know and understand the command lines inside of
the HASERVER.INI, so that you can tune NHAS specifically to your requirements. Below is a
sample HASERVER.INI with a description of each line:

'HSOR\LQJWKH3UR/LDQW&OXVWHUIRU1HW:DUH 
0096-0699-A
[HAServer]
Server=PL7000-2 This identifies the NetWare Servers in the NHAS Cluster.
Server=PL7000-1 This identifies the NetWare Servers in the NHAS Cluster.
NWLinkTimeout=30 This sets the NWLink timeout value before NHAS triggers the link
down. Keep in mind that this value is also used when the disk and
direct links fail.
FailoverDelay=30 This specifies the delay time of NHAS after the links are down to
failover the disk volumes to the surviving cluster node.
Broadcast=Off .
Enabling this will broadcast a network message informing users the
status of the servers in the cluster when a failover or recovery
happens.
Rollback=Automatic Setting this value to manual will disable automatic failback of the
volumes should the server that went down go back online.
DiskLink=258E5776 This is a unique number used for quorum arbitration by the disk link.
Note that if this value is accidentally changed or erased, we
recommend that you reinstall NHAS.
[Server.PL7000-2] This subsection details the information for this particular server,
PL7000-2.
Volume=VOL2 Volume settings identify which volumes should remain exclusive to this
server.
Volume=SYS Volume settings identify which volumes should remain exclusive to this
server.
PrimaryLicense=20614660 This License Number is used as the main Server License for this node
during normal operating conditions.
RecoveryCommand=REC1.NCF By specifying a command or an executable script, when a server
enters the RECOVERY STATE, i.e. has failed, it processes the
commands listed here. This specified command should call a file that
contains all commands necessary for a successful IP failover.
FailoverCommand=FAIL1.NCF By specifying a command or an executable script, when a server
enters the FAILOVER STATE, i.e. is taking over for the other failed
server, it processes the commands listed here. This specified
command should call a file that contains all commands necessary for a
successful IP failover.
CostandbyCommand=CO2.NCF By specifying a command or an executable script, when a server
enters the COSTANDBY STATE, i.e. Normal condition, it processes
the commands inputted here. This specified command should call a
file that contains all commands necessary for a successful IP failover.
DiskLinkSector=0 This specifies the segment of the disk used for quorum arbitration by
the Disk Link.
SecondaryLicense=60022796 This License Number is used to append the other Server’s License
should a failover happen.
[Volume.PL7000-2.VOL2] This subsection details the information for this particular server,
PL7000-2, in respect to Volume 2.
Shared=True Setting the value to false will disable the ability for a drive to failover.
Mounted=True Setting the value to false will disable NHAS from automatically
mounting this volume.