Docstoc

Unisys_Sentinel_Tech

Document Sample
Unisys_Sentinel_Tech Powered By Docstoc
					A highly integrated suite of server and system
management tools for the Unisys ES7000 enterprise
servers
Introduction
Server Sentinel is a centralized operations environment for your server. It provides
world-class data center server management capabilities for greater business continuance
and system resiliency within your server environment. It can dramatically enhance the
manageability, availability, and performance of your server environment while
considerably reducing total cost of operations.

Server Sentinel simplifies system management, automates corrective and preventive
measures to avoid a system failure, and increases the reliability, scalability, and
performance of Unisys servers.

There are four major drivers for IT management today: Cost containment, security,
availability, managing complexity. These drivers have prompted a move toward self-
managing server systems (includes HW, OS, management software). The Server Sentinel
Program directly impacts three capabilities…

•   Provisioning
       – Capacity planning, asset discovery, self-configuring tools
•   Optimization
       – Workload management, performance management, end-user accounting &
           billing
•   Availability
       – Root-cause analysis, failover clustering, component isolation, automatic
           recovery, minimization of unplanned downtime

Unisys Server Sentinel is a highly integrated suite of server and system management
tools for the Unisys ES7000 enterprise servers. These tools enhance the manageability,
availability, and performance of your server environment all while reducing your total
cost of operations. Server Sentinel is ideally suited for deployment on enterprise server
platforms in a MS Datacenter infrastructure and as the core of a mission-critical
application environment.

Sentinel Management Strategy

The overriding purpose of Server Sentinel is to provide users of Unisys ES7000
Enterprise Servers with:
 A simple, flexible, and robust computing environment that meets or exceeds changing
   operational requirements.
 An enterprise class management system that supports service-level business processes
   at the lowest possible operating costs.
 A secure server management environment that scales, integrates, and delivers the
   agility to meet our customers’ business objectives.

Features that Support the Sentinel Strategy
•   Set-up Assistant for easy installation and system configuration
•   Server Builder for simple drag-and-drop partition configuration
•   Processor Autonome to optimize processor usage
•   Call Home for automatic notification to Unisys of fail conditions*
•   Health Monitoring provides rapid notification of fault conditions*
•   Diagnostic Data Collection
•   Self Healing takes automated corrective actions to faults*
•   EZupdate delivers automated notification of updates
•   Customizable Security Features including vulnerability analysis, baseline security
    analysis, and role-based security*
•   Remote Access for lights out data center management*
•   AppManager 7.0 for enhanced application management and integration with
    enterprise management software

* These features are one of the seven operational and management tasks available from the Task
Map located on the Server Sentinel Interface. They are described in detail further down in this
document under the heading Server Sentinel Task Categories

Set-up Assistant

The Server Sentinel Setup Assistant simplifies your installation and configuration process
by enabling you to use a wizard interface to collect Server Sentinel configuration data.
This information is saved to a single Server Sentinel configuration information file. This
configuration information file is used during the system installation process to provide a
consistent view of the initial configuration to all Server Sentinel components. If required,
you can input configuration information over time. You will be prompted to supply the
completed configuration information file during the installation process.

Server Builder

The Server Sentinel Server Builder feature enables you to easily create and operate
partitions. The Server Sentinel Server Builder tab, available from the Server Sentinel user
interface, graphically displays the cells in your system and the partitions in which they
are organized.

Using the Server Builder feature, you can quickly configure one or more cells into a
partition, monitor the active and inactive partitions, start or stop active partitions, and
activate partitions that are currently inactive. Physical cells in the system can be logically
organized into multiple partitions, which you can activate or deactivate with a few clicks.

Processor Autonome

Processor Autonome enables you to manage processor usage. The Processor Autonome
service takes snapshots of processor rates over a period of time. These snapshots are
combined to establish processor usage by the running processes. Using the information
you learn from the Processor Autonome service, you can use the Processor Autonome
utility to assign certain processes to specific affinity groups and obtain more effective
processor utilization. You can rerun the Processor Autonome utility at any time to
establish updated processor rates to meet the changing needs of the organization.


Diagnostic Data Collection

Diagnostic Data Collection enables you to automatically gather diagnostic information
about the Server Sentinel components installed on your system, including the Sentinel
Management Server, Service Processors, and partitions. The information collected
includes log files, database files, event logs, and registry data. Unisys support personnel
use this information to analyze and solve system-related problems without the delays
associated with the manual collection of diagnostic information. Diagnostic Data
Collection is available through the Server Sentinel interface.

EZupdate

The EZupdate feature provides information about updates for all computers managed by
the Sentinel Management Server—for example, partitions and Service Processors—in a
single report. Reviewing one consolidated report to determine whether any Server
Sentinel components are in need of updates saves you time and effort. You can download
any of the updates available on the Unisys Product Support Web site through a
convenient EZupdate link on the Server Sentinel interface.

Note: Internet connectivity is only required from the Sentinel Management Server and
not from the various computers managed by the Sentinel Management Server.

AppManager

A key component of Server Sentinel is the integrated AppManager® program which
helps reduce the support costs associated with the management of distributed clients and
server networks and helps provide increased efficiency within the network. Server
Sentinel now includes AppManager 7.0, which offers several new enhancements,
including improved performance, scalability, security, and monitoring. These
enhancements enable IT administrators to more efficiently monitor and manage
their systems. See the Server Sentinel User’s Guide for more information.

Server Sentinel Infrastructure & Architecture
•   One Server Sentinel program running from a single Sentinel Management Server can
    manage up to 50 Unisys ES7000 systems (200 Servers)
•   Server Sentinel can be used in heterogeneous environments with multiple operating
    systems including Windows, Linux and Unisys ClearPath OS’s
•   The two categories of hardware required in a Unisys Server Sentinel environment
        – One or more servers on which the Server Sentinel software is installed
               • Such servers are called Sentinel Management Servers
        – And the servers that are managed by Server Sentinel
              •   Such systems are called managed systems and have a Server Sentinel
                  agent installed in the Server and a Service Engine in the platform
                  hardware

Server Sentinel is a comprehensive software suite that provides users with:
• Powerful Management and Monitoring …
       – Server Sentinel automates processes required to manage systems proactively
• An easy to use interface …
       – Access to the enterprise of Unisys systems from anywhere and anytime for
           complete lights out management
• Integration with workgroup and enterprise management software programs such as…
       – Microsoft MOM, Tivoli, NetIQ, CA Unicenter, and BMC




Server Sentinel Task Categories
Server Sentinel organizes operational and management tasks into seven categories:

   Health Monitoring. Delivers policy definition and event notification to monitor the
   health of the hardware and the Microsoft Windows software environment. Monitoring
   is continuous and in real time, providing rapid notification of fault conditions.

   Self Healing. Enables the Server Sentinel to take automated corrective actions in
   response to hardware and software faults, allowing the system to continue to operate
   without human intervention.

   Security. Provides security features for intrusion prevention, vulnerability analysis,
   network security, and user access. You can customize these features to meet your
   specific server security requirements.

   Call Home. Provides a proactive capability that automatically detects and responds to
   system service events. Service events are automatically reported to the Support
   Center for rapid resolution.

   Remote Access. Ensures that normal operations can be conducted from any location
   at any time or a connection can be made to a technical expert at a remote facility to
   diagnose hardware and software faults.

   Health Reporting. Provides detailed performance, inventory, and availability
   reporting.

   System Configuration. Provides intuitive configuration and setup procedures for the
   server platform and Server Sentinel.
Task Map
Each of the Server Sentinel tasks is grouped in task categories. Access to each task is
controlled by the user role on the service processor and by the user role on the partition.
See the following task map for a list of tasks that are available in each category, and for
the minimum user role required to access each task.

Tasks are only visible if they are available to the user role and the task applies to the
particular server configuration.

Task Map

                                                                 Minimum User Role on
Task Category                   Task Name                        Target Server and Service
                                                                 Processor
                                Hardware alerts                  Observer
                                System alerts                    Observer
                                Event Notification               Administrator
                                System, partition, power
                                domain, and environmental        Observer
Health Monitoring
                                status
                                                                 Administrator on target
                                EZupdate
                                                                 machine
                                Cluster diagnostics              Administrator
                                Cluster groups status            Administrator
                                Predictive Health actions        Observer
                                Partition recovery options       Operator
                                Policy-based management
                                                                 Operator
                                rules
                                Windows Server processor
                                                                 Administrator for partitions
                                autonome
                                                                 Administrator for partitions
                                Windows Server processor
                                                                 observer for Service
Self Healing                    affinity
                                                                 Processors
                                                                 Administrator for partitions
                                Windows Server
                                                                 observer for Service
                                automated log backup
                                                                 Processors
                                Cluster self-healing policy      Administrator
                                AppManager Operator
                                                                 Administrator
                                Console (Expert Mode)
                                Self-healing capabilities        Observer
                   (How-To)
                   Unattended operations
                                                 Observer
                   capabilities (How-To)
                   Security alerts               Observer
                   Server Sentinel Firewall      Operator
                   Partition software
                                                 Operator
                   vulnerability analysis
                   Partition software baseline
Security                                         Operator
                   security analysis
                                                 Local operator at the
                   Server Sentinel users         Sentinel Management
                                                 Server
                   Enterprise Server security
                                                 Observer
                   (How-To)
                   Configuration wizard          Operator
                   Service request status        Observer
                   Send test service request     Operator
Call Home          Service request delivery
                   schedule (only available on
                                                 Observer
                   systems prior to Server
                   Sentinel 4.0)
                   Health Advisor                Administrator
                   Partition desktop             Operator
                   Service Processor desktop

                   This link will initiate
                   Remote Desktop
Remote Access      Connection if the Service     Operator
                   Processor is running
                   Windows XP or VNC if
                   the Service Processor is
                   running Windows 2000.
                   Console Manager Card          Administrator
                   Cluster health                Administrator
                   Cluster Manager log and
                                                 Administrator
                   report
                   Configuration verification    Operator
Health Reporting
                   Operating system
                                                 Observer
                   configuration
                   Partition hardware
                                                 Observer
                   properties
                             Partition hardware units
                                                              Observer
                             inventory
                             Software inventory               Observer
                             System hardware
                                                              Observer
                             properties
                             System hardware units
                                                              Observer
                             inventory
                                                              Administrator for partitions
                             Windows Datacenter
                                                              observer for Service
                             certification
                                                              Processors
                             Windows Server
                                                              Observer
                             operational reports
                                                              Administrator for partitions
                             Windows Server packages          observer for Service
                                                              Processors
                                                              Administrator for partitions
                             Windows Server snapshots
                                                              observer for Service
                             and monitors
                                                              Processors
                             Hardware platform                Operator is the minimum
                             management                       role to access hardware
                             See System Tasks for a           platform management.
                             description of the tasks         Menu items that require the
                             available from hardware          administrator role are
                             platform management.             hidden from operators.
                             System definition                Administrator
                             Monitored Windows
                                                              Operator
System Configuration         Server computer systems
                             Peripheral Power Control
                             System
                             (Only available on systems       Administrator
                             with native MCP
                             partitions)
                             Additional management
                                                              Operator
                             tools
                             Maintenance mode                 Administrator

Tasks available after selecting Additional Management Tools

                             Cluster Manager                  Administrator
Additional Management
                             Cluster Verifier                 Administrator
Tools
                             Application Sentinel             Administrator
Understanding User Roles
The user roles of administrator, operator, and observer are implemented using Microsoft
groups. User roles are established during Server Sentinel installation.

Here is a general definition of each role:

       Administrator

       Can administer the management environment (for example, configure the
       management software or delete items from a log) as well as manage systems.
       Functions that require administrative privileges to execute are included in this role.

       Operator

       Can manage systems. Operator has read/write access to most management
       functions. Functions that require administrative privileges are excluded from this
       role.

       Observer

       Can observe systems. Observer has read access to most management functions.

If you want to change user roles, close Server Sentinel and log on to Server Sentinel with
a user name and password from a different user group. For example, if you are an
observer and you want to perform an operator function, close Server Sentinel then log
back on with a user name and password from the operator group.

For more information, see the installation documentation appropriate for your system.




Health Monitoring
Purpose

Unisys health monitoring software monitors your system’s resources such as Disk space,
memory availability, page faults, disk throughput, memory allocation, status of the
network interface card (NIC) and much more…this includes the ability to view and
configure the policies that set the criteria for monitoring…
The health-monitoring capabilities of Server Sentinel deliver policy definition and event
notification to monitor the health of the hardware and of the Microsoft Windows
Monitoring is continuous and in real time, providing rapid notification of fault conditions.
For example, real-time views of CPU and memory usage for each partition in your
enterprise server are presented in graphical representations and all on one screen—the
Health Dashboard—for quick access.

Detailed information about each partition or about any system event is accessible with a
single click on the Health Dashboard.

User Benefit

The Health Monitoring feature contributes directly to the ES7000’s extremely high rate
of reliability and availability (often measured as high as 99.996%). Health Monitoring
indirectly contributes to a lower total cost of ownership since it can identify pending
failures and allow Sentinel and the ES7000 platform top take corrective actions before
expensive downtime even occurs.

Description

Server Sentinel provides convenient access to a Health Monitor page that enables
you to view and configure the policies that set the criteria for monitoring. To access this
page, select the Self-Healing category of Server Sentinel, and then click Policybased
management rules.

The Configure Health Monitoring Policies page contains the following policy groups:
• System Resources
        These policies determine the status of disk space, memory availability, page
        faults,
        disk throughput, process handles, and memory allocation.
• Hardware Resources
        These policies monitor hardware for the status of network interface card (NIC)
        packet receive and transmit errors, NIC failure, and SMART disk errors.
• Critical Services
        These policies support the monitoring and self-healing capabilities of Server
        Sentinel,
        including Call Home, AppManager, clustering, and SNMP.
• Critical Software
        These policies support the monitoring and self-healing capabilities of Server
        Sentinel, including monitoring for the presence of the Unisys SNMP Agents, the
        status of Call Home Health Advisor connectivity, clustering events in the event
        log, and unexpected service states.
• Long-Term Analysis
        These policies cover the collection of data about the normal range of values on
        this system for several monitored metrics.
Monitoring Partition Activity
Server Sentinel simplifies system management by providing a consolidated view of the
operating environments on each partition of your enterprise servers. The existing
Microsoft tool set, Unisys SNMP enterprise server object manager extensions, and up-to-
date Web technology are all incorporated to expose both open and Unisys enterprise-
specific Management Information Base (MIB) objects. The combination of these tools
enables you to monitor and manipulate MIB objects associated with each component
residing on the enterprise server.

Server Sentinel includes partition management tasks that provide

      Real-time access to the operating characteristics and real-time views of the
       networking and system components for each partition
      A consolidated view of critical configuration, status, and performance information
       relative to the Windows, UnixWare, Linux, MCP, and OS 2200 operating
       environments
      Wizards with which to define policy-based management rules on system objects
       such as processors, memory, and disk volume utilization
      A report of installed software packages on any given partition and the current
       software versions for each package
      A view of critical events and management policy exceptions over the last 24-hour
       period

Alert Notifications
Viewing Alert Notifications
Alert notifications provide information about the health and security of the system being
monitored. For example, alert notifications can inform you of such events as abnormal
voltage, temperature, and power readings as well as unauthorized access of system files.

Viewing Hardware Alerts
Hardware alerts occur when hardware units fail or reach critical operating thresholds.

To view hardware alerts, select Health Monitoring and then choose Hardware alerts.

Viewing System Alerts
System alerts occur when a business object is out of policy. If processor or memory
utilization exceeds the associated threshold value, or if a AppManager alert is present,
there may be a system alert.
To view system alerts, select Health Monitoring and then choose System alerts.

In addition, from the System Alerts page, you can click the System Messages link to see a
real-time 24-hour history of all the Simple Network Management Protocol (SNMP) traps
sent by the enterprise server. SNMP traps (or notification messages) are sent to the Status
Messages page each time a change or an error occurs in a managed object.

Viewing Security Alerts
Security alerts are surfaced to provide notification about virus attacks and about such
activities as unauthorized access of system executable files, registry, databases.

To view security alerts, select Security and then choose Security alerts.

Viewing Predictive Health Actions
Predictive health actions are surfaced based on these Predictive Health System Policies.

To view predictive health actions do one of the following from Server Sentinel. Either of
these actions displays the Predictive Health Actions page.

      Select Self Healing and then choose Predictive Health actions.
      Click the Predictive Health on the Health Dashboard.

Understanding Event Notifications
Server Sentinel issues event notifications when system events such as partition, Service
Processor, operating system, maintenance, and recovery errors occur on the system. For
example, if an attempt to stop the partition fails, the action taken for this event might be
an e-mail notification sent to the system administrator.

Event notifications are also issued to notify you when software or firmware updates are
available for you to download from the support site.

Viewing Event Notifications
You can view a report that shows a history of the events and actions taken for a particular
event.

To view an event notification report

   1. On Server Sentinel, click Health Monitoring and then click Event Notification.

       The Events Report page appears.
   2. For more details, click Task Help in the upper-right corner of the Events Report
      page.

Configuring Event Notifications
Before you can be notified of system events that require your attention, you need to
specify the type of action to be taken when an event occurs. For example, you can specify
that an e-mail or a script be run in response to a particular event.

To configure event notifications:

   1. On Server Sentinel, click Health Monitoring and then click Event Notification.

      The Events Report page appears.
   2. Click Configure Event Actions at the top of the page.

      The Configure Event Actions page appears.
   3. For more details, click Task Help in the upper-right corner of the Configure
      Event Actions page.

Obtaining Updates
If updates to the Server Sentinel software and platform firmware are available, you can
download them by using the Server Sentinel EZupdate feature or by using the Unisys
Product Support Web site.




    Caution

To avoid corrupting data, stop all partitions before updating Service Processor software.
Update both master and standby Service Processors, if available. After updating the
software, back up the Service Processors.


Internet Access Requirements Depending on your system, Internet access is required to
obtain updates through EZupdate, as summarized in the following table.

Server        Management Server,      Service Processor           Partition
Sentinel      Client Workstation, and
Level         Monitored Server
3.x           Access required.        Access required.            Access required.
4.0 and 4.1   Access required.        Not supported; use the      Access required only if
                                      Product Support Web         no management server
                                          site.                   is available.
4.2 and          Access required.         Access required only if Access not required.
higher                                    no management server
                                          is available.

If you access EZupdate from the Start menu, the system must have Internet access.

Obtaining Updates Using EZupdate
EZupdate periodically checks the Unisys Product Support Web site for available updates
to Server Sentinel and platform firmware and notifies you if updates are available.

You can access EZupdate from the Server Sentinel user interface on all systems as
outlined in the preceding table. This is the preferred method.

EZupdate is not supported on systems running IA 64-bit partitions and Server Sentinel
4.3.

Alternatively, to access EZupdate from Service Processors, client workstations,
monitored servers, or management servers, click the Start menu, point to Programs and
then Unisys Server Sentinel, and click EZupdate. Internet access is required if you use
this method.

         If updates are available on client workstation or monitored server desktops, the
          EZupdate icon is displayed in the system tray, and a balloon message appears
          periodically until you download the available updates or choose to be reminded at
          another time.
         If updates are available on the Server Sentinel user interface, the EZupdate
          Available icon is prominently displayed.

Perform the following steps to download the available updates using the EZupdate
feature of Server Sentinel:

   1. Perform one of the following steps:
           Click the following configuration summary icon:

                   Then select System Tools, and click EZupdate.

                  Click the EZupdate Available icon.

          Depending on your system, the View Latest Report page or the System Report
          page appears.

   2. Click Task Help in the upper-right corner for more information on downloading
      and installing the updates.
   3. Copy the files to all systems that need software or firmware updates.
         Note: Do not download the software or firmware updates directly onto the
         embedded Service Processor desktop because there is insufficient disk space.
         Instead, download the updates to a network share and map a drive to the share.

   4. Be sure to update all partitions when you update the software and platform
      firmware on Service Processors, client workstations, monitored servers, and
      management servers.

Obtaining Updates Using the Unisys Product Support
Web Site
If your system does not include access to the Internet, you can download Server Sentinel
software and platform firmware updates from the Unisys Product Support Web site by
using a workstation that is connected to the Internet.

Updates to Server Sentinel software and platform firmware are available in the form of
Interim Corrections (ICs) and quick-fix releases (QFRs). ICs and QFRs are replacement
modules that contain fixes for your software or platform firmware. Any customer with a
maintenance agreement is entitled to use these modules.

To access the latest updates and apply them to your system, perform the following steps:

   1. Access the Unisys Product Support Web site at http://www.support.unisys.com.
   2. Log in with your user name and password.

         The Product Support Home page is displayed.

   3. Select your system type and click Go.

         The Support page for your system is displayed.

   4. Click Releases.

         The releases for the system are displayed.

   5. In the Fixes column, select the relevant release to see any updates.
   6. If necessary, close all Server Sentinel windows before installing the Server
      Sentinel updates.
   7. Review the description of the update, and then follow the instructions provided to
      download and install it.

Notes:

        If present, suspend the Cisco Security Agent before installing the Server Sentinel
         updates on the Service Processor and management server.
      Uninstall the Cisco Security Agent before installing the Cisco Security Agent
       updates.
      Install the Cisco Security Agent updates on the Service Processor and
       management server only if they previously included this software.

Before you can be notified of system events that require your attention, you need to
specify the type of action to be taken when an event occurs. For example, you can specify
that an e-mail or a script be run in response to a particular event.

To configure event notifications

   1. On Server Sentinel, click Health Monitoring and then click Event Notification.

      The Events Report page appears.
   2. Click Configure Event Actions at the top of the page.

      The Configure Event Actions page appears.
   3. For more details, click Task Help in the upper-right corner of the Configure
      Event Actions page.

Partition Management Tasks
Real-Time Views of Operating Characteristics and
Components
Real-time views of CPU and memory usage for each server in your enterprise server are
presented in graphical representations in the center of the Server Sentinel home page,
known as the Health Dashboard. The Health Dashboard provides the consolidated view
as well as providing access to detailed information about each server.

The Health Dashboard uses the existing Microsoft tool set, Unisys SNMP enterprise
server object manager extensions, and up-to-date Web technology to expose both open
and Unisys enterprise-specific MIB objects. The combination of these tools enables you
to monitor and manipulate MIB objects associated with each component residing on the
enterprise server.

By clicking any server to the right of the Health Dashboard, you have access to a wealth
of operating characteristics and components for each server, for example:

      Hardware information, such as IRQs, NICs, and storage controllers
      BIOS, processor, and processor utilization information
      Disk information, including disks, and disk volumes
      Memory properties
       Operating system information, such as devices, services, shares, domain access
        data, printer queues, Transaction Server data (MCP only), currently running
        software, and a list of installed software (OS 2200 only)
       Networking information, including interfaces, TCP/IP, IP address table, routing
        table, mask table (MCP only), SNMP details, and network connection groups
       Enterprise Output Manager jobs and job history

Accessing Server Details Through the Health
Dashboard
Note: Access to partition details is granted based on your user role.

To access detailed information about a specific partition:

       Click the server name from the list on the right of the screen to populate the health
        dashboard with general information about the server.
       Click the words View Details to access detailed information.

       Note: A server can be a Windows server or a server running the MCP, OS 2200,
       Linux, or UnixWare operating system.

Viewing Configuration, Status, and Performance
Information
The detailed objects and properties described in Real-Time Views of Operating
Characteristics and Components are also available on one screen as a comprehensive
report.

Accessing This Information
Note: Access to this report is granted based on your user role.

To access this report:

   1. Select the Health Reporting category from the Server Sentinel navigation pane.

   2. Select the task labeled Operating System Configuration.

Defining Policy-Based Management Rules
You can set management policies for your system business objects: processor, memory,
disk volume utilization, services desired state, and page faults. SNMP Agents are invoked
to provide visibility to this data.
The Server Sentinel SNMP extensions include

       Network information agent - Reports configuration information about the
        installed network card and its associated driver.
       System configuration information agent - Reports configuration information about
        the current system, including information about the installed BIOS, processor,
        memory, and I/O devices and controllers.
       Storage information agent - Monitors the disks and file systems and reports
        operating conditions.
       System software information agent - Reports the device drivers, services, and
        software components installed on the system.
       System statistics information agent - Monitors page faults, memory usage, and
        processor usage, and reports exceptions that exceed expected values.

Accessing This Information
Note: Access to this task is granted based on your user role.

To access the policy-based management rule wizards

   1. Select the Self Healing category from the Server Sentinel navigation pane, and
      then select the task labeled Policy-based management rules.

       A list of computer names appears.

   2. Click the desired computer name. From here you can set policies for different
      business objects.

Reporting on Installed Software Packages
At times there may be a need to produce a list of all installed software applications on
any Windows partition on your enterprise server or Service Processor. A task within
Server Sentinel provides such a report. It lists the software in alphabetical order and
includes the release version of each package.

Accessing This Information
Access to this report is granted based on your user role.

To access the installed software report

   1. Select the Health Reporting category from the Server Sentinel navigation pane,
      and then select the task labeled Software inventory.

       A list of computer names appears.
   2. Click the desired computer name. The Software Inventory Report appears.

Viewing System Alerts and Policy Exceptions
Whenever a change or error occurs in a managed object on your enterprise server, an
SNMP trap is logged for that event. A rolling 24-hour history of these SNMP traps is
easily accessible within Server Sentinel on the System Alerts page.

If you receive a warning in the System Alerts page, you can either take corrective action
directly from the page or use Policy-Based Management Rules to modify the associated
threshold to an acceptable level. When you return to the System Alerts page, you might
see that a Normal message was added. Note that the warning message will remain in the
queue until the queue is emptied at the end of the day.

Accessing This Information
Note: Access to this page is granted based on your user role.

To access the System Alerts page select the Health Monitoring category from the
Server Sentinel navigation pane, and then select the task labeled System alerts.

The resulting page displays all events for all partitions on your enterprise server.




Self-Healing
Purpose

More than just monitor your system, Sentinel can also take corrective actions to repair
faults. This self-healing feature is another unattended operation that contributes
significantly to your servers’ uptime. Self-healing actions can include reconfiguration of
hardware, re-initialization or reboot of the OS, or restarting of operating system services.
Normally when a fault is detected, sentinel alerts the operator, and if necessary, initiates
Call Home.

Predictive Health is a primary feature of the Server Sentinel self-healing capabilities. It
provides real-time monitoring, detection, and prediction of unusual conditions to prevent
system downtime.

User Benefit

The most obvious benefit is increased availability of the enterprise server environment
with no (or at least minimal) human intervention – which contributes to reduced
administration costs. Often the corrective actions can occur before any alerts are even
detected.

Description

The Self-Healing feature of Server Sentinel includes:
• The ability for the platform to continue operation with minimal human intervention
• Out-of-the-box self-healing of the OS environment
• Health Monitoring focuses on early predictions of system health problems
       – Service failures often are preceded by a period of instability and poor
           performance and a degradation of end-user service levels
• Combining Health Monitoring with Self-healing and end-user service levels are often
   restored without effecting operations
• Server Sentinel health-monitoring notifications often cause automatic actions that
   correct or even avoid a problem

Some self-healing actions require redundant hardware or Microsoft Cluster Service
(MSCS) to facilitate automatic actions

These automated self-healing actions can include:
• Isolating memory, including cache, when multiple uncorrectable errors are detected
   during memory initialization
• Reinitializing or electronically isolating a failing component
• Detecting and recovering from memory, I/O, processor, clock, and Service Processor
   failure
• Rebooting a partition and restarting a service
• Deleting files or folders
• Initiating a recovery program or script
• Moving a failing application to another cluster node
• Providing event notification
• Correcting single-bit memory errors
• Rebooting partitions that have stopped responding
• Restarting critical Windows services

If MSCS is installed on even one server, other recovery actions include…
          Moving specified cluster groups to another node
          Taking offline or bringing online specified cluster groups\
          Reinitializing specified cluster groups

Predictive Health
The Server Sentinel Predictive Health environment includes the following capabilities:

              Health monitoring
              Self healing
Health Monitoring
Server Sentinel includes built-in hardware and software monitoring capabilities that are
activated at installation. Some of these capabilities monitor current conditions, while
others monitor conditions over time to predict areas that might cause problems later.

In many cases, Server Sentinel offers the self-healing capability paired with health
monitoring. That is, Server Sentinel monitors the health of a system and, in response to
unhealthy indications, heals the system. These capabilities result in improved system
availability.

You can view all monitoring events from the Sentinel Health Dashboard. When a
predictive alert is generated, the Predictive Health text glows yellow until you close the
alert. When a nonpredictive alert is generated, the Alerts indicator at the bottom of the
page displays a red X.

To view a list of predictive alerts, access the Predictive Health Actions page. To view a
list of nonpredictive alerts, access the Hardware Alerts, System Alerts, or Security Alerts
page as appropriate. For information on accessing these alert pages, see Viewing Alert
Notifications.

On the alerts pages, you can perform the following tasks for a specific alert:

      Click Details to view detailed information about the alert.
      Click Acknowledge to indicate awareness of the alert.
      Click Close to indicate that the issue causing the alert has been addressed.

The following monitoring activities are initiated immediately after software installation:

      Hardware environment monitoring (Service Processor redundancy, voltage,
       temperature, impeller, and power supply)
      Resource monitoring (available disk space, available virtual memory, and
       processor utilization)
      Critical Windows service monitoring for failure
      Network error monitoring (network interface card (NIC) status and input and
       output errors)
      Disk error monitoring (SMART disk errors and disk throughput)
      Platform monitoring (single- and double-bit errors in memory and cache, and
       hardware unit failures)

Some monitoring is performed over time and uses trend analysis to predict possible
failures. This type of monitoring includes

      Disk throughput (when disk read time is degrading)
      Process handle leaks (when the number of process handles is increasing)
      Memory leaks (when the number of private bytes is increasing)
On a system that has Health Monitor installed, you can view and configure the policies
that set the monitoring criteria. For more information, refer to Predictive Health System
Policies.

If Microsoft Cluster Service is installed on even one server, you can activate the
following additional monitoring capabilities:

      Processor and memory utilization, page fault rate, and page pool fragmentation
       are monitored to ensure that they are within acceptable ranges.
      Application-specific events are monitored for most applications that use the
       Windows application event log for notification.
      Cluster-specific events are monitored (cluster faults, cluster log size, and other
       cluster health information).
      Cluster verifier can perform periodic tests of the cluster service, the cluster
       network, and the cluster shared disk.

You can use the cluster manager operator console to customize monitoring rules to
trigger automatic self-healing actions. The rules can use any of the following information
to trigger a self-healing action:

      Processor and memory utilization, page fault rate, and page pool fragmentation
      Cluster-specific events (cluster faults, cluster log size, and other cluster health
       information)
      Network events pertaining to the public or private LAN
      Application-specific log entries

These rules can also cause operator notifications by e-mail, pager, or event log entry as
well as by an entry on the Server Sentinel Predictive Health Actions page.

You can use the AppManager Operator Console to deploy Knowledge Scripts® with
custom parameters and actions. This activity extends the monitoring capabilities of
Server Sentinel to include

      Resource monitoring (logical disk activity, physical disk activity, network
       activity, processor utilization overall or by process, memory utilization overall or
       by process, server utilization, main consumers of memory or processor resources,
       and number of open shared files)
      Content monitoring (event log entries and file changes)
      Windows service status (service up, down, changing, or hung)
      Windows process status (up or down)
      Windows registry changes
      System availability (uptime)

You can also configure Knowledge Scripts to cause operator notifications by e-mail or
event log entry as well as by an entry on the Server Sentinel System Alerts page.
Self Healing
Because health monitoring focuses on early predictions of system health problems, the
resulting self-healing actions can often be taken before end-user service levels begin to
degrade. Many service failures are preceded by a period of instability and poor
performance—a degradation in end-user service level. In many cases, the combination of
Server Sentinel health monitoring and self-healing capabilities can eliminate the
degradation period and restore end-user service levels without affecting operations—all
without manual intervention.

The Server Sentinel health-monitoring notifications often cause automatic actions that
correct or avoid a problem. Some of these actions require redundant hardware or
Microsoft Cluster Service to enable the self-healing capability. These automatic actions
include

      Isolating memory, including cache, when multiple uncorrectable errors are
       detected during memory initialization
      Reinitializing or electronically isolating a failing component
      Detecting and recovering from memory, I/O, processor, clock, and Service
       Processor failure
      Correcting single-bit memory errors
      Rebooting partitions that have stopped responding
      Restarting critical Windows services

You can use the AppManager Operator Console to define self-healing actions in response
to error conditions detected by Knowledge Scripts. This automatic activity includes

      Terminating processes that are exceeding predefined processor utilization
       thresholds
      Initiating a stopped service or a service that has stopped responding
      Deleting a set of files or folders to free up disk space
      Initiating a Windows program to take further recovery actions
      Rebooting the server

If Microsoft Cluster Service is installed on even a single server, you can take additional
recovery actions. These actions include

      Moving specified cluster groups to another node
      Taking offline specified cluster groups
      Bringing online specified cluster groups
      Reinitializing specified cluster groups
      Initiating a Windows application to take further recovery actions

Predictive Health System Policies
The Health Monitor service monitors your server and reports conditions that meet criteria
for predictive alerts and system alerts. The service generates both warnings and errors
based on the criteria, which are established by the health monitoring policies.

Note: A warning event indicates a potential problem and is a lower priority than an error,
which indicates an actual problem or failure.

Predictive events are reported on the Predictive Health Actions page and server events
are reported on the System Alerts page. See Viewing Alert Notifications for information
on accessing the different alert pages.

Accessing the Health-Monitoring Policies
On a system that has Health Monitor installed, you can view or modify the health-
monitoring policies on the Configure Health Monitoring Policies page. To access this
page, perform the following steps from Server Sentinel:

           1. Click Self Healing.
           2. Click Policy-based management rules.
           3. Select the computer name for which you want to view or modify health
              monitoring policies.

Understanding the Health Monitoring Policies
The health-monitoring policies are categorized in the following groups:

              System resources
              Hardware resources
              Critical services
              Critical software
              Long-term analysis

The following text describes the policy groups.

System Resources
These policies cover the monitoring of the system to determine the status of disk space,
memory availability, page faults, disk throughput, process handles, and memory
allocation. In addition to enabling or disabling these policies, you can specify the error
thresholds and other criteria used in monitoring.
When a system resource policy is enabled, Health Monitor generates alerts if the
conditions specified in the policy occur.
By default, the system resources policies are enabled.
System resource policies generate alerts for the following conditions.
Predictive Warnings                              Server Errors and Warnings
Disk space in use exceeds the specified        Disk space in use exceeds the specified
warning threshold.                             error threshold.

Virtual memory usage exceeds the               Virtual memory usage exceeds the
specified warning threshold.                   specified error threshold.

Processor usage exceeds the specified          Processor usage exceeds the specified
warning threshold and queue depth for a        error threshold and queue depth for a
sustained period.                              sustained period.

Potential disk throughput problem exists.      Available physical memory is below the
The time spent servicing read requests and     specified error threshold.
the number of requests to be serviced
exceed specified thresholds for a sustained    Page fault rate exceeds the specified error
period.                                        threshold.

Sustained upward trend occurs in the
number of process handles allocated to a
specified process.

Sustained upward trend occurs in the
number of private bytes allocated to a
specified process.

Hardware Resources
These policies cover the monitoring of hardware for the status of network interface card
(NIC) packet receive and transmit errors, NIC failure, and SMART disk errors. In
addition to enabling or disabling these policies, you can specify the error thresholds used
in monitoring certain resources.
When a hardware resource policy is enabled, Health Monitor generates alerts if the
conditions specified in the policy occur.
By default, the hardware resources policies are enabled.
Hardware resource policies generate alerts for the following conditions.
Predictive Warnings                             Server Errors and Warnings
Number of packet receive errors on the          Operating status of the NIC does not
network interface card (NIC) exceeds the        match the administrative status.
specified error threshold.

Number of packet transmit errors on the
NIC exceeds the specified error threshold.

Transmit queue depth on the NIC exceeds
the specified error threshold.
SMART disk error is reported in the error
log.

Critical Services
These policies cover the services that support the monitoring and self-healing capabilities
of Server Sentinel. These services include Call Home, AppManager, clustering, and
SNMP.
When a critical service policy is enabled, Health Monitor monitors the status of the
service. If that service is not running, Health Monitor generates an alert and attempts to
restart the service if the service itself is not disabled.
By default, the critical services policies are enabled.
Critical services policies generate alerts for the following conditions.
Predictive Warnings                                 Server Errors and Warnings
                                                    Critical service that has not been disabled
                                                    is not running.

                                                 Critical services include those needed for
                                                 the operation of Call Home, clustering,
                                                 AppManager, and the Service Processor.

                                                 If Health Monitor can restart the service, it
                                                 reports a server warning. If Health Monitor
                                                 cannot restart the service, it issues a server
                                                 error.

Critical Software
These policies cover the software that supports the monitoring and self-healing
capabilities of Server Sentinel. For example, these policies include monitoring for the
presence of the Unisys SNMP Agents, the status of Call Home Health Advisor
connectivity, clustering events in the event log, and unexpected service states.
When a critical software policy is enabled, Health Monitor generates alerts if the
conditions specified in the policy occur.
By default, the critical software policies are enabled.
Critical software policies generate alerts for the following conditions.
Predictive Warnings                              Server Errors and Warnings
                                                 One or more of the five SNMP Platform
                                                 Agents is absent.

                                                 Cluster Health event is reported in the
                                                 error log.

                                                 Cluster Load Balancer event is reported in
                                                the error log.

                                                Call Home has been unable to transmit a
                                                packet in more than 14 days.

                                                A service is in an unexpected state.

Long-Term Analysis
These policies cover the collection of data about the normal range of values on the
system for several monitored metrics.
By default, the long-term analysis policies are disabled.




Heartbeat, Recovery, Faults, and Dumps
Understanding Heartbeat
The Service Processor has the ability to receive a periodic signal, called a heartbeat, from
each partition to determine if the host operating system is functioning. If a signal is not
received within a specified time period, the system automatically begins a predetermined
set of actions. By selecting actions for the system and each partition, you can determine if
and how the system and partition recover when an operating system does not send a
heartbeat signal.

Understanding Recovery
Recovery is the ability of the system to restart partitions after detecting a hardware or
operating system failure. How the system recovers depends on the settings that you
choose for the system and each partition during the configuration process.

System Recovery Attribute
The system can take certain actions to repair faults and recover partitions. The system can
analyze the error and determine what to do automatically. For example, if a unit fails, the
system can automatically bring the unit down, set the access denied attribute, and reboot
the partition if necessary.

You can set the system recovery attribute, Failing Unit Removal Enabled, on the System
Settings page.

System recovery depends on the settings of partition recovery and heartbeat attributes,
particularly for shared units; see Interdependence of Recovery Actions later in this topic.
Partition Recovery Attributes
Different recovery attributes apply to different partition types and systems. However, all
recovery attributes work in conjunction with the system recovery attribute and various
partition stop, heartbeat, and timeout attributes to recover, reboot, or shut down a
partition in case of failure.

You can set partition recovery attributes on the Partition Settings page.

Interdependence of Recovery Actions
The settings for the system and partition recovery and reboot actions can affect each
other. For example, assume that recovery actions for the system are not enabled (false)
and automatic reboot for a partition is enabled (true). If a unit fails, the system does not
remove the failed unit from the partition. However, the system attempts to initialize all
units in the partition, including the failed unit, and reboot it. If units are initialized
successfully, the partition reboots with the same configuration. Otherwise, the reboot
fails.

If recovery actions are not enabled, partitions that enable automatic reboot should not
share units, because the system cannot initialize a shared unit that is being used in another
partition.

Understanding Halts and Dumps
A halt is a planned or unplanned stop of a partition or the entire system. A dump is a
report of the state of the system and partitions at the time a halt occurs. You can control
the behavior of a partition or the system during stops and initiate recovery by setting stop,
halt, timeout, and dump attributes on the Partition Settings page.




APPMANAGER
What Is the AppManager Program?
This topic provides an overview of the AppManager® program:

      Centralized Management
      Capabilities
      Architecture
      Managing tools
Centralized Management
Server Sentinel provides a comprehensive software solution to monitor and manage the
performance, availability, and server health of an enterprise server. A key component of
Server Sentinel is the integrated AppManager program, which helps reduce the support
costs associated with the management of distributed clients and server networks and
provides increased efficiency within the network.

System administrators can view all servers from a central, easy-to-use console, providing
complete visibility of critical server resources across the enterprise.

From the Operator Console, system administrators and operators can

      Optimize system performance and availability.
      Detect current and potential problems.
      Automate system management tasks and corrective actions.
      Gather real-time and historical performance information.
      Maintain an inventory of managed computers.

Server Sentinel takes advantage of the Knowledge Scripts® feature, offered by the
AppManager program, to automate the monitoring and management capabilities of a
server. For example, scripts can run platform tools that enable you to

      Enhance application performance and manage system workload.
      Create a ―snapshot‖ description of the current Windows system environment at a
       particular point so that the state can be reviewed or compared with a previous
       state.

Other scripts monitor

      Service request notifications and status information
      Critical services associated with the clustering capabilities of Server Sentinel

Additional scripts to solve specific customer problems can be supplied by Unisys
technical services personnel or by technical personnel at a customer facility.

Capabilities
The AppManager program provides

      Centralized event management

       Provides a single console to monitor hardware and operating system status and the
       availability of enterprise servers. From this central location, you can process
       events and notify key personnel of critical issues.
      Powerful scripting

       Enables customized data collection and automated responses to typical server
       management problems. Scriptable objects are provided to monitor system
       parameters and facilitate programmable reactions. These scriptable objects are
       deployed on one or more systems by a simple ―drag-and-drop‖ operation.
      Historical data analysis

       Maintains historical server and operating system data for capacity planning and
       trend analysis. Through the extensive graphic and reporting capabilities, you can
       identify and resolve system problems before outages occur.
      Automated response

       Provides automation through event-driven actions such as pager alerts and SNMP
       traps.
      A unified front end for launching other applications

       Provides an easy and standard way for launching other system management
       applications, such as Windows Server Manager.

Architecture
The AppManager program consists of the following major components, which you can
selectively install:

      Operator Console

       The Operator Console is a user interface that enables you to configure and control
       the execution of Knowledge Scripts on the systems and applications you manage.
      Chart Console

       The Chart Console is a user interface that enables you to generate and view charts
       of repository data.
      Distributed Event Console

       The Distributed Event Console is a user interface that enables you to view and
       respond to event information on one or more repositories.
      Agent

       The agent is an intelligent, autonomous Windows service that resides on managed
       clients to monitor the system and application resources, such as printers, disks,
       processors, and database files associated with those clients.
      Management server

       The management server is a process that communicates with agents on managed
       clients and updates the repository database with information received from the
       clients. This component runs on the same server as the repository.
      Repository

       The repository is a database that contains information about discovered
       configurations, Knowledge Scripts, events or alerts, jobs (instances of Knowledge
       Scripts), and graph data. The repository is implemented on a Microsoft SQL
       Server database. If a SQL Server is unavailable, Installation Assistant
       automatically installs the Microsoft Server Desktop Engine (MSDE).
      Web management server

       The Web management server is a set of Active Server Pages (ASP) that display
       reports based on information in the repository. These pages can be accessed from
       Server Sentinel.

       If you purchase an extra license for the Operator Web Console, the Web
       management server also provides the Web pages that make up the Operator Web
       Console.

For a graphical representation of these components, see AppManager Architecture.

Managing Tools
Server Sentinel makes use of the AppManager program to provide the following tools for
monitoring and managing an enterprise server:

      Knowledge Scripts
      Actions
      Reports

Knowledge Scripts

Server Sentinel provides data collection and automated responses to typical server
management problems through the use of scriptable objects. These scriptable objects are
deployed on one or more systems by a simple ―drag-and-drop‖ operation. However, some
scriptable objects are launched automatically during installation. An example is the
IntrusionDetection Knowledge Script, which monitors events during attempts to gain
unauthorized access into the Service Processor.

When a Knowledge Script is deployed on a target, one or more jobs might be created and
sent to the target computers being managed. If the Knowledge Script is deployed on a
single computer, one job is created and sent to that computer. If the Knowledge Script is
deployed at a higher point in the hierarchy, there might be multiple managed client
computers and thus multiple jobs might be created.
Parameters can be supplied when a script is deployed. The parameters include using a
particular threshold level, raising an event, or running a particular script if an error is
encountered. The following list is a small sample of the tasks controlled by the scripts
included with Server Sentinel:

      Monitor total processor usage and queue length to determine loads, count open
       files, or monitor physical and virtual memory utilization.
      Monitor the logical disk activity on one or more disks.
      Manipulate a single affinity group or job object by adding or deleting one or more
       rules or processors, or by assigning the priority or affinity mask.

Actions

Certain configurable actions can be invoked when a Knowledge Script encounters a
particular event. Once an event is raised by a script, an action is triggered. Examples of
these actions include

      Send mail to one or more e-mail users.
      Allow multiple actions to be started when an event is raised.
      Write an event to the Windows event log.
      Send a paging call to one or more recipients in response to an event.
      Shut down and restart a computer when an event is raised.
      Take a snapshot of the current system state.
      Create a package of diagnostic information.

Reports

The Report category of Knowledge Scripts enables you to generate HTML-formatted
reports from data collected from a variety of Knowledge Scripts.

AppManager Integration with Server Sentinel
The AppManager program provides a comprehensive software solution to monitor and
manage the performance, availability, and server health of an enterprise server. For an
overview of these capabilities, see What's the AppManager Program.

Server Sentinel offers ease of use by integrating the AppManager features into the overall
management environment. Server Sentinel also provides extensions that enhance the
ability of the AppManager program to manage specific features on Unisys systems.

The AppManager program is integrated into Server Sentinel operations through the
following features:

      Status and diagnostic information for display in Server Sentinel
      Expert mode operations through the AppManager Operator Console
      The Unisys managed object, which enables the AppManager program to discover
       the components that are unique to Unisys servers
      Unisys scripts, which monitor Unisys hardware and invoke other features of
       Server Sentinel, such as platform tools and Call Home.

Server Sentinel
Server Sentinel enables you to view alert notifications and Windows operational reports
generated by the AppManager program.

Alert Notifications

AppManager alerts notify you about conditions for possible problems or about the need
for operator intervention. You can access these alerts through

      The Alerts list on the left side of Server Sentinel.

      The navigation categories in the left pane of Server Sentinel, as follows:
          o Select Health Monitoring and then click System alerts to display the
              system alerts.
          o Select Security and then click Security alerts to display the security
              alerts.
          o Select Health Monitoring and then click Predictive Health to display a
              list of predictive health actions.


Windows Operational Reports
You can view reports created through the AppManager program by selecting Health
Reporting and then clicking Windows Server operational reports. For more
information, see "Generating and Viewing Windows Operational Reports."

AppManager Operator Console (Expert Mode)
You can use the Operator Console to access more advanced AppManager features, such
as

      Starting or stopping scripts that monitor the system for specific problem
       conditions
      Specifying parameters for a script
      Specifying self-healing actions to be performed automatically when a script raises
       an event

To launch the Operator Console from the Server Sentinel Web pages, select Self Healing
and then click AppManager Operator Console (Expert Mode).
Unisys Managed Object
The Unisys managed object extends the AppManager capabilities by providing built-in
knowledge of the CMP system architecture.
Installation Assistant automatically installs the Unisys managed object on any ES7000
component where the AppManager agent is present. Once the managed object is
installed, the Discovery_Unisys script detects components unique to Unisys, such as the
Service Processor, platform tools, and specific system services. These components are
displayed in the AppManager TreeView pane. You can then deploy scripts on these
components.
Installation Assistant automatically initiates the Discovery_Unisys script when you install
the AppManager agent on an ES7000 component. In some cases, you might need to
manually launch this script, as described in the Discovery_Unisys script.

Unisys Scripts
The AppManager program includes a full suite of scripts that specifically monitor various
major components of the system as well as industry-standard applications. Unisys has
extended this suite with a variety of scripts that provide access to specific Unisys
hardware and software features, such as platform tools and Call Home. For a description
of the Unisys scripts, see Knowledge Scripts Overview.

Monitoring Policy

To simplify your operations, Unisys installs a default monitoring policy that applies to all
Unisys servers, including such components as the Service Processor and partitions. This
monitoring policy automatically initiates certain standard scripts on all servers or
components where applicable. For more information, see Unisys Monitoring Policy.

Obtaining Additional AppManager Licenses
The AppManager components integrated into Server Sentinel are regulated by the same
licensing rules as Server Sentinel. That is, the AppManager license provided with Server
Sentinel enables you to manage all the elements of a single system that is delivered with
Server Sentinel. These system elements include partitions, the Service Processor, the PCI
Blade Appliance (formerly known as Slot Appliance), and virtual machine software, such
as VMWare.
When you purchase Server Sentinel, you get a license key that enables 30 AppManager
agents to be managed from the management server. The 30-agent limit applies even if
you purchase multiple ES7000 servers. In a few cases, a larger number of licenses might
be needed to manage one or more ES7000 systems. You can obtain an additional key at
no charge to provide additional licenses in increments of 30.
If you need to obtain additional AppManager licenses, contact your Unisys representative
or complete the following steps:
   1. Visit the Unisys Product Support Web site at www.support.unisys.com.

   2. Confirm your support agreement by providing your user name and password.

   3. Select ES7000 from the list of servers.

   4. Select Server Sentinel.

   5. Select Software.

   6. Locate AppManager by NetIQ in the supplied table, and follow the
      instructions in the Comments column to obtain additional licenses.


Security – AppManager
Granting User Access to the AppManager Program
This topic discusses the log-on information that is required for users to view resources,
start jobs, respond to events, and perform other tasks with the AppManager program.

Server Sentinel User Roles
The easiest way of managing access to the AppManager repository is to use the Server
Sentinel roles: Sentinel Observers and Sentinel Operators.

Installation Assistant automatically creates these groups and configures the AppManager
repository to grant the appropriate access to members of these groups. Members of the
Administrators group are implicitly granted complete access to all AppManager
functionality.

      Sentinel Observers

       Members of the Sentinel Observers group can only view the information in the
       Operator Console. This group corresponds to the AppManager Read-Only User
       role.
      Sentinel Operators

       Members of the Sentinel Operators group have full access to the Operator
       Console but are not allowed to configure security. This group corresponds to the
       AppManager Standard User role.

You can use the Windows Computer Management Administrative Tools utility to add
accounts to the Server Sentinel user groups on the repository server. Through this utility,
you can automatically provide groups with appropriate access to the repository without
any further configuration.
Security Requirements on Client Workstations (Power Users
Group)
To run the AppManager Operator Console from a client workstation, the following
requirements must be met:

       The user account of the Windows session must be a member of the local Power
        Users group.

       If the user account is not part of the Power Users group on the client workstation,
       attempts to log on to the Operator Console can fail, and the following message is
       displayed:

             Another console is logging with the same username or fails to create KS
             cache file.

       The user account of the local Windows session must be a member of one of the
        following groups on the management server: Administrators, Sentinel Operators,
        or Sentinel Observers.

Handling Windows Workgroup Security
If you use Windows workgroup security, create an account on the repository server for
each user with the same user name and password that the user employs to log on to his or
her workstation. In addition, you must add these accounts to the appropriate Sentinel
group: Sentinel Observers or Sentinel Operators.

Handling Windows Domain Security
When a new domain account is created that requires access to the AppManager program,
add that account to the Administrators, Sentinel Operators, or Sentinel Observers group
on the repository server.

Handling SQL and MSDE Security
Server Sentinel requires SQL software to support SQL Server and Windows
authentication (mixed mode). Installation Assistant installs the Microsoft SQL Server
2000 Desktop Engine (MSDE) with SQL security enabled. MSDE is installed with a
default password of SentinelPasswordForSA.

If you plan to use an existing SQL Server instance, use the SQL Enterprise Manager to
check the server properties. Ensure that the server is configured for SQL Server and
Windows authentication.
Mixed mode enables you to use Windows security credentials or to use SQL log-on
names when accessing the AppManager Operator Console.

If you installed a new MSDE version through Installation Assistant, then Installation
Assistant assigns the default password SentinelPasswordForSA. For security purposes, it
is highly recommended that you change this password by performing the following two
procedures.

If you are using an existing MSDE version and the AppManager user log-on name in
Server Sentinel is sa, then perform the procedure ―Updating the sa Password in Server
Sentinel.‖

Changing the MSDE Password
To ensure secure access to the AppManager repository, change the system administrator
(sa) password after installing the AppManager program by completing the following
steps:

   1. On the computer hosting the MSDE instance to which you are connecting, open a
      command prompt window.
   2. Type the following command:
   3.              osql -E


        This command connects you to the local default instance of MSDE by using the
        user account of your Windows session.

   4. Type the following commands. Replace 'mynewpassword' with your new
      password.
   5.              exec sp_password null, 'mynewpassword', 'sa'
   6.              go


        When the password is changed, the following message is displayed:

         Password changed



Updating the sa Password in Server Sentinel
To enable communication between Server Sentinel and the AppManager repository, you
must supply a SQL log-on name. Typically, the sa log-on name is used for this purpose.
However, you can use any SQL log-on name that has been configured with the NetIQ®
Security Manager to use the AppManager Standard Operator role.

To configure the SQL log-on name for Server Sentinel, complete the following steps:
   1. From Server Sentinel, select the Configuration Summary icon, shown below,
      and then click Server Sentinel Settings on the Configure menu.




       The Server Sentinel settings page appears.

   2. Update the AppManager User Logon Name and AppManager User Logon
      Password boxes to reflect the new name and password.

       The AppManager user log-on password must be the same as the MSDE sa
       password.

   3. Click Update.

Knowledge Scripts Reference
Knowledge Scripts Overview
This User's Guide provides information about the scripts developed by Unisys as an
additional value to the AppManager® program.

Server Sentinel provides data collection and automated responses to typical server
management problems through the use of scriptable objects. These scriptable objects are
deployed on one or more systems by a simple ―drag-and-drop‖ operation. However, some
of these scriptable objects are launched automatically during installation. An example of
this is the IntrusionDetection Knowledge Script, which monitors events on attempts to
gain unauthorized access into the service processor.

When a Knowledge Script is deployed on a target, one or more jobs might be created and
sent to the target computers being managed. If the Knowledge Script is deployed on a
single computer, one job is created and sent to that computer. If the Knowledge Script is
deployed at a higher point in the hierarchy, there might be multiple managed client
computers and thus multiple jobs might be created.


   Caution
If you have previously installed the AppManager program and have changed the default
parameter values for any of the Knowledge Scripts developed by Unisys, reinstalling the
AppManager program will cause those parameters to be restored to the original default
value.

Knowledge Script Categories
Knowledge Scripts are grouped into various categories. For example, Knowledge Scripts
in the Action category are used to perform corrective or responsive actions when events
are detected. Knowledge Scripts in the Discovery category are used to find or discover
configuration or reporting information about your managed system.

Click any of the following links for more information about the Knowledge Scripts
developed by Unisys. (No help is available for these Knowledge Scripts from within the
AppManager Operator console.)

      Action Knowledge Scripts
      Discovery Knowledge Scripts
      Unisys Knowledge Scripts

Other Knowledge Scripts

For information about the many other Knowledge Scripts available with Server Sentinel

      Refer to the AppManager Knowledge Script Reference Guide included on the
       Server Sentinel AppManager Supplement Software CD-ROM.
      From within the AppManager Operator Console, select a Knowledge Script in the
       Knowledge Script pane and press F1.




Unisys Monitoring Policy
When the AppManager program is installed, the installation process creates a monitoring
policy that applies to all Unisys servers. This policy launches the scripts from the
Unisys_ServerHealth Knowledge Script group on all Unisys servers.

The Unisys_ServerHealth group includes the following scripts:

             Unisys_CertificationCheck
             Unisys_IntrusionDetection
             Unisys_LogBackup
             Unisys_PredictiveAlerts
             Unisys_ServerAlerts

For information about these scripts, refer to Unisys Knowledge Scripts Overview.

Because these scripts are in a group that is part of a monitoring policy, you cannot start or
stop the scripts or modify their parameters in the usual way. Instead, you can

      Modify the contents of the Unisys_ServerHealth group as described in the
       AppManager User Guide under "Changing a Knowledge Script Group" and
       "Propagating Knowledge Script Group Properties to Running Jobs." The User
       Guide for the AppManager software is included on the Server Sentinel
       AppManager Supplement Software CD-ROM.
      Modify the Unisys monitoring policy as described in the AppManager User
       Guide under "Changing a Monitoring Policy."

Action Knowledge Scripts Group
Action Knowledge Scripts Overview
The Action category provides Knowledge Scripts that perform corrective or responsive
actions when events are detected.

The following table summarizes the Action Knowledge Scripts. Click any of the
Knowledge Script names for complete details.

For information about other Action Knowledge Scripts

      Refer to the AppManager Knowledge Script Reference Guide included on the
       Server Sentinel AppManager Supplement Software CD-ROM.
      From within the AppManager Operator Console, select a Knowledge Script in the
       Knowledge Script pane and press F1.

Knowledge Script            What it does
                            Creates an Analysis Manager package and optionally creates
CreatePackage
                            a service request packet when an event is raised.
DeleteFilesOrFolders        Deletes one or more files or folders.
EZupdateMail                Generates an e-mail notification in response to an event
                            raised by the Unisys_EZupdate script.
Snapshot                    Takes a snapshot of the local configuration in response to an
                            event.


CreatePackage
Note: This script is unavailable on partitions or Service Processors running Server
Sentinel 4.0 or higher. However, this script is available for managing systems running
earlier versions of Server Sentinel.

Use this Action Knowledge Script to create an Analysis Manager package, and optionally
create a service request packet, when an event is raised. For information about Analysis
Manager and packages, refer to the Platform Tools Help.
Remember to set the Managed Client Action option whenever you specify
Action_CreatePackage in the Actions property of a script. The Managed Client Action
option causes the action script to run on the managed client rather than the management
server.

Parameter values
Set the following parameters as needed.

Description                     How to set it
Choose a package type:          Choose one of the available package types from the
                                drop-down list:

                                         Application Failure. Use this type for an
                                          application that terminated abnormally.
                                         Application Not Responding. Use this type for an
                                          application that has not terminated, but is not
                                          responding to input.
                                         Base Package. Use this type to create a package
                                          with information about all the processes that are
                                          currently running on the system.
                                         Service Not Responding. Use this type for a
                                          service that has not terminated, but is not
                                          responding to input.
                                         System Failure. Use this type after rebooting
                                          from a system failure. The package created will
                                          include crash dump data, if a crash dump was
                                          taken.

Headline (Required)             A short (up to 100 characters) description of the
                                problem.
Detailed description            A more detailed description of up to 2000 characters.
(Optional)
Name of failed application      Enter a complete path name, such as C:\Program
or service (Optional)           Files\Internet Explorer\Iexpore.exe. This parameter is
                                required if the package type is Application Failure,
                                Application Not Responding, or Service Not
                                Responding.
Additional files to package,    Enter complete path names, separated by commas. Do
separated by commas             not enter spaces before or after the commas. Enclose a
(Optional)                      path name that includes spaces or commas in angle (<>)
                                brackets. For example: c:\daily.bat,
                                <c:\Program Files\MediaPlx\mdp.exe>
Send Call Home packet?          Specifying y causes service request data to be included in
(y/n)                            the package, and causes a service request packet to be
                                 created. If Call Home is not available, a warning event is
                                 given, but the package is still created.




DeleteFilesOrFolders
Use this Action Knowledge Script to delete one or more files or folders (directories).
Folders can contain zero or more files.

Default schedule
None. This Knowledge Script runs when Action_DeleteFilesOrFolders is selected in the
Actions tab of the Properties of a Knowledge Script.

Parameter values
Set the following parameters as needed.

Description                      How to set it
Enter ―File‖ or ―Folder‖ and     Enter the type and name of each item to be deleted. Enter
the name of the file or folder   the fully qualified path and name for each item to delete.
to be deleted (repeat as         For example:
necessary, separated by
commas)                                 File C:\*.tmp, File C:\*.log, Folder
                                        C:\temp.
Delete read-only files? (y/n)    Enter y to delete a file even if it is marked as read-only.
                                 The default is n.
Delete hidden files? (y/n)       Enter y to delete a file even if it is marked as hidden. The
                                 default is n.
Delete system files? (y/n)       Enter y to delete a file even if it is marked as a system
                                 file. The default is n.
Minimum age of file or           Enter the minimum age for the items to delete. Files and
folder to be removed             folders created more recently than the minimum are not
                                 deleted. The default is 0 days old, meaning that no age
                                 restriction exists and all specified files are to be deleted.

Handling Read-Only, Hidden, and System Files
If the parameters you set for this script lead to an ambiguous situation, the script errs on
favor of not deleting some files that you expected to be deleted rather than deleting files
you expected to be retained.
Examples

      If the script is run to delete the directory specification .Folder C:\TEMP. with
       Delete hidden files set to n and hidden files are in the directory, the script does not
       remove the directory or any files in the directory.
      If the script is run to delete the file specification .File C:\TEMP\ *.*. with Delete
       hidden files set to n, all files in C:\TEMP except for those marked as hidden are
       removed and the folder is retained.

EZupdateMail
Note: This script deploys only on client systems that are running the EZupdate software
with Server Sentinel 2.3 or 2.4. On systems running a later version of Server Sentinel, the
notification of software updates is provided without the use of the AppManager program.

Use this Knowledge Script to generate an e-mail notification in response to an event
raised by the Unisys_EZupdate Knowledge Script. The e-mail message is sent in HTML
format.

This script uses a Simple Mail Transfer Protocol (SMTP) server. Because the script can
specify a remote SMTP server, there is no need for the SMTP service to run locally.

The script does not provide user authentication, so it cannot be used with SMTP servers
that require authentication.

Default schedule
None. This Knowledge Script runs automatically in response to a Unisys_EZupdate script
event.

Parameter values
You can leave most parameters unchanged because by default the parameters use the
values configured during the installation of the EZupdate software and the AppManager
extensions. However, you can modify one of the following parameters to specify the
handling of software or firmware updates on a management server:

      To receive e-mail notifications of updates for the management server, specify
       recipients in the List of recipients parameter, and allow the Disable script on
       management server parameter to default to n.
      To avoid creating AppManager events, reports, and e-mail notifications for the
       management server, set the Disable script on management server parameter to
       y. Then the List of recipients parameter is not used on the management server

Set the following parameters as needed.
Description                    How to set it
List of recipients             Enter a list of recipients (for example,
                               JohnDoe@aol.com).

                               Use a comma to separate multiple addresses, and do not
                               include any spaces.

                               This parameter affects only those systems for which you
                               have not defined a list of recipients for the EZupdate
                               component. For example, the management server never
                               includes a predefined recipients list. By contrast,
                               partitions typically have a predefined recipients list
                               unless you did not provide one during the installation of
                               the EZupdate component.

                               If you leave this parameter blank, and the system does
                               not have a predefined recipients list, then no e-mail
                               notification is sent.
Disable script on              Enter y to prevent the Unisys_EZupdate script from
management server? (y/n)       creating EZupdate events and e-mail messages for the
                               management server.

                               The default is n.
SMTP server name               Enter the SMTP server name.

                               If you leave this parameter blank, the SMTP server
                               specified during the installation of the AppManager
                               extensions is used.
SMTP port                      Enter the SMTP port number.

                               If you leave this parameter blank, the port number
                               specified during installation of the AppManager
                               extensions is used.

Changing Action_EZupdateMail Parameters for the
Monitoring Policy
The Action_EZupdateMail script is designed for use by the Unisys_EZupdate script,
which runs as part of the Unisys monitoring policy.

To modify Action_EZupdateMail parameters, complete the following steps from the
AppManager Operator Console:

    1. In the Knowledge Script pane, select the Action tab.
   2. Double-click EZupdateMail.

      The Properties for the Action_EZupdateMail dialog box appears.
   3. Make changes as needed and then click OK.
   4. In the Knowledge Script pane, select the Unisys tab.
   5. Right-click the EZupdate Knowledge Script, point to Properties Propagation,
      and then click Knowledge Script Group Members.

      The Properties Propagation dialog box appears.
   6. Ensure that the Script and Properties options are enabled and then click OK.

All the scripts in the monitoring policy stop and then restart within 2 minutes
approximately. After restarting, the latest parameter values are propagated to the
EZupdateMail action of all running instances of the Unisys_EZupdate Knowledge Script.

Snapshot
Use this Action Knowledge Script to take a snapshot of the local configuration in
response to an event.

Default schedule
None. This Knowledge Script runs when Action_Snapshot is selected in the Actions tab
of the Properties of a Knowledge Script.

Parameter values
Set the following parameters as needed.

Description                     How to set it
Snapshot directory              Required. Enter the drive letter and directory where the
(drive:\directory)              snapshot file is to be saved. The default value is
                                C:\Snapshots.
Snapshot file name without      Required. Enter the file name under which the snapshot
.snap extension                 is to be saved. Do not include a file extension. The
                                default name is New.
Overwrite existing Snapshot     Enter y to replace an existing snapshot file with the
with same name? (y/n)           current system configuration data. The default is n.
Include processes? (y/n)        Enter y to include process data in the snapshot. The
                                default is n.
Include hardware                Enter y to include hardware configuration information in
configuration? (y/n)            the snapshot. The default is y.
Include files? (y/n)            Enter y to include file data in the snapshot. The default is
                                 y.
Use existing File Map (file      Enter the name of a previously saved File Map to use.
name without .snap               Do not include a file extension. By default, the script
extension)                       looks for this file in the Snapshot directory that you
                                 specified. Do not enter a value in this field if you are
                                 specifying a file path and mask.
Include files from this path?    Limit the file data to be included in the snapshot to files
                                 from this path only. The default path is
                                 C:\WINNT\SYSTEM32.
In addition to .exe. and .dll    Executable files (.exe and .dll extensions) from the
files, include files with this   specified directory are always included in the snapshot.
mask                             Use this parameter to include files of an additional type.
(*.* for all)                    The default mask is ―‖ to indicate that no additional files
                                 are to be included.
*Include registry? (y/n)         Enter y to include registry entries in the snapshot. The
                                 default is y.
Use existing Registry Map        Enter the name of a previously saved Registry Map to
(file name without .snap         use. Do not include a file extension. By default, the script
extension)                       looks for this file in the Snapshot directory specified
                                 previously. Do not enter a value in this field if you are
                                 specifying a registry path and mask.
Registry path to snapshot        Limit the registry data to be included in the snapshot
                                 only to entries from this Registry Map. The default is
                                 HKEY_LOCAL_MACHINE\Hardware\description
Limit to registry entries        Limit the type of registry data to be included in the
matching this key (―*‖           snapshot to entries matching the specified key. The
matches all entries)             default key is ―*‖ to include all entries in the specified
                                 registry map.
Include certification? (y/n)     Include drivers information related to Windows
                                 Datacenter certification in the snapshot. The default is y.
Event when Snapshot              Enter y to generate an event when the Knowledge Script
completed? (y/n)                 completes. The default is n.
Event level when error           Enter the severity level for the event generated when the
during Snapshot                  snapshot cannot be created. The default level is 5 (red
                                 event indicator).




UnisysAlerts
The UnisysAlerts action Knowledge Script is the default action for the ServerEvents
script and is responsible for performing any AppManager actions associated with a
HealthEvent object.
Each event can take zero or more actions, as determined by the health monitoring policy
for the system. Some actions, such as stopping or starting a service, are performed
directly by the Health Monitor service itself. Other actions are performed in the
AppManager environment by the UnisysAlerts script. These action values include

        Email
        Page
        Net Send
        DOS Command

Because there are no user-provided monitoring policies in the initial release of Predictive
Health, no events exist that require customization. In future releases, you will be able to
customize actions such as e-mail and paging.

Discovery Knowledge Scripts Group
Discovery Knowledge Scripts Overview
The Knowledge Scripts in the Discovery category provide information about your
Microsoft Windows operating system and application configuration. Each Discovery
Knowledge Script has a specialized task, such as finding configuration details about the
Windows operating system installed on the computer where you run the script.

The following table summarizes the Unisys Discovery Knowledge Script. Click the
Knowledge Script name for complete details.

For information about other Discovery Knowledge Scripts,

        Refer to the AppManager Knowledge Script Reference Guide.
        From within the AppManager Operator Console, select a Knowledge Script in the
         Knowledge Script pane and press F1.

Knowledge Script             What it does
                             Discovers configuration and resource information for Unisys
Unisys
                             servers.




Unisys
Use this Knowledge Script to discover configuration and resource information for Unisys
servers.
The Discovery_Unisys script runs automatically when you install the AppManager agent
on a server. As a result, it should rarely be necessary for you to deploy this script, except
in the in the following circumstances:

      If you notice that the Unisys Server object is not present on a Unisys partition or
       service processor.
      If you add a computer to the repository using the Add Computer feature of the
       Operator Console.

       The Add Computer feature automatically runs the Discovery_NT script to
       discover standard Windows objects. When Discovery_NT has completed, run the
       Discovery_Unisys script against the newly added service processor or partition.

Resource objects
Unisys servers.

Default schedule
By default, this script is only run once for each computer.

Setting parameter values
Set the following parameters as needed.

Description                      How to set it
Event for successful             Set to y to generate an event indicating the discovery
discovery                        completed successfully. The default is n.
SNMP community string            Enter the SNMP community name to be used. The
                                 default value is the community name entered in the
                                 AppManager Security Manager.

                                 If no community name has been entered in AppManager
                                 Security Manager, then public is the default value.
Event severity levels            Set the event notification level to give you the desired
                                 visibility for a successful, failed, or partially completed
                                 discovery. By default, the severity levels are

                                         Discovery was successful: severity 25 (blue event
                                          indicator)
                                         Discovery failed: severity 5 (red event indicator)
                                         Discovery was partially done: severity 10 (red
                                          event indicator)
                                         Windows NT server not found on the machine:
                                       severity 15 (yellow event indicator)




Unisys Knowledge Scripts Group
Unisys Knowledge Scripts Overview
The Unisys Knowledge Script category provides Knowledge Scripts for monitoring
specific components, such as fans and network cards. Some of these scripts also enable
you to use platform tools, such as Contrast Manager.

The following table summarizes the Unisys Knowledge Scripts. Click any of the
Knowledge Script names for complete details.

For information about other Knowledge Scripts,

      Refer to the AppManager Knowledge Script Reference Guide.
      From within the AppManager Operator Console, select a Knowledge Script in the
       Knowledge Script pane and press F1.

For information about checking in custom Knowledge Scripts, see Checking in Custom
Knowledge Scripts

Knowledge Script              What It Does
CallHomeAlert                 Monitors Call Home services for abnormal conditions. The
                              script can raise an alert when a certain number of service
                              request packets cannot be sent to the Unisys Support
                              Center due to a connection or transmission failure when
                              errors are encountered sending packets, when packets are
                              refused by the support center, or when a certain total
                              number of packets are generated since the script last ran.
CallHomeHealth                Monitors the status of Call Home services.
CallHomeInfo                  Reports Call Home statistics for a number of days or over
                              a specified time period.
CallHomeUCFInfo               Collects the information necessary to write a UCF against
                              Call Home.
CertificationCheck            Verifies that a Windows Datacenter Server remains
                              compliant with a certified configuration.
ClusterHealthMonitor          Monitors the status of the Server Sentinel Cluster Health
                              Monitor service.
ClusterOptimizer              Monitors the status of the Availability Optimizer for
                           Microsoft Cluster Service.
Compare2Systems            Compares the local configuration with a snapshot from a
                           remote system.
ConfigMonitor              Detects and reports changes since a designated snapshot.
ConfigSNMP                 Configures the Windows SNMP Service.
CorrectableMem             Monitors the condition of correctable memory and the
                           number of new correctable memory errors.
DetectDump                 Monitors the system for new dump files and creates
                           Analysis Manager packages for them.
DiskArrayFail              Monitors the status of physical drives in an array set.
DoAffinityGroup            Manipulates a single affinity group or job object by adding
                           or deleting one or more rules or processors, assigning the
                           priority or affinity mask, or setting the time limit of the job
                           or of given processes within the job.
DynamicRules               Monitors CPU utilization of affinity groups and job
                           objects.
EZupdate                   Checks for the availability of updates for Server Sentinel
                           and platform firmware releases and then creates a report
                           about these updates.
FanSummary                 Monitors the status of system and processor fans.
GetFileOrRegMap            Creates a file map and/or a registry key map for a Unisys
                           server.
GroupMonitor               Monitors the status of all or selected affinity groups, job
                           objects, and rules, and raises an event when a change is
                           detected.
GroupUsage                 Tracks the processor usage of all or selected affinity
                           groups or job objects, and, optionally, dynamically
                           allocates additional processors to maintain a specified
                           threshold.
IntrusionDetection         Monitors OKENA StormWatch or Cisco Security Agent
                           log files for events that indicate attempts to gain
                           unauthorized access into the Service Processor.
LogBackup                  Backs up selected event logs.
MonitorBottlenecks         Monitors a system for possible bottleneck conditions as
                           reported by Perfmon counters.
MonitorFileAndFolderSize   Monitors the size of one or more files or directories.
MonitorSNMPAgents          Verifies that platform SNMP agents are installed and
                           running.
NICError                   Monitors network interface transmission errors.
NICFail                    Checks whether the network interface subsystem is down.
PowerSupplySummary         Monitors the status of system power supplies.
PredictiveAlerts              Runs in the background and queries the HealthMonitor
                              service for predictive server events that have occurred
                              since the last polling cycle.
RestartSystem                 Restarts the system with an optional crashdump.
ScheduleTask                  Schedules a task for execution at a specified time.
ServerAlerts                  Runs in the background and queries the HealthMonitor
                              service for server events that have occurred since the last
                              polling cycle.
Snapshot                      Creates and saves a snapshot containing configuration
                              information for a Unisys server.
SPHealth                      Monitors external Service Processor status, and optionally,
                              restarts any critical services that are not running.
TemperatureSummary            Monitors the status of system and processor temperatures.
TSLogoff                      Generates an event that contains the number of active
                              Terminal Services sessions on a system and optionally logs
                              off the sessions.
UndoAffinityGroup             Deletes all or selected affinity groups or job objects.
UPSBatteryLow                 Monitors the UPS battery life.
VoltageSummary                Monitors the status of system and processor voltages.
WebSiteMonitor                Monitors whether a particular Web site is available.




Launching External Applications
From the AppManager Operator Console, you can launch the Server Sentinel navigation
interface or the Console Manager.

Server Sentinel Navigation Interface
Server Sentinel is a centralized operations environment for your server. It provides
world-class data center server management capabilities for greater business continuance
and system resiliency within your server environment. It can dramatically enhance the
manageability, availability, and performance of your server environment while
considerably reducing total cost of operations.

Server Sentinel simplifies system management, automates corrective and preventive
measures to avoid a system failure, and increases the reliability, scalability, and
performance of Unisys servers.
To launch this application

   1. From the TreeView pane, expand a machine icon corresponding to a Service
      Processor.
   2. Click Server Sentinel Home on the Extensions menu.

Console Manager
Server Sentinel provides a method for you to communicate with and control Intel
partitions from one or more remote workstations. You can redirect the functions of the
partition peripherals—including the partition keyboard, video, and mouse—to a remote
workstation.

To launch this application

   1. From the TreeView pane, expand a machine icon corresponding to a server or
      partition, and select the Console Manager icon.
   2. Click Console Manager on the Extensions menu.


Security
Purpose

Protection from external intrusions, deliberate or otherwise, is critical for enterprise
applications. Server Sentinel provides a number of security features to minimize service
interruptions from such intrusions.

User Benefit

 Increased server availability and lower administration costs attributable to corrective
actions required from vulnerabilities, network intrusions, and human error.

Description

Server Sentinel’s security features include intrusion prevention, vulnerability analysis,
network security and user access; all of which can be customized to meet your specific
server security requirements.

The vulnerability analysis feature detects security vulnerabilities in the Service
Processor, Server Sentinel Management Server, and server partitions, which can consist
of missing Microsoft patches and systems not configured for optimum security.
Server Sentinel also provides access to the Microsoft Baseline Security Analyzer which
checks computers for common security misconfigurations.
And Server Sentinel’s role-based security limits access to tasks through the use of role
based security. The user roles of Administrator, Operator, and Observer are implemented
using Microsoft groups, and are established during Server Sentinel installation.

Vulnerability Analysis

The vulnerability analysis feature detects security vulnerabilities in the Service Processor,
Server Sentinel Management Server, and server partitions. Vulnerabilities consist of
missing Microsoft patches and systems that are not configured for optimum security.
Server Sentinel produces a report that lists the vulnerabilities of the target systems, and
provides recommendations on how to improve the security of the systems.

Note that this feature is enabled for a limited period of time. You have the opportunity to
extend the time period, purchase a vulnerability analysis service, or extend the analysis
capability to other systems in the enterprise.

Network Security

See the topic Understanding Security in an Enterprise Server Environment for
network Security information.

Baseline Security Analysis

Server Sentinel provides access to the Microsoft Baseline Security Analyzer. This utility
checks computers running Microsoft Windows NT 4.0, Windows 2000, or Windows XP
for common security misconfigurations. If the Microsoft Baseline Security Analyzer is
not installed on the workstation you are using, the Server Sentinel interface redirects you
to a Web site where you can download this utility program.

Role-Based Security

Server Sentinel limits access to tasks through the use of role-based security. The
user roles of Administrator, Operator, and Observer are implemented using Microsoft
groups and are established during Server Sentinel installation.


Understanding User Roles
The user roles of administrator, operator, and observer are implemented using Microsoft
groups. User roles are established during Server Sentinel installation.

Here is a general definition of each role:

      Administrator
       Can administer the management environment (for example, configure the
       management software or delete items from a log) as well as manage systems.
       Functions that require administrative privileges to execute are included in this role.

       Operator

       Can manage systems. Operator has read/write access to most management
       functions. Functions that require administrative privileges are excluded from this
       role.

       Observer

       Can observe systems. Observer has read access to most management functions.

If you want to change user roles, close Server Sentinel and log on to Server Sentinel with
a user name and password from a different user group. For example, if you are an
observer and you want to perform an operator function, close Server Sentinel then log
back on with a user name and password from the operator group.

For more information, see the installation documentation appropriate for your system.

Partition Software Baseline Security Analysis
This task opens the Microsoft Baseline Security Analyzer (MBSA).

MBSA checks computers running Microsoft Windows NT 4.0, Windows 2000, or
Windows XP for common security misconfigurations. You must have administrator
privileges for each computer you want to scan.

If the MBSA is not installed on the workstation you are using, you will be redirected to a
Web site where you can download MBSA. Download and install the MBSA to continue
the task.

When MBSA is run, it recommends the use of Microsoft's IIS Lockdown Tool, which
disables several optional features of IIS to secure your IIS server against attack. If the IIS
Lockdown Tool is run on the Service Processor and the "Other" server template is
selected, the following options can be selected or deselected. Unisys has tested Server
Sentinel with the IIS Lockdown Tool and recommends the following settings to ensure
that Server Sentinel will continue to work properly. See Microsoft Knowledge Base
Article 817807 for more information on the IIS Lockdown and URLScan tools.

INTERNET SERVICES

Web service (HTTP) - enable

SCRIPT MAPS
Active Server Pages - do NOT disable

Index Server Web interface - do NOT disable

Server side includes - disable

Internet Data Connector - disable

.HTR scripting - disable

Internet printing - disable

ADDITIONAL SECURITY

Remove the selected virtual directories: - okay to remove these

      IIS Samples
      MSADC
      IIShelp
      Scripts
      IISAdmin

Set file permissions to prevent anonymous IIS users from performing these actions:

      Running system utilities
      Writing to content directories

Disable Web Distributed Authoring and Versioning (WebDAV). - do NOT disable

URLSCAN

Install URLScan filter on the server - Check the box to install

Understanding Security in an Enterprise Server
Environment
Planning ahead helps ensure that your network enables you to take advantage of the
benefits that your enterprise server configuration and Server Sentinel software provide.

System Network Traffic
Your system requires the following types of information flow across your network
environment:
      Server Sentinel operations traffic. Server Sentinel is the centralized operations
       environment for your enterprise server. Communication between the various
       Server Sentinel components of your system depends upon your LAN topology.
      Server Sentinel support traffic. The Server Sentinel support traffic, which includes
       traffic from the Call Home and Call Home Health Advisor components,
       communicates through the Internet with the Unisys Support Center.
      Enterprise management solution traffic. If you are using an enterprise
       management framework server, communication between the enterprise
       management solution and your system depends upon your LAN topology.

LAN Configurations
Your LAN topology will determine how you integrate the system into your environment.
You can integrate the system in a typical LAN environment where all traffic passes
through your public LAN. An operations LAN, where operations and enterprise
management solution traffic is isolated from your other network traffic, adds additional
security. Because LAN topology is so diverse, there can be other considerations that
apply to your configuration.

LAN Ports
Each system can contain one or more cells. There are up to five LAN ports accessible on
the rear panel of each cell.

      One maintenance LAN port – There is an RJ45 port on the left rear of each cell
       which provides a 10/100 Mb Ethernet network connection to the maintenance
       LAN. A system can have only one active maintenance LAN port; however,
       multiple maintenance LAN ports can be connected for redundancy.
      Two server LAN ports – There are two RJ45 ports on the right rear of each cell
       which provide 10/100/1000 Mb Ethernet network connections for the operating
       system to use. These are normally connected to your public LAN.
      The first cell (cell 0) contains a Control Module (CM), which has two LAN ports.
       If there are two or more cells, there can be an additional Control Module in
       another cell, and in this case the CM LAN ports are used as a secondary CM-to-
       CM communications path.



Typical LAN Configuration
Advantages

      This is a simple LAN topology.
      Additional network components are not required.
Disadvantages

      In this configuration, operations and enterprise management solution traffic pass
       through your public LAN.
      If you have multiple ES7000 enterprise servers, additional traffic loads are placed
       on your public LAN.
      Setting up and maintaining strong network security can be a challenge.

Operations LAN Configuration

The operations LAN is an isolated network which connects the maintenance LAN to
management components (such as workstations or management servers) associated with
one or more enterprise servers.

Operations LAN Advantages

      The operations LAN isolates operations and enterprise management solution
       traffic from your public LAN either by using a firewall or by being disconnected.
      This configuration helps overcome the challenges of setting up and maintaining a
       secure network. It provides strong network security for the maintenance LAN
       while offering simple, straightforward access for authorized users.
      This configuration provides remote serviceability without compromising system
       security.
      If you have multiple ES7000 enterprise servers, you can manage them with a
       single operations LAN.
      If you are using an enterprise management system, such as HP OpenView or
       Tivoli NetView, these components can be members of your operations LAN.

Operations LAN Disadvantages

      Additional network components are required.
      Additional network administration is required.

Selecting a Method for Communication with the Support
Center
The system must be able to communicate support information to the Unisys Support
Center. The system can send information through your corporate Internet connection (the
preferred method) or a USB modem.

To obtain optimum performance, an Internet connection is required. Using an Internet
connection simplifies communication configuration and management and ensures that all
of the Server Sentinel components function properly. This method is likely to fit better
than a modem into your corporate security policy, and when your corporate firewalls are
configured correctly, it introduces minimal additional risk. However, if your security
policy does not permit your system to communicate through your corporate Internet
connection, you can use a USB modem.

To use the USB modem, a dedicated telephone line must be installed and maintained in
close proximity to the system and you must configure your system to use it. The Server
Sentinel Call Home and Call Home Health Advisor functionality will operate properly,
but the EZupdate feature is not supported over a USB modem.

Microsoft Active Directory
In a server environment that uses Microsoft Active Directory, you can provide
authorization for users based on their user accounts. When users attempt to access the
operations LAN, they are required to provide a valid user name and password if the user
account that the workstation is currently running under is not sufficient. If they cannot,
access is denied. Using Active Directory enables you to easily add or remove user
authorizations based on information in the directory service and to share security groups
among different servers.

If Active Directory is not available in your server environment—or if you want to further
restrict traffic within an environment that has Active Directory— you can authorize
traffic based on allowable IP addresses or subnets. This provides the benefit of being able
to restrict access to Console Manager from the public LAN; only users attempting to
access Console Manager from selected IP addresses or subnets would be permitted to do
so.

For detailed information about Active Directory, see the Microsoft corporate Web site or
your Windows server documentation.

Microsoft Terminal Services
For users at the Service Processor, Console Manager is the primary user interface to each
partition. However, you might consider using Microsoft's Terminal Services as an
alternate user interface solution for accessing a Windows environment. You can use
Terminal Services to remotely execute applications on a Windows-based server from a
wide range of devices over virtually any type of network connection (such as from a
workstation on your public LAN). With Terminal Services, you can use all display
settings available on the client workstation.

For detailed information about Terminal Services, see the Microsoft corporate Web site
or your Windows server documentation.


Call Home
Purpose
Call Home monitors and collects system and application errors according to a predefined
set of conditions.

User Benefit

Call Home is another feature contributing to the increased reliability of the ES7000 with
no reliance on human interaction.

Description

The Call Home functionality enables you to monitor the generation and delivery of
service request packets to the support center. These packets contain information on
system, controller, or application errors that are determined according to a predefined
set of conditions.

Internet access is the preferred method for Server Sentinel Call Home and Health
Advisor to communicate with the Unisys Support Center. Using this communication path
to the support center is beneficial because a modem and additional phone line are not
required.

Note: If configuring Internet access conflicts with your corporate security policy, you
can use a USB modem. The Server Sentinel Call Home and Health Advisor functionality
will operate properly, but the EZupdate feature is not supported over a USB modem.

Health Advisor

Health Advisor is a part of the Call Home software that gathers stability and
configuration data from a system and transports that information to Unisys on a regular
basis. This information is used within Unisys to generate stability reports on individual
systems, specific partitions, or all systems in a network. Although Health Advisor is
designed to automatically run without customer intervention, customers can use the
Server Sentinel Call Home Health Advisor Web pages to change certain settings or to
configure other systems. The Health Advisor Web pages enable you to perform such
tasks as:

•   View information about partitions and Service Processors on your system
•   Send stability data to the Unisys Support Center
•   Set up a schedule for sending information to the Unisys Support Center

Call Home is software that provides proactive, automatic detection and reporting of faults
and other server events to the Unisys support center, without any need for human
intervention. These faults or errors are included in a service request data packet which is
created automatically and sent electronically to the Unisys support center by the Health
Advisor utility for diagnoses, action and resolution.
These service events are also reported to the Sentinel operator console and to the user’s
support staff via email or pager if desired, all of which assist in rapid problem
resolution…and increased system availability.

The Call Home application consists of three major modules:

      Generator Service
      Central Service
      Health Advisor

Generator Service and Central Service Interaction
The Generator Service and the Central Service interact as follows:

   1. The Generator Service monitors information about system or application errors in
      the Windows event log.
   2. If the errors meet a specified set of conditions, the Generator Service extracts
      information from the Windows event log and creates a data packet.
   3. The Generator Service sends packets of raw data describing the errors to a
      Formatter program for data formatting. The Formatter program then sends the
      packets to the Central Service.
   4. The Central Service records the packet information in the Central database and
      delivers the packets to the support center according to a user-defined schedule.
   5. The support center acknowledges receipt of the packet information, and if the
      packet is valid, it generates a customer contact number. This contact number is
      displayed in the Packet Status page.

Call Home Heartbeat Monitoring
The Central Service periodically checks the Generator Service status by generating a
heartbeat request entry in the Windows event log of the computer on which the Generator
Service is installed. The Generator Service processes this event into a heartbeat packet,
which it then sends to the Central Service.

If the Central Service does not receive a reply to the heartbeat request within an hour of
generating the request, a service request is generated and sent to the support center.
Because the Generator Service processes event logs based on a set of specified error
conditions, the event log entry for the heartbeat request has an event type of Warning.
This is normal and does not imply that Call Home is not functioning correctly.

Typical Call Home Event Scenarios
This topic provides detailed scenarios for typical system events to which Call Home
responds.

Hardware Event Scenario

When server error-detection hardware determines that a hardware error has occurred, the
support center is notified that there is a hardware problem. This notification occurs in any
of the following ways:

      The system sends a Call Home service request packet.

       The Call Home hardware service request packet typically identifies the problem at
       the replaceable-unit level.
      The customer calls the support center.

      The customer initiates a support request through the Web by submitting a service
       request on the Unisys Support Online Web site at www.service.unisys.com.

When the notification is received, a systems analyst reviews the information. If the
replaceable unit can be identified, the analyst contacts the customer with information
about the problem and schedules maintenance based on the impact to the customer and
service coverage hours.

If the systems analyst cannot identify the problematic hardware, the analyst contacts the
customer for further information. Identifying the replaceable unit might require a Remote
Service session.

If the replaceable unit cannot be identified remotely (either the customer cannot enable
Remote Service or the information sent is inconclusive), the customer is informed of the
problem, the situation is assessed, and arrangements are made for maintenance. If the
requirement is for immediate service, a service representative is dispatched immediately
with replacement parts to perform on-site diagnosis. In many cases, depending on the
impact of the failure, the customer can continue operations and schedule the repair for a
more convenient time.

Controller-Related Event Scenario

The Call Home Controller feature currently supports LSI and Mylex PCI RAID
controllers when used for partition boot devices. The software that provides Call Home
Controller functionality must be installed on the Windows partition where either the LSI
Power Console Plus software or the Mylex Global Array Manager Server software is
installed.

Call Home works as follows on systems with these LSI or Mylex controllers:
   1. Errors and warnings generated by the LSI Power Console Plus software or the
      Mylex Global Array Manager Server software are collected.
   2. Received traps are placed in the Windows event log as informational, warning, or
      error events.
   3. The Call Home Generator (fault criteria database) determines whether these
      events are to be reported to the support center. If the events must be reported, the
      event data is collected in a service request packet, formatted, and sent to the
      Central Service for transmission to the support center.
   4. When the support center receives a Call Home controller-related service request
      packet, it routes the reported event for further analysis and processing.

The major types of events reported are as follows:

      Hard disk failure
      Consistency check errors
      Bad blocks

The support center contacts the customer and provides information about the problem.
The situation is assessed and arrangements are made for maintenance. If the requirement
is for immediate service, a service representative is immediately dispatched with
replacement parts to perform on-site diagnosis. In many cases, depending on the impact
of the failure, the customer can continue operations and schedule the repair for a more
convenient time.

Peripheral-Related Event Scenario

The Call Home Peripheral feature is supported only on approved storage subsystems.
Navisphere and the CX Event Response software provide Call Home peripheral
functionality.

Call Home works as follows on systems with CX storage system peripherals:

   1. Errors and warnings generated within the storage subsystem are collected.
   2. Received events are placed in the Windows event log as informational, warning,
      or error events.
   3. The Call Home Generator (fault criteria database) determines whether these
      events are to be reported to the support center. If the events must be reported, the
      event data is collected in a service request packet, formatted, and sent to the
      Central Service for transmission to the support center.
   4. When the support center receives a Call Home peripheral-related service request
      packet, it routes the reported event for further analysis and processing.

The major types of events reported are as follows:

      Controller failure and failover
      Path failure and failover
      Any other major replaceable unit failure
      Any failed recovery action by the subsystem
      Cache-related errors

The support center contacts the customer and provides information about the problem.
The situation is assessed and arrangements are made for maintenance. If the requirement
is for immediate service, a service representative is immediately dispatched with
replacement parts to perform on-site diagnosis. In many cases, depending on the impact
of the failure, the customer can continue operations and schedule the repair for a more
convenient time.

Accessing Call Home Tasks
To initiate any of the Call Home functionality,

   1. From the left side of Server Sentinel, select the appropriate system.
   2. Select Call Home from the list of tasks.
   3. Click any of the available tasks listed under Call Home.

The following table summarizes how to perform various tasks in Call Home.

To perform this task:                      Click . . .
Configure Call Home for the first time     Configuration Wizard.
or configure additional systems.
Update the Generator Service               Send test service request. Click the computer
configuration.                             name to be updated.
Update the Generator Service               Send test service request. Click a computer
configuration on multiple systems.         name. On the Generator menu, click Modify
                                           Multiple Generator Information.
Update the Central Service                 Service request delivery schedule.
configuration.
Send a service request test packet.        Send test service request.
Update the service request delivery        Service request delivery schedule.
schedule.
Change the maintenance mode                Service request status. Click Central
configuration.                             Service and then Maintenance Mode.
Specify e-mail notification.               Service request delivery schedule. Specify
                                           your choice in the Additional Packet Delivery
                                           Options form.
List and delete systems on which the       Send test service request.
Generator Service has been configured.
Copy configuration information from        Send test service request. Click the computer
another system.                            name for which you are updating data, select
                                            the computer from which you want to copy
                                            data, and then click Copy Data From.
Set Central Service trace options to be     Service request delivery schedule. Specify
used for diagnostics.                       your choices in the Trace Output form.
Set Generator Service trace options to      Send test service request. Click the computer
be used for diagnostics.                    name for which you want to set trace options.
                                            Specify your choices in the System
                                            Information form.
Access Health Advisor to view               Health Advisor.
information about partitions and
Service Processors on your system and
to send stability data to the support
center.

Context Help Information

For help information while viewing any of the Call Home pages, click Task Help in the
upper-right corner of the page.

Additional Information

The following links provide additional help information for Call Home:

   Copying Generator Service Information from Another System
   Deleting Service Request Packets
   Initiating the Send Packet Process
   Rescheduling Delivery of Service Request Packets
   Setting Trace Options
   Specifying E-mail as an Additional Service Request Notification


Remote Access
Purpose
The Remote management feature enables you to control the Service Processor desktop
from a remote workstation. This accessibility ensures that normal operations can be
conducted from any location at any time or that a connection can be made to a technical
expert at a remote facility to diagnose hardware and software faults.

User Benefit

The capability to manage, control and access the system remotely naturally will improve
system reliability and availability without requiring on-site personnel (lower staff costs).

Description
Server Sentinel also allows for a single remote console to act as the desktop for all
partitions in an ES7000. With the elimination of the need to have a keyboard, video
screen and mouse (KVM) tethered to the system, distance limitations are eliminated, and
a true ―lights out data center management‖ is achieved.

The remote desktop performs Sentinel management functions including navigation, event
management, troubleshooting and repartitioning. Also supports remote diagnostics by
Unisys technical service experts. The Sentinel home screen provides remote access to
these areas
• Service Processor Desktop
            o Event log information
            o Setup
• Partition Desktop
            o Operating System desktop
• Console Manager card
            o Administering configuration
            o Operating System desktop



Partition Desktop
After you click Partition desktop, a web page opens that asks you to select one of the
following computer names to view its partition desktop.

If you select a Windows partition:

The Console Manager Partition Desktop opens for the selected partition.

Console Manager enables users to communicate with and control Intel partitions from
one or more remote workstations. It does this by redirecting the functionality of the
partition peripherals, including the partition keyboard, video, mouse, floppy disk and CD-
ROM drives, to the remote workstation. Console Manager is made up of the following: a
Console Manager card with embedded firmware, Console Manager configuration or
utility/administration screens, and the Console Manager Partition Desktop. For Windows
environments, additional Console Manager partition drivers provide enhanced video
resolution and performance.

When a user enters at the Console Manager card’s IP address into a browser, the Web
server built into the card responds with HTML pages that validate the user and then
enable access to Console Manager.

The Console Manager hardware and software components provide the interface to the
operating system running on each Intel partition.

Note: Refer to the Console Manager Help for more information.
If you select an MCP Partition:

Selecting an MCP partition and clicking on the Continue button will initiate a telnet
session with the partition console. A user name and password prompt will appear.

Service Processor Desktop
This task allows you to control the Service Processor desktop from a remote workstation.

To initiate this task, click Service Processor desktop.

This link will initiate Remote Desktop Connection if the Service Processor is running
Windows XP or VNC if the Service Processor is running Windows 2000.

If a Security Warning opens asking if you want to install JVMxxx.tmp (where xxx may be
any combination of alphanumeric characters), click Yes.

Enter the password for Virtual Network Computing (VNC) in the application area then
click OK. The password for VNC is set during the VNC installation process. See the
installation documentation appropriate for your server for more information on the VNC
installation process.

After you enter the password, you will see the Service Processor desktop. From there you
can control the service processor remotely.

The disconnect button terminates an active VNC connection. There are other options
available see Virtual Network Computing (VNC) help for more information.

If you already have initiated the Service Processor desktop task and select it again, your
original session will be reset. Only one session from a remote client to the Service
Processor may be used.




Health Reporting
Purpose

The Server Sentinel Health Reporting feature provides the user an at-a-glance status of
the system’s health in the form of detailed performance and availability reports.

User Benefit
You cannot manage what you cannot see or define. Health Reporting enables users to
obtain ad hoc reports on the systems’ health. Uptime is increased when reports enable
users to proactively take precautionary corrective actions.


Description

Examples of health-reporting information include:

•   Cluster health
•   Configuration verification reporting
•   Operating system configuration reports
•   Inventory and properties for both partition hardware and system hardware
•   Installed software inventory

Configuration Verification Report
The Server Sentinel Configuration Verification Report performs an extensive check of
the configuration of the Server Sentinel software component configuration and identifies
problems, inconsistencies and suggested corrective actions.

Upon completing the initial installation, this report can be used to verify that all the
management components are operational and communicating properly.

Problems are identified with this icon at the end of the current line of the report.
 Additional information relative to the problem is provided at the end of each report
subsection.


System Configuration
Purpose

Server Sentinel provides intuitive configuration and setup procedures for the server
platform and the Server Sentinel interface. These procedures enable you to:

•   Update configuration options for Server Sentinel.
•   Add additional computer systems to be monitored.
•   Manage the hardware platforms with various utilities and wizards.

For example, one of the available wizards defines a partition profile by specifying its
name, type, hardware units, and other attributes. The wizard presents each selection
sequentially and lets you choose the values for your partition. You can save the profile
for later use or install it right away.
User Benefits

Wizards facilitate the ease by which users can add servers to the environment quickly and
easily, manage servers and modify servers when needed, saving time and labor costs.

Description

Server Sentinel provides an interface for monitoring configuration information associated
with the systems and components running in your enterprise environment. It provides
access to a series of pages on which authorized users can define, modify, and delete
specific configuration information monitored by Server Sentinel.

System configuration options are available from both the level of the Sentinel
Management Server and from a system level. On the Sentinel Management Server level,
the configuration summary pages contain information about all components configured
on all systems currently being monitored by the management server. On a system level,
the system definition pages contain information about all components configured on a
specific system.

Whenever you change the network configuration for a partition — such as the IP address,
host name, or SNMP settings — you should verify that the system configuration pages
are correct.

From the configuration summary pages, you can

      Modify the list of monitored systems (for example, add a new system or stop
       monitoring an existing system).
      Initiate a manual refresh of the configured systems information being monitored
       by Server Sentinel and to display information for newly discovered systems on
       the network.
      Modify the SNMP or AppManager configuration settings associated with the
       Sentinel Management Server.
      View a list of Server Sentinel users configured on the Sentinel Management
       Server and their associated user roles.
      Access information about available software and firmware updates to the Sentinel
       Management Server.
      Access the system definition pages.

A description of the system definition pages and the tasks you can perform using them
follow.

View Configuration Information Page

This page displays configuration information about all components configured on a
specific system. This includes automatically detected system and partition information
and manually defined component information. You access the View Configuration
Information page by clicking the following links in the left pane of the Server Sentinel
page:

   1. The name of the system you want to view
   2. System Configuration link
   3. System Definition link

Authorized Server Sentinel users can initiate edit and deletion operations of specific
configuration information from this page. The View Configuration Information page
automatically refreshes to provide a consistent up-to-date view of the configuration
information.

Links under the Firewall Information and Add Manual Component headings enable you
to add a Server Sentinel firewall or partition to your configuration. You can also select
Add a Manual Component from the Add menu to manually add a partition to your
configuration. When you perform these operations, you integrate the added component's
information into Server Sentinel, which allows the component to be managed and
monitored by the navigation interface.

Note: The Add Firewall link under the Firewall Information heading is only available if
there is not another firewall currently configured.

Edit and Delete Information Pages

The component information that you can edit or delete is identified by Edit and Delete
links in the Options column of the tables. Clicking these links displays corresponding
Modify (edit) or Delete information pages. Information fields that you can edit or delete
are identified by an asterisk (*). Information in fields without an asterisk cannot be
overwritten or modified.

A Modify information page is provided for firewalls, Service Processors, and
automatically detected and manually added partitions. A Delete information page is
provided for firewalls and manually added partitions.

Changes and deletions take effect after you click the Save Changes or delete confirmation
(Yes) button at the bottom of the page. The updated View Configuration Information
page, on which you can view your modified configuration information, automatically
displays.

Add Component (Firewall/Partition) Page
The Add Components page provides the task interface that enables you to manually add a
component (firewall or partition). You can access this page from by clicking

      Add Firewall under the Firewall Information heading to add a Server Sentinel
       firewall. Note that this link is visible only if there is not another Server Sentinel
       firewall currently configured.
      Add Manual Component under the Manual Component Information heading, or
       from the Add menu on the menu bar, to add a partition. Note that only partitions
       that run operating systems that are supported by Server Sentinel (Windows, OS
       2200, MCP, UnixWare, or Linux) can be added.

Required information boxes are identified with an asterisk (*).
The information you enter from this page is integrated into Server Sentinel, enabling the
component you add to be managed and monitored by the navigation interface. Your input
takes effect after you click the Save Changes button at the bottom of the page. The
updated View Configuration Information page, on which you can view your modified
configuration information, automatically appears.

Hardware Platform Management
Understanding the System Environment
This system is a cellular multiprocessing (CMP) server that offers the benefits of
symmetric multiprocessing (SMP) and clustering technologies. System hardware
components are called units and are grouped in cells for cell-based systems. Units are
apportioned according to certain criteria to create partitions. Partitions support operating
systems that run user applications to perform useful work. Storage is assigned to
partitions through virtual memory units, called MEMs.

The following topics provide high-level information about the main concepts of your
system:

      Understanding Partitions
      Understanding Units
      Understanding Memory
      Understanding System Data Sets

The following topics provide direction for performing work on the system:

      System Tasks
      Partition Tasks

Detailed procedures are contained in task-specific help available from the interface. Other
topics in the user's guide and operations guide provide information about monitoring
partitions, performing work, and maintaining your system.

Understanding Partitions
A partition is a logical computer in the cellular multiprocessing (CMP) server. The
purpose of a partition is to support an instance of an operating system that, in turn, runs
your applications. A partition uses the hardware components, also called units, according
to certain rules. It must be associated with a specific system data set (SDS), which
controls how the units function.

The following topics describe the relationship between partitions and system hardware,
the advantages and purposes for using partitions, the types of partitions available, and the
rules for creating partitions.

      Understanding Partitions in Cell-Based Systems with Predefined Memory Units
      Understanding Partitions in Cell-Based Systems without Predefined Memory
       Units

See Partition Tasks for information on creating and configuring partitions.

See Also

See other topics in the user's guide and operations guide for more information about
partition tasks, units, memory, power components, and system data sets.

Understanding Units
Units are hardware resources. Partitions must own units to run applications and do useful
work. A unit owned by a partition is said to be "up" in that partition (partitions acquire
ownership using the Up command).

Units have the following features:

      Unit visibility: who can see and access units through the user interface. Visible
       units are members of partitions and are known to the operating system of the
       partition; you can bring visible units up and down in partitions. Internal units are
       not directly known to the operating system of the partition but are known to the
       system itself.
      Unit ownership: which partition can use a unit. Partitions must acquire ownership
       of visible units before using them and release ownership when finished. Acquiring
       ownership is part of configuring a partition and is accomplished by bringing units
       up in the partition.
      Unit dependencies: whether a partition needs to own additional units. A unit can
       be dependent on other units, both visible and internal. When one visible unit is
       dependent on another visible unit, the operating partition must own both units. For
       example, if a partition needs to use a CPB, the partition must own the PCI bridge
       and PCI bus that connect to the CPB. No requirement exists regarding the order of
       acquiring the dependent visible units. The partition cannot become viable until the
       dependency is met (other conditions besides resolving unit dependencies also
       affect viability). When a partition acquires a visible unit, the system automatically
       attempts to acquire ownership of all dependent units, including internal units. If
       ownership of the dependent units cannot automatically be acquired, the user-
       initiated operation of acquiring the visible unit is rejected and the unit state is not
       owned.

Acquiring ownership depends on

      Unit dependencies
      Availability of internal units
      The physical location of the unit

Releasing ownership is accomplished by bringing units down. Units also can be released
from a partition automatically. For example, the system can release units from a partition
if a fault occurs against the unit that makes the unit unavailable to the partition.

See Also

See other topics in the user's guide for more information about partitions, memory, icons,
and other terminology.

Note: For each task, you view only the functionality that applies to the system and your
role.

Understanding Memory
In partitions, memory is managed using memory units (MEM). When a memory unit is
brought up in (owned by) a partition, the system allocates the physical memory for the
partition from available resources.

Depending on your system type, memory is handled differently:

      Understanding Memory for Cell-Based Systems with Predefined Memory Units
      Understanding Memory for Cell-Based Systems without Predefined Memory
       Units

See Also

See other topics in the user's guide for more information about units and partitions.

Understanding Memory for Cell-Based Systems with
Predefined Memory Units
In partitions, memory is accessed using memory units (MEM). In cell-based systems with
predefined memory units, MEMs are visible hardware units with fixed allocations in the
physical memory, using either stacked or interleaved access. A stacked MEM uses all
available memory in one cell. An interleaved MEM uses memory in both cells of a cell
pair.
The following MEMs are defined for each cell pair, where x and y are cell numbers:

      Two full stacked MEMs, one in each cell (MEM_x, MEM_y)
      One full interleaved MEM that uses all memory in both cells (MEM_x_y)
      Two half interleaved MEMs that each use half the memory in each cell
       (MEM_x_y_H1, MEM_x_y_H2)

In addition, a special MEM (called MEM_ALL) uses all memory in all cells, provided
that all memory is available and not being used in another partition. MEM_ALL provides
interleaved access to all memory in the system for the highest possible performance. It is
useful when you want one partition to use all system resources.

Using advanced techniques, you can select the MEMs for partitions. For example, you
can configure the half interleaved MEMs in two partitions.

See Also

See other topics in the user's guide for more information about units and partitions.

Understanding Memory for Cell-Based Systems
Without Predefined Memory Units
In partitions, memory is managed using memory units (MEM). In cell-based systems
without predefined memory units, a MEM is a logical unit of memory that represents
physical space on one or more main storage areas (MSU). When a MEM is brought up in
(owned by) a partition, the system allocates the physical memory for the partition from
available resources. You can create a maximum of 128 MEMs in the system. Each
partition can own a maximum of seven MEMs at one time unless the operating system
for the partition enforces stricter rules. If sufficient resources are not available for the
MEM, the system cannot bring it up.

Creating Memory Units
MEMs have default hardware unit attributes, such as addresses, but you can configure the
following attributes:

      Type. MEMs should be exclusive, meaning that one partition at a time can use
       each MEM.
      Range. For Intel partitions, the possible ranges are
          o System Determined. Memory mapping is determined when the partition
             starts, depending on other MEMs already brought up in the partition. Most
             MEMs should be system determined to allow for flexibility.
          o Low. Memory is mapped below 4 GB and below the PCI hole. The total
             size of all low memory units plus the fixed PCI hole size cannot exceed 4
             GB. At least one memory unit of 4 GB or less must be mapped in the low
               range for the boot memory. If no such MEM is defined but a suitable
               system determined MEM exists, that MEM becomes the boot memory.
           o High. Memory is mapped above 4 GB and above the PCI hole. The total
               size of all high memory units is constrained only by the available MSU
               space.
      Storage area assignment. The MSU allocations are determined by which cells
       are in the partition that uses the memory unit.
      Size. You can specify the number of gigabytes and megabytes to be allocated for
       each MEM.

Changing Attributes of Memory Units
To change the range or size of a MEM, you must bring it down.

To change the type or storage area assignment for a MEM, you must delete and re-create
it.

Deleting Memory Units
You can delete an existing MEM that is not owned by any partition.

Note: You cannot recover a deleted MEM. You can only create a new unit to replace it.

See Also

See other topics in the user's guide for more information about units and partitions.

Understanding System Data Sets
A system data set (SDS) is a set of hardware-related data contained in a single logical
container. Appropriate system data sets are installed with other software on the Service
Processor. Occasionally, additional system data sets are released as additional or revised
hardware becomes available.

A newly installed SDS is identified on the user interface as the installed default SDS.
This is the preferred SDS for use with the platform firmware. Under normal
circumstances, the system and partitions use this SDS automatically if it is compatible
with the previous SDS in use. If the new SDS is not compatible and you want to use it,
you must stop the partitions before installing the new SDS. Refer to the installation
documentation for specific details.

The user interface enables you to determine which SDS to use for the system and
partitions. Under normal circumstances, you should allow the system automatically to use
the preferred SDS that is installed with the other software on the Service Processor. If
you wish to use another SDS, you should consult with qualified support personnel first.
An SDS installation does not overwrite or delete existing SDSs. You can delete unwanted
SDSs manually.

System Tasks
The term "system" refers to all hardware and software in the server. Any task that refers
to the system refers to all server components.

The tasks that are available and other information that appears on each page depend on
the defined role of the user who logged on to the system.

Click Task Help for procedural details and more information on the displayed page.

View the message log for informational messages that appear when you submit
commands. You can choose to view the log in the lower pane of the user interface or in a
separate window.

System Monitoring Tasks
The banner at the top of the interface offers the following information and navigation to
some common system tasks.

Monitor Hardware Operating Status
You can discern the aggregate operating status of hardware alerts from the status
indicator icon. Click the icon for more information about individual alerts.

Operate in Maintenance Mode
The wrench icon provides access to maintenance mode, which enables qualified
personnel to perform maintenance tasks on the system. Click the icon to turn maintenance
mode on and off. Maintenance mode is off when the wrench is oriented vertically and on
when the wrench is oriented horizontally.

Partition Tasks
A partition is a logical computer in the system. A partition runs its own instance of an
operating system and application software, has ownership of hardware and memory units,
is associated with a specific system data set (SDS), and has its own requirements for stop
and recovery behavior. See Understanding Partitions for a complete description of
partitions.

The tasks that are available and other information that appears on each page depend on
the defined role of the user who logged on to the system.
Click Task Help for procedural details and more information on the displayed page.

View the message log for informational messages that appear when you submit
commands. You can choose to view the log in the lower pane of the user interface or in a
separate window.

Creating Partitions
Creating partitions is a system-level task. You can create partitions with any of the
following methods (see System Tasks):

      A wizard, which enables you to create and start typical partitions or define
       partition profiles with advanced options.
      The Create Partition system command
      The Partition tab on the System Commands page
      Other software that creates a partition profile, which you then install using the
       Install Profile system command or the Partition tab on the System Commands
       page

To change the name of a partition, you must delete it and recreate it.

Working with Existing Partitions
You can access information and settings for existing partitions from detailed partition
pages. To display these pages, click the partition name on the Active Partitions list or
the default Server Sentinel page.

Additional Management Tools


Clustering
Clustering Features
Server Sentinel provides access to two Application Sentinel clustering features:

Cluster Manager - The Application Sentinel for MSCS cluster manager feature provides a
Windows application service with a Microsoft Management Console (MMC) snap-in
interface that helps maintain the high availability of resources and applications in a
running Microsoft Cluster Service (MSCS)
Cluster Verifier - The Application Sentinel for MSCS cluster verifier feature is another
MMC snap-in that provides a diagnostic and health monitoring tool for enterprise server
MSCS clusters.

Cluster Manager
This topic explains what the Application Sentinel for MSCS cluster manager feature is,
what its uses and benefits are, and how it works.

It contains the following sections:

      Product Description
      Functions
      How the Software Starts
      The Manager Console
      Cluster Management Strategies
      Additional Information

Product Description
The cluster manager feature of Application Sentinel for MSCS is a Windows application
service with a Microsoft Management Console (MMC) snap-in interface that helps
system administrators maintain the high availability of resources and applications in a
running Microsoft Cluster Service (MSCS) based cluster. It does this by continually
monitoring system health and cluster health, as well as the health of applications in the
cluster. It monitors the event log and also monitors the performance of each cluster node
through instrumentation activity in the software and hardware environments. It then
interprets events by referring to a comprehensive list of possible problems and initiates
predefined or custom user-defined actions as needed to protect both application data in
the cluster and the cluster itself. The cluster manager feature provides a default set of
rules, each of which is predefined to respond to certain cluster events but can be modified
to fit the needs of a particular cluster.

The cluster manager feature is for use with clusters that are running MSCS in a Server
Sentinel environment under any of the following operating systems:

      Windows Server 2003 Enterprise Edition
      Windows Server 2003 Datacenter Edition
      Windows 2000 Advanced Server
      Windows 2000 Datacenter Server

Functions
The cluster manager feature includes both client and server components. The client
component (called the Manager Console) is installed on the local server or on any remote
workstation. The server component must be installed on the cluster nodes and can be
installed on all of the nodes using the ease of deployment feature available from the
Manager Console.

The main cluster management functions are viewed and controlled through the Manager
Console. These functions include:

      Defining rules that respond to different events (which address the health of the
       cluster nodes, the cluster software, and the applications in the cluster) with
       specific actions
      Reviewing reports of events detected and actions taken
      Setting user notification preferences
      Monitoring the status of cluster groups and applications

During cluster operations, the cluster manager feature provides continuing system level
and application level monitoring. For example, it can detect that:

      A cluster node is about to fail due to its operating system health deterioration
      One or more cluster resources are not able to come online due to an invalid
       configuration or another reason
      There is an impending application failure due to a known problem

Cluster and application health is surveyed through reference to event log and
instrumentation activity. Events occurring on a cluster are interpreted by symptomatic
comparison to a database of known events and actions. Whenever a potentially harmful
event is recognized, one or more of the existing default management rules are applied.
Some of these predefined rules even allow the cluster manager feature to act on its own to
fix the detected problem.

The power and flexibility of the feature lies in the ability it gives users to customize rules
and expand upon the built-in rules functionality. You can modify existing rules, create
new rules, include additional application events for rules to respond to, and set up the
priority that multiple rules are to respond in. In effect, rules can be tailored to fit the
needs of the cluster. The software can thus react exactly as needed to problem conditions
and take specified actions prior to a failure.

For example, one important action the cluster manager feature can take is to move
applications and resources to another node in the cluster when signs of a node failure
appear. This proactive failover happens before a halt or hang can occur, thus shortening
the recovery period normally associated with the unplanned failover of an application.

Therefore, the primary benefit of the feature is higher availability of services and
applications provided by the cluster.

Not all events are necessarily threatening to the operation or stability of a cluster. Some
are benign or merely notable. However, when an actual failure occurs, the event/trap
analysis that can be performed saves recovery time by pointing you to specific error
conditions in the cluster. The cluster manager feature can also provide timely
notifications of problems through e-mail, fax, pager, or system console messages.

How the Software Starts
When you launch the cluster manager feature by choosing a corresponding Server
Sentinel task, an automatic check takes place to determine that the node is ready for
remote management, then a prompt appears for installation of, or upgrade to, the latest
version of the software. Finally, you are prompted to start the Manager Service on the
node, at which point the Manager Console is displayed.

The Manager Console

The Manager Console is the user interface. It is an MMC snap-in for centralized
management adapted to the requirements of cluster management. It presents a familiar
tri-pane view with:

      A navigation toolbar at the top
      High level feature options listed in the left pane
      Detailed feature selections and fields shown in the right pane

After software launches, the Manager Console appears with the following action options
listed in the left pane for each cluster:

      Report
      User Preferences
      Rules Wizard
      Cluster Groups

It is also possible for you to add or remove a cluster from the list of managed clusters
using menu and toolbar options provided on the Manager Console. The remote
management snap-in provides data to the user interface controls and thus allows for
general cluster management.

Report

The Application Sentinel for MSCS cluster manager feature maintains a local event
log that is a record of cluster events and can be used as desired to build an event
report.

You can use the Report option on the Manager Console to open and view the log that the
Manager maintains. This log is a printable report of events and actions.

When this report option is selected, a report is generated from the events logged by the
cluster manager feature, showing all the events and actions in order, with the latest listed
first. There is no filtering for particular events. The report is generated in HTML.
Use these printed reports to analyze the sorts of events being logged on the cluster.
Depending on your analysis, you may want to add or modify rules or actions specified in
the rules or re-evaluate your cluster management strategy in general.

User Preferences

You can use the User Preferences option on the Manager Console to bring up a screen
that allows you to select how you want to be notified if a possible problem with the
cluster is detected.

Your choice of methods by which to receive notice of events that may require some
administrative attention includes:

      SMTP e-mail
      Console messages

Notification options should be configured according to your event notification
strategy and the methods that apply to your administrative environment.

Rules Wizard

The Rules Wizard option on the Manager Console helps you set up the monitoring and
reporting capabilities. It allows you to view the default rules applied and to customize
the response of the cluster to specific events by defining your own rules.

Using the Rules Wizard, you build or modify rules that consist of four parts:

      Events to be handled as part of the rule
      Actions to be taken when any of the listed events occurs
      Exception conditions under which the action should not be executed
      Name and description of the rule

The Rules Wizard works on one rule at a time, and the steps correspond to the parts of
the rule. When you use the Rules Wizard, you are selecting, defining, reorganizing, or
building individual rules in the Rules Database. The choices you make in using the Rules
Wizard should be made with an overall cluster management strategy in mind.

Cluster Groups

The Cluster Groups option on the Manager Console is provided for ready viewing of the
MSCS cluster groups. Using this feature, you can see the MSCS status without having to
open a second application window and toggle between the applications.

Remember, however, that if you need to perform cluster actions, you will do this through
MSCS itself. You must use MSCS Manager to affect the cluster, as the cluster manager
feature does not have this type of direct control over the cluster.
Adding or Removing Clusters

It is possible to add a cluster or remove a cluster from the cluster management list using
either the Action menu or selecting the appropriate toolbar icon on the Manager Console
to bring up a Cluster Management Dialog window. Not only does this window give you
the ready capability of selecting clusters to add or remove, but it also facilitates the
remote management of clusters.

Context-Sensitive Application Help

Once you bring up the Manager Console, online help is always available to you to
explain the various features and options and to describe how to complete cluster
management tasks. All you have to do is click on the Help icon (the question mark) on
the toolbar or the dialog Help button to open the online Help at the topic related to the
current screen view or dialog. You can also select from the topics in the help table of
contents or search by entering the key words of your choice. Help on dialog fields and
controls is available by clicking on the question mark in the upper right hand corner of
the dialog box, then selecting the desired field or control with the question mark cursor.

Cluster Management Strategies
Your strategy for managing your target cluster environment should include a plan to:

      Carefully monitor cluster status
      Protect applications on the cluster
      Set up failover conditions to keep the cluster running

Monitoring of Cluster Status

When first opened, the cluster manager feature establishes its interfaces and begins
immediately to monitor node and cluster health by capturing event information. This real
time monitoring continues as long as the feature remains active. You can check the
cluster status yourself as often as you wish by looking at the cluster group status and
event reports. If any critical events occur, you will be notified according to the
notification preferences you set.

Use the Reports tools to show logged events for the recent past. Consult this report when
investigating or attempting to avert problems.

Use the Cluster Group tools to show the online status of the cluster and the individual
resource groups as displayed by MSCS. Check this display to get a quick overview of
cluster status. Any resource shown as offline requires investigation and possible repair
action.
Protection of Clusters

A Rules Database provides default actions that protect the cluster against gross failure
events. However, the Manager Service running on the cluster is much more useful when
the system administrator chooses actions, exceptions, and parameters that apply to the
conditions of the particular workplace. For example, as the cluster administrator.
consider:

      The types of applications running
      The processing load on individual nodes
      Hours of heaviest transaction traffic on the network
      Work hours of the system administrator and the majority of users
      Number of employees available to support system maintenance
      Customer sensitivity to slow service
      Advertised service commitments
      Mission urgency or criticality

The Application Sentinel for MSCS cluster manager feature can be customized according
to system administrator objectives operating through a careful choice of events, actions,
exceptions, and notification options.

Understanding and Selecting Events
When selecting events from the Rules Database for a particular rule to manage, always
select from options available with the goal of maintaining cluster service and reducing
requirements for human intervention.

For example, assume that a particular rule is to be sensitive to the event or condition of
insufficient disk space remaining in the quorum device. This is important because if there
is not enough space on the disk for the quorum log files, then changes to the cluster
registry will be prevented. When available disk space is low on the quorum disk, the
problem must be resolved. The Rules Database includes recommendation information
for this event that some space on the quorum device must be freed up. Typically this is
accomplished by removing data or unnecessary files from the quorum disk or even by
designating another disk with adequate free space as the quorum device. Then sufficient
free space exists for the cluster to operate.

As the administrator, you can configure the cluster manager feature not only to monitor
for such events but also to execute one or more of several actions when a problem like
quorum disk space shortage is reported. One action could be to designate another disk
with adequate free space as the new quorum device. This is actually the recommended
default action for this scenario, but other relevant actions could also be selected such as to
stop the cluster service on one of the nodes or to simply notify administrative personnel
of the event. It is the administrator's job to configure the most appropriate action for the
given circumstances, keeping utmost in mind the goal of ensuring continued smooth
functioning of the complete cluster in mind.
Choosing Actions for Events
Consider the following facts and guidelines when deciding which actions to choose to
include in a rule for execution when specified events are detected.

      Any event you have selected for the rule will run all the actions in the rule.
      Any actions selected are subject to the exceptions you specify after selecting the
       actions.
      When an event occurs that is part of more than one rule, the actions defined in all
       relevant rules are executed.
      Actions in separate rule sets can conflict. When this happens, the action defined in
       the rule with the highest priority will be executed. If this produces an undesirable
       result, you have to reprioritize your rules. Warnings are issued when such rules
       are created.
      Some actions (called Advanced Actions) can bring down the cluster on all nodes.
      The Stop Cluster selection prevents automatic failback.

If an event is reported, and a rule addresses that event, then the actions described in the
rule will take place. It might be the failover of a group, or the whole node might fail over,
depending on how you configure the actions in the rule.

The choices under Cluster Actions are important to your cluster management
strategy. The following listing shows how each action option can be useful.

Action                       Result
Failover Cluster             If an event is reported that triggers this action, the cluster
Group(s)                     manager feature will fail over (to any available node )
                             one or more groups (selected under the parameter part of
                             this action in the Rules Wizard) owned by the affected
                             node. It will not fail over a group that is selected in an
                             exception.

                             This action is useful when a node reports unstable
                             conditions, but the other node may become overloaded
                             with all groups moved over.
Take Cluster Group(s)        If an event is reported that triggers this action, the cluster
Offline                      manager feature will take one or more groups (selected
                             under the parameter part of this action in the Rules
                             Wizard) owned by the affected node offline. It will not
                             affect a group that is selected in an exception.

                             This action is useful when a node reports unstable
                             conditions and is relevant (compared to the previous one)
                             when the other node should not be burdened to take on
                             all groups.
Bring Cluster Group(s)   If an event is reported that triggers this action, the cluster
Online                   manager feature will bring one or more groups (selected
                         under the parameter part of this action) owned by the
                         affected node online. This action will not affect a group
                         that is selected in an exception.

                         This action is useful when an event is reported that
                         indicates a system is returning to normal health.
Bounce Cluster Service   If an event is reported that triggers this action, the cluster
                         manager feature will take one or more groups (selected
                         under the parameter part of this action) owned by the
                         affected node offline and then immediately bring them
                         back online. This action will not affect a group that is
                         selected in an exception.

                         This action is useful in case of certain applications and
                         can also be used in single-node clusters.
Start Cluster Service    If an event is reported that triggers this action, the cluster
                         manager feature will bring MSCS online on the affected
                         node. This action will not apply to a node selected in an
                         Exception.

                         This action is useful to restart MSCS when an event
                         indicating MSCS service failure occurs.
Stop Cluster Service     If an event is reported that triggers this action, the cluster
                         manager feature will stop MSCS on the affected node. It
                         will not do so if the node is selected in an Exception.

                         This action is similar to a failover with additional ability
                         to override any node preferences some groups may have.

                         Note: This action selection prevents automatic failback.
Increase Quorum Log      This action increases the quorum log size by 64 KB to a
Size                     maximum of 512 KB.

                         This action is a quick fix in situations where the quorum
                         log runs out of allocated space.
Send an E-mail           If a triggering event is reported, this action directs the
Notification             cluster manager feature to send an e-mail notification to
                         the address specified on the User Preferences pane of the
                         Manager Console.
Send a Pager             If a triggering event is reported, this action directs the
Notification             cluster manager feature to send a pager notification to the
                         number on the domain specified on the User Preferences
                         pane of the Manager Console.
Send a Fax Notification     If a triggering event is reported, this action directs the
                            cluster manager feature to send a fax notification to the
                            fax number specified on the User Preferences pane of the
                            Manager Console.
Send a Console              If a triggering event is reported, this action directs the
Message                     cluster manager feature to send a console message
                            notification to the machine name or IP address specified
                            on the User Preferences pane of the Manager Console.
Log Application Event       If a triggering application event is reported, this action
                            directs the cluster manager feature to log the occurrence
                            of the event to the Application Events Log, which can
                            then be viewed using the standard Windows Event
                            Viewer. The event can also be monitored by other
                            components of Server Sentinel.
Run a Program               This action executes a program to which the path is
                            provided by the administrator in the action parameter. If
                            the program has any user interface, the interface will not
                            show.

                            The primary purpose of this action is to provide a
                            capability to execute custom scripts.

Note: When customizing notification options, consider the following:

      Some rules may be useful at night but not during your working hours. Consider
       running some rules separately (day versus night).
      Certain times of the workday or night may be less critical for downtime. Decide
       how long the cluster can be down without attention. Decide what the periods of
       greatest risk are.
      Make rules to enable or disable for convenience or to wait for something (for
       example, power cuts and predicted natural disasters).
      Some rules do not call for failover. For example, on a quorum disk failure you
       must have warning, but your action is to change the quorum to another resource.

Setting Failover Conditions

A failover is the transfer of workload from a failed node to a healthy node in response to
a node failure. Proactive failover is the deliberate transfer of workload from a failing
node to a healthy node in response to pre-indications of failure. The cluster manager
feature initiates a proactive failover based on default conditions or on custom conditions
defined in the Rules Database. A failover can be caused by problem conditions such as:

      Logged errors that indicate a system (cluster node) is about to fail
      Performance changes that indicate system resources are stressed or exhausted
      Faults in hardware devices managed by the operating system and its drivers
When setting up custom rules in the Rules Wizard, you must select actions for the cluster
manager feature to execute. For each rule you work in, the failover action has a default
setting (selected or not selected). If you wish to change the default setting, consider the
following before making the change:

      The Microsoft Cluster Service failover works after a failure occurs, and the
       failback works after the recovery occurs. Set your rules up to reduce failover time
       using manager rules. Failback does not occur when the cluster manager feature is
       in charge of the cluster.
      The Manager can take over and run the cluster itself. While the cluster manager
       feature operates with MSCS, not all of the actions taken are done through the
       administration of MSCS. Some are done directly through Server Exchange and
       other software interfaces of the Manager Console.

If Taking a Cluster Off Line
Some service events may require you to bring the cluster down for a substantial time.
Examples of such events are:

      Halts for hardware replacement (resident disk drives, shared disk drives,
       communication controllers, entire node)
      Halts for software upgrades

During the cluster down time, services may be provided by one node without benefit of a
second server, or both nodes may be taken off line. In these cases, planning the down
time is important. Most enterprise clients can tolerate a maximum service interruption of
five minutes.

Deciding on Rule Exceptions
Some exceptions take parameters as input, and these can be assigned by clicking on the
hyperlink in the exception name. If any exception condition is true, the rule is disabled
and the Manager takes no action. Four exception types are offered within the cluster
manager feature:

    Execute the selected   actions in this time window only – This exception allows
       you to run certain actions in a limited time window. If the time at which the
       action is triggered is not within the window, the action will not take place.
    Execute actions in this rule no more than once in a set time interval – This
       exception ensures that actions do not go into a loop. It is very important that this
       be used as part of each rule.

    Do not affect this Cluster group – This option excludes the selected group from
       the action being taken.
    Do not affect this Cluster node – This option protects a selected node from all
       cluster actions even if events are received that indicate problems with the node.

Additional Information

Once the MMC snap-in is running, access the online help for information related to the
cluster manager screen displayed.



Cluster Verifier
This topic explains what the Application Sentinel for MSCS cluster verifier feature is,
what its uses and benefits are, and how it works.

It contains the following main sections:

      Product Description
      Functions
      How Cluster Verifier Starts
      The Verifier Console
      Cluster Diagnostics
      Cluster Health Monitoring
      Additional Information

Product Description

The cluster verifier feature of Application Sentinel for MSCS provides system
administrators with a diagnostic and health monitoring tool for use on enterprise servers
running Microsoft Cluster Service (MSCS). You can use the cluster verifier feature to
diagnose cluster configuration problems, run targeted diagnostic tests, view test results,
monitor cluster health, and notify users of detected cluster problems. The feature presents
itself as a Microsoft Management Console (MMC) snap-in that accomplishes
communication between cluster nodes by way of Windows services running on a client
machine. The client maintains state information and allows multiple sessions. Services on
each server machine perform the diagnostics and health monitoring against that particular
cluster node. This arrangement facilitates remote execution of the cluster verifier feature
set as well as rapid ease of deployment (program installation and configuration across all
cluster nodes).

The cluster verifier feature is for use with clusters that are running MSCS in a Server
Sentinel environment under any of the following operating systems:

      Windows Server 2003 Enterprise Edition
      Windows Server 2003 Datacenter Edition
      Windows 2000 Advanced Server
      Windows 2000 Datacenter Server

Functions

Using Cluster Verifier, you can:

      Diagnose various environmental, communications, and cluster disk problems
      Specify and run targeted diagnostic tests
      View results of diagnostics that you scheduled through the Windows Scheduled
       Task Wizard to run at predefined times
      View diagnostic test result summaries or detailed reports
      Monitor the continuing health of clusters
      Send notifications about problems detected on a cluster by way of e-mail or
       console messages

The primary benefits of using the cluster verifier feature are faster problem solving and
help in maintaining higher availability of services and applications provided by the
cluster. The cluster verifier feature supports high availability standards by reducing
configuration and maintenance times.

Both client and server components are installed automatically during the software
installation process. Once the client is launched, the cluster verifier feature can be
programmatically deployed (installed or updated) on all remote cluster nodes. Cluster
diagnostic tests can then be run as needed from a local node or a remote node to verify
the configuration of any of the members of a cluster.

Although very useful for analyzing and troubleshooting cluster nodes, the cluster verifier
feature is not a repair tool. It needs to be used in conjunction with Windows Cluster
Administrator.

How Cluster Verifier Starts

When you launch the cluster verifier feature by choosing a corresponding Server Sentinel
task, a check takes place to determine if nodes detected are ready for remote
management. Automatic installation of, or upgrade to, the latest software version on the
cluster nodes takes place if you are launching on a clustered system. If not, the Verifier
Console appears and you can choose the Find Clusters option to add one or more cluster
as needed. Software upgrade takes places then, after which you can select a cluster and
start the cluster diagnostics function.

The Verifier Console

The Verifier Console is the user interface for the cluster verifier feature. It is an MMC
snap-in adapted to the specialized requirements of cluster diagnostics and monitoring. It
presents a familiar tri-pane view with:
      A navigation toolbar at the top
      High level feature options listed in the left pane
      Detailed form and field selections shown in the right pane

The Verifier Console appears with the following options listed in the left pane:

      Cluster
           o  Notification Options
           o  Schedule Results
      Find Clusters

Cluster Option
This option allows you to display the main diagnostics form for any selected cluster. By
making cluster node, test type, test execution, and preferred diagnostic routine choice on
this form, you set up the way cluster diagnostic testing is to occur. Two additional
options for each cluster allow you to establish user notification preferences by way of
addresses supplied (e-mail, page, fax, and console) and selected options as well as to
view diagnostic results from past scheduled runs of the cluster diagnostics.

Note: The actual scheduling of cluster diagnostics is accomplished through the Windows
Scheduled Task Wizard.

Find Clusters Option
This option allows you to inventory the system for available clusters and then to select
from the list displayed, adding one or more cluster to the cluster diagnostics and health
monitoring configuration. You can also use this feature to remove selected clusters form
the configuration.

Cluster Diagnostics

The cluster verifier diagnostics form provides the interface for running cluster tests.
It allows you to quickly and easily make choices about the tests you want to run.

Using cluster diagnostics, you can verify that a cluster is present, that communications
connections on the cluster are working, and that shared disks are online and available to
the cluster servers. Also use the diagnostics to find cluster configuration problems to fix.

Besides choosing to run selected diagnostics tests from the main diagnostics form, you
can:

      Follow the progress of tests as they run
      View and log diagnostic results
      View node properties
Diagnostic Testing
Cluster verifier diagnostic routines test the physical conditions essential to cluster
operation. These tests go beyond the static informational configuration of the cluster,
with live exercises conducted between the clustered nodes to confirm real activity. Any
problems or failures are identified immediately.

You can choose multiple Source Nodes to test on the selected cluster as well as multiple
Target Nodes to test against (nodes with a known good cluster configuration). Before
starting the tests, which can be launched immediately or scheduled to run at some future
time, decisions need to be made about the type of tests to be run, desired test execution
parameters, and which test of the available test categories are needed.

Types of Tests

There are two basic types of cluster tests you can choose to run:

      Non-obtrusive tests do not take over any cluster resources, and this type of test
       can always be safely run in a production setting. If you elect to run non-obtrusive
       tests only, all live data on the cluster is protected from interruption or corruption
       while this set of diagnostic routines is running.
      The Complete Suite of Tests option includes tests that may take over and affect
       some cluster resources, and data on the cluster will not necessarily be protected
       over the entire length of this diagnostic routine execution. Running the complete
       suite of tests is not safe in a full production environment although it may be
       appropriate in other cases (for example, when you are tuning the nodes in the
       cluster before coming online with live data).

Test Execution Parameters

By default, the selected test suite will run once and then report results. However, you can
specify a number (up to 999) to have the diagnostic routines run through multiple
iterations before displaying the results of the tests.

By leaving the Run Tests Without Intervention check box selected (as it is by default),
you can set the selected cycle of tests to proceed without the program intervening after
each test cycle and prompting you to continue. This check box should be selected if you
want run the diagnostics all the way through unattended. However, if you are going to be
present and monitoring the tests as they run, you probably want to uncheck this box.

In addition, all selected tests are set by default to run to completion without stopping after
errors are detected. Select the Stop on First Error check box if you want the diagnostic
tests to stop as soon as a test in which an error has been detected completes. In this case,
the results form appears immediately and all remaining tests are ignored. You can
use this stop-on-error feature to isolate problems in real time if you are attending to the
testing.
Test Categories
There are three test categories from which you can choose to run specific tests:

       Environmental – checks static server configuration data and node clusterability
       Communications – checks private, public, and mixed network connections,
        domain controller connections, and remote procedure call functionality
       Cluster disk – checks quorum disk and shared disk operation and verifies the
        SCSI bus rest and reserve function

The default selection is to run all the tests in all three categories. If you want to skip one
or more tests within any of the categories, you can deselect the tests you do not want to
run. You can also deselect an entire category of tests and thereby ensure that no tests
within that category will run.

Following Test Progress
A diagnostics progress form automatically displays after you start up cluster verifier
diagnostics. A test status bar progresses for each test as the selected tests are run, and a
count is kept for how many tests are yet to be run. Test status statements are also
continuously refreshed to inform you about tests still running, tests completed (with
errors or warnings if there are any), and tests aborted. If you selected multiple passes of
tests to be run, you can follow the count of the number of passes run so far. You can also
change test execution parameters, track exceptions encountered, and even stop the testing
if you need to from this form.

Changing Test Execution Parameters

Whatever you selected with respect to the Run Tests Without Intervention and Stop on
First Error options earlier, you can change these options on the progress form. It is
recommended that if you select Stop on First Error, you also elect to Display All Details
on this form. These are the selections that will help you most in system-attended real-
time troubleshooting. Conversely, if you are running the tests unattended, you probably
would do best to select Run Tests Without Intervention along with Display No Details.

Tracking Exceptions Encountered

Information about the tests is displayed on the diagnostics progress form, including a
summary of exceptions (errors and warnings) encountered in the testing. This shows you
at a glance the number of exceptions encountered in the test currently running and also
displays a running total of all exceptions encountered during this run of diagnostics.
Viewing and Logging Test Results
When you can tell on the progress form that the selected diagnostic tests have completed,
a diagnostics results form opens automatically. You can also display this results form
by choosing the Show Results button on the main diagnostics form.

The diagnostics results form has three tab views that allow you to review statistics on
the diagnostics run, limit the inclusive view (shown by default) of all diagnostic messages
generated, and write the results to a permanent log file.

Diagnostic Statistics

Reviewing the available statistical summary of results can help you to make
determinations about how your tests proceeded, whether your testing found any serious
errors, and whether there are any indications that the cluster is not viable.

Statistical fields displayed include the number of diagnostics that were:

      Attempted
      Completed with errors (problems that prevent cluster operation)
      Completed with warnings (violations of cluster best practices that still allow the
       cluster to operate) but no errors
      Completed with no exceptions (no errors or warnings)
      Skipped or stopped

Diagnostics Messages

The most important results to look at are the diagnostic messages themselves. By default
all test messages are reported, not just messages related to exceptions (errors and
warnings). You can elect, however, to filter out certain messages and view only error
messages or just error and warning messages if you prefer.

The diagnostic messages tab of the results form shows you each diagnostic that
generated a notable diagnostic result, clear (simple language) message text
indicating what the problem encountered was, and recommendations for actions to
take to resolve the problem.

Going through the result messages and taking the recommended actions should resolve
the majority of the problems reported, after which you can run more diagnostics as need
be.

Writing Results to a Log File

The tab for log file options lets you quickly create a customized log file of the results
data. You can select the kinds of messages to log (all messages, only warnings and
errors, or only errors), the date and time format you prefer, and the sort of message
details you want logged. Message detail options allow you to log message numbers,
long message descriptions, and even associated recommendations.

Viewing Node Properties
The Show Properties button on the main diagnostics form brings up a node properties
form that allows you to view the cluster configuration form data under the following tab
headings:

      Cluster – shows the cluster name and IP address along with the name of the local
       node whose properties are being shown on the current properties form
      Node – shows the node name and operating system information for the local node
      Connections – shows the connections by name and IP address
      Network Domain – shows the name of the domain in which the local node resides
      Cluster Disks – shows the shared disk information relevant to the local node
      Mount Point – shows the volume mount data in support of Windows Server 2003
      Quorum – shows the disk information
      User – shows the user and user group assigned to the node
      Cluster Groups – shows the names of cluster resource groups used by MSCS to
       manage the cluster

These tab views are for informational purposes (not for configuration modification).

Notification Options
The Notification Options selection (under Cluster) in the Verifier console allows you to
decide on the methods of user notification (e-mail, pager, fax, or console message
posting) you want to use on the cluster and the level of severity of messages you want
reported. You need to supply the required delivery address information to allow these
desired notifications to take place.

Schedule Results
Although cluster diagnostics are scheduled to run through the Windows Scheduled Task
Wizard, the results of the scheduled run(s) of diagnostics are viewed from the Verifier
Console through the Schedule Results option (under Cluster). Results of scheduled
cluster diagnostics are reported on a per-cluster basis. Therefore, to view scheduled
diagnostic results for another cluster, first make sure you have the desired cluster name
selected.

Cluster Health Monitoring

The Application Sentinel for MSCS cluster verifier feature provides continuous cluster
health monitoring by way of its Health Monitor Service, which starts automatically when
the feature is launched. The service detects problems on the cluster, logs them, and sends
alerts to the Server Sentinel interface.
Additional Information

Once the MMC snap-in is running, access the online help for information related to the
cluster verifier screen displayed.




Server Sentinel Guided Tours
The Server Sentinel Guided Tours provide detailed information about the world-class server
management capabilities provided by Server Sentinel.

Important Access Tips
   The guided tours are designed to run at a display resolution of 1024x768 and high color (16 bit), and
    they are best viewed with Microsoft Internet Explorer version 6.0 or later.
   Audio is available for each guided tour.
   Use the Restart button, located on the last frame of each tour, to start the current tour again.
   If you cannot see the entire display of the guided tours, press F11 to maximize the viewing area by
    hiding the browser navigation; press F11 again to return to the original display.


Overview
Four Guided Tours are available to help you learn about Server Sentinel and how it can help you
in your business.

Tours

Using Server Sentinel to Manage Your Business Processing
http://epas1.rsvl.unisys.com/pcproducts/docs/Sentinel/Sentinel4.4/PDF/SS4_BusinessProcessing.html

Demonstrates how to manage system resources to support applications by starting and stopping
partitions at specific times to suit the processing needs of your business.

Using Server Sentinel to Resolve Alerts
http://epas1.rsvl.unisys.com/pcproducts/docs/Sentinel/Sentinel4.4/PDF/SS4_ResolveAlerts.html

Demonstrates how to recognize and resolve an alert on the Server Sentinel interface.

Using Server Sentinel to Allocate Resources
http://epas1.rsvl.unisys.com/pcproducts/docs/Sentinel/Sentinel4.4/PDF/SS4_AllocateResources.html

Demonstrates how to use the Server Sentinel interface in an Intel partition to check memory
usage for a specific application and configure additional memory.

Using Server Sentinel to Monitor Your Enterprise
http://epas1.rsvl.unisys.com/pcproducts/docs/Sentinel/Sentinel4.4/PDF/SS4_MonitorEnterprise.html

Demonstrates how to use Server Sentinel to monitor an enterprise that starts with a single system
at one site and grows to multiple systems at dispersed locations.


Note: The Server Sentinel Guided Tours are also available from the Server Sentinel interface. To
access the tours from Server Sentinel, click Libraries and tours on the Server Sentinel Help
menu.
The Sentinel Management Server
A Sentinel Management Server is simply an independent server that is running Sentinel
Management Server software. You can have one or more independent servers acting as
Sentinel Management Servers in your environment. You can provide your own servers or
purchase them from Unisys. (See the ―Requirements‖ section for more information.)
Note that to access the full range of Server Sentinel management capabilities, your
environment must also include AppManager® management server software.


Installation Scenarios
There are four basic scenarios that you can follow to obtain the full benefits of a Server
Sentinel management:

    1. Install Sentinel Management Server software and AppManager management
       server software on one independent server in your environment. This option
       produces one Sentinel Management Server and conserves your hardware
       resources.

    2. Install Sentinel Management Server software on one independent server, and
       install AppManager management server software on another independent server
       in your environment. You might want to choose this option, for example, if you
       are a current customer of NetIQ Corporation and you already have a dedicated
       server running AppManager software. This option produces one Sentinel
       Management Server.

    3. Install Sentinel Management Server software and AppManager management
       server software on one independent server, and install Sentinel Management
       Server software only on another independent server in your environment. This
       option produces two Sentinel Management Servers for redundancy.

    4. Install AppManager management server software on one independent server, and
       install Sentinel Management Server software on two independent servers in your
       environment. You might want to choose this option, for example, if you are a
       current customer of NetIQ Corporation and you already have a dedicated server
       running AppManager software. This option produces two Sentinel Management
       Servers for redundancy. For even greater redundancy or for environments with a
       large number of systems, you can add any number of Sentinel Management
       Servers.


Advantages to Using a Server Sentinel Management Server

A Sentinel Management Server is an optional component, but using a Sentinel
Management Server can provide the following advantages:

•   Including a Sentinel Management Server in your environment enables you to obtain
    the benefits of full Server Sentinel functionality. If you do not include a Sentinel
    Management Server, only a core set of Server Sentinel functionality is provided by
    default for platform management.

•   Each Sentinel Management Server can manage up to 50 qualified systems (or 200
    operating system instances) that are running Server Sentinel releases 2.0 and higher.

Notes:
- If a selected system is running a Server Sentinel version prior to 3.0, a new
browser instance will open for that system when it is selected.
- Systems running different releases of Server Sentinel (3.x versus 4.x) might
display the same features in different ways; however, there is no loss of
functionality.

•   You can manage systems
          o Locally, by using Internet Explorer on the Sentinel Management Server to
             view the status of the systems in your environment

•   Remotely, by using Internet Explorer on one or more client workstations to connect to
    the Sentinel Management Server and view the status of the systems in your
    environment

Servers That Can Host Sentinel Management Server Software
You can install Sentinel Management Server software and AppManager management
server software on any independent server in your environment that meets the
appropriate requirements. (See the ²Requirements² section for more information.) You
can provide your own hardware, or you can purchase one or more qualified servers from
Unisys.

If you already have an independent Sentinel Management Server (which might have been
referred to simply as a management server in a previous release), you will be able to
upgrade its software and use it to manage systems running Server Sentinel 4.x as well as
systems running earlier releases of Server Sentinel.


Management Server Requirements
Hardware Requirements

All management servers require the following minimum hardware configuration:

•   2.4-GHz Xeon processor (single processor) with at least 512 megabytes (MB) of
    memory
•   Super VGA color monitor with a display area of at least 1024 x 768 pixels; a
    1280x1024 display area is recommended

•   Fast Ethernet network interface card (NIC)

•   Connection to the network through a network segment that can access the partition
    operating systems and Service Processors

Software Requirements

A management server can run any of the following Windows operating systems.

Note: The following are the latest supported Service Packs for each operating system.
• Windows 2000 Server, with Service Pack 4
• Windows 2000 Advanced Server, with Service Pack 4
• Windows Server 2003 R2, Standard Edition, with Service Pack 2
• Windows Server 2003 R2, Enterprise Edition, with Service Pack 2
• Windows XP Professional, with Service Pack 2

Note: If, after you complete the Sentinel Management Server configuration, you decide
to upgrade the management server operating system, see Appendix B for information
about batch files that you might be required to download.

All management servers require the following software configuration:
• Internet Explorer 6.0 and higher
• TCP/IP

Each management server must be configured to run TCP/IP with either static or
dynamic IP addressing.

•   SNMP service

SNMP must be configured on the Sentinel Management Server and on all partitions
and Service Processors that it manages.

Note: For information on how to configure the SNMP service on partitions and
Service Processors, refer to your system installation and configuration documentation.

•   Valid administrator group privileges on the system

•   User authentication requirements

User authentication on all of the components of each system you plan to manage
(including Service Processors, partitions, and client workstations) must be the same
as the Sentinel Management Server. Unisys systems are installed using the Microsoft
workgroup model, meaning that you must configure the same user name, password,
and group membership on each component. See your system planning documentation for
more information.

•   Internet Information Services (IIS)

•   If you intend to configure the Call Home Central Service from your Sentinel
    Management Server, hardware and software that enables you to connect to the
    Internet (either a modem or a network interface card)

•   If a default instance of SQL Server is present, it must be SQL Server 2000 or
    Microsoft SQL Server Desktop Engine (MSDE). The default instance must be
    configured to use SQL Server and Windows authentication mode (also known as
    mixed mode). You can use SQL Server Enterprise Manager to change the
    authentication mode for a SQL instance.

Note: If you do not have SQL Server installed on your system, the Sentinel
Management Server installation program installs MSDE, properly configured to use
SQL Server and Windows authentication.

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:1
posted:9/26/2011
language:English
pages:98