|
|
Windows
NT 4.0 Server - Cramsession
|
|
|
|
- Installation
- Minimum requirements for NT Server
4.0:
- 125MB Free disk space
- 16MB RAM
- 486/DX33
- CD-ROM (if not installing over
network)
- VGA compatible display
- Windows NT Setup
- WINNT32.EXE is used to upgrade only from
a previous version of Windows NT.
WINNT.EXE is used for the regular Windows NT
setup, or an installation through DOS or Windows
95.
Upgrading from Windows 3.1x or a previous
version of NT will keep all user, network and program
settings.
There are no conversion options from Windows 95
to Windows NT that will allow you to maintain user
settings. To dual boot between the two, install NT in
a separate directory and reinstall all your
applications.
Command modifiers for installation:
/B |
Put boot files on hard drive instead of
using boot floppies (takes an extra 4-5MB of
hard disk space). |
/S |
Specify source file location(s) - multiple
locations will speed up installation. |
/U |
Specify answer file location for use with
unattended installation - MUST be used with /s
to specify source file location(s). |
/T |
Specifies location of temp directory created
for install (/t:<path>). |
/OX |
Create the setup disks from CD-ROM or shared
network folder. Used to replace damaged boot
disks. |
/F |
Don't verify files. Can speed up
installation. |
/C |
Don't check for free space when creating
boot disks. |
/I |
Specify setup information (.inf) file. This
file tells setup how to run. The default name is
DOSNET.INF. |
Setup disks can be created by running WINNT.EXE
/OX or running WINNT.EXE from the cdrom.
- Answer file - Used when performing unattended
installs. Provides information that would
normally be answered by the user during setup.
Default name is UNATTEND.TXT.
- UDF (Uniqueness Database File) - Used in
conjunction with the answer file when performing
unattended installs. Provides information for settings
that are user or group specific. Default name is
$UNIQUE$.UDF.
To uninstall NT on a FAT partition, you will
need to boot to DOS, run SYS.COM, and remove the WINNT
directory and files.
In the server properties menu, there are
options to optimize server memory for certain
situations.
Minimize Memory Used |
Allows memory to be allocated for up to 10
network connections. |
Balance |
Provides memory for up to approximately 64
connections. |
Maximize Throughput for File Sharing |
Optimizes server memory for file sharing
operations(default). |
Maximize Throughput for Network
Applications |
Optimizes server memory for server-based
network applications. Key word is
SQL. |
- Virtual memory
- Virtual memory can be controlled in the Control
Panel -> System properties under the Performance
tab.
The paging file size can be in/decreased here,
and even distributed across multiple drives to speed
up access.
The most efficient paging file is distributed
on several drives but not on the boot or system
drive.
The initial paging file size equals the amount
of RAM in the system plus 12MB for NT Workstation and
the amount of physical RAM for NT Server.
Paging file size can increase during operation,
but will not shrink. Page file size will be reset when
the computer is restarted.
-
Multiple Disk Sets
Disk Striping |
Divides data into 64k blocks and spreads it
equally among all disks in the array. Needs a
minimum of two hard disks. Does not provide
fault tolerance. |
Disk Mirroring |
Duplicates a partition on another physical
disk. Provides fault tolerance by keeping data
stored on two different disks, in case of drive
failure. |
Disk Duplexing |
Duplicates a partition on another physical
disk which is connected to another Hard Drive
Controller. Provides fault tolerance by keeping
data stored on two different disks, in case of
drive failure, and by having two hard drive
controllers, in case of drive controller
failure. |
Disk Striping with parity |
Distributes data and parity information
across all disks in the array. The data and
parity information are arranged so they are
always on separate disks. A parity stripe block
exists for each row across the disk. The parity
stripe is used for disk reconstruction in case
of a failed disk. Supports a minimum of three
disks and a maximum of thirty-two disks. |
Volume Set |
Merges numerous partitions into one drive
mapping. Drives are read one at a time. Does not
provide fault
tolerance. |
-
System and boot partitions cannot be part of a
stripe or volume set, but can be a part of disk
mirroring and duplexing partitions.
- Speed factors
- Disk striping will provide the fastest
read/write performance as it can read multiple disks
at a time.
- Disk striping with parity is slower, as it
has to write the parity information, but is still
faster than disk mirroring and volume
set.
- Disk mirroring is slow due to the redundancy
factor of writing the same information to two drives
at once.
- Volume set can only read/write one drive at a
time.
To recover from drive failure with disk
mirroring, you must install the new drive, boot the
system into NT, run Disk Administrator, break the
mirror from the Fault Tolerance menu, and then
re-establish the mirror. This will not be done
automatically.
To recover from drive failure with disk
striping with parity, you must install the new drive,
boot the system into NT, run Disk Administrator, and
choose the Regenerate option.
To recover from multiple drive failure with
disk striping with parity, you must install the new
drives, boot the system into NT, and restore the
system backup from tape.
- File systems
- NTFS has file level security, and is faster
over 400M, but has a larger overhead (cannot format a
floppy disk with NTFS) and cannot be read by DOS, WIN
3.1, WIN 3.1.1 or WIN95.
FAT16 is compatible with MS-DOS & WIN95 but
has no file-level security.
FAT32 is not NT
compatible.
- NTFS vs. FAT
- FAT
- Files and directories on a FAT partition only
contain the standard attributes of Archive, Read-Only,
System and Hidden.
- Cannot set local security access on a FAT
volume.
- Can convert the partition to NTFS by running
CONVERT.EXE
- A FAT partition can be defragmented by booting
with a DOS diskette and running DEFRAG.EXE
- File moved from a FAT partition to an NTFS
partition retain their attributes and long-filename.
NTFS
- NTFS partitions contain the standard
attributes, as well as security descriptors basing
file access from file-level security.
- Can set local security access on an NTFS
volume.
- Partition cannot be converted to FAT. The
partition must be deleted and recreated as a FAT
partition.
- NTFS partitions cannot be defragmented. To
defragment an NTFS partition, it must be formatted and
restored from backup.
- Files moved from an NTFS partition to a FAT
partition do not retain their attributes or security
descriptors, but will retain their long filenames.
-
Security
- Share-Level Security - Governs user access to a
resource through the network. Can be implemented on
NTFS or FAT partitions. Applied through the Sharing
tab of the resource's properties.
File-Level Security - Governs local user file
and folder security on NTFS partitions only. Applied
through the Security tab of the resource's
properties.
Share Security Levels
Full Control |
- Is assigned to the Everyone group by
default.
- Allows user to take ownership of files
and folders.
- Users can change file access
rights.
- Grants user all permissions assigned by
the Change and Read levels.
|
Change |
- User can add and create files.
- Grants ability to modify
files.
- User can change the attributes of the
file.
- User can delete files.
- Grants user all permissions assigned by
the Read level.
|
Read |
- User can display and open
files.
- User can display the attributes of the
file.
- User can execute program files.
|
No Access |
- User cannot display, access, or modify
files.
|
NTFS Permissions (For a Folder - a user
can...)
-
Read (R) |
Display folder names, attributes, owner,
and permissions. |
Write (W) |
Add files and folders, change a folder's
attributes, and display owner and
permissions |
Execute (X) |
Display folder attributes, make changes
to folders within a folder, and display owner
and permissions. |
Delete (D) |
Delete a folder |
Change Permission (P) |
Change a folder's
permissions |
Take Ownership (O) |
Take ownership of a
folder |
-
NTFS Permissions (For a file - a user
can...)
-
Read (R) |
Display file data, attributes, owner, and
permissions. |
Write (W) |
Display owner and permissions, change
file attributes, create data in, and append data
to, a file. |
Execute (X) |
Display file attributes, owner and
permissions. Run a file if it is an
executable. |
Delete (D) |
Delete a file. |
Change Permission (P) |
Change a files's permissions |
Take Ownership (O) |
Take ownership of a
file. |
-
-
Permissions are cumulative, except for No
Access, which overrides anything.
When a resource has both File-Level and
Share-Level Securities enabled, the most
restrictive security is given to the user.
File permissions override the permissions of
its parent folder.
Anytime a new file is created, the file will
inherit permissions from the target folder.
The priority of attributes to a file
is: 1)
File 2)
Directory 3)
Share File attributes override directory
attributes, which override share
attributes.
Copying within a partition |
Creates a new file resembling the old file.
Inherits the target folders permissions. |
Moving within a partition |
Does not create a new file. Simply updates
directory pointers. File keeps its original
permissions. |
Moving across partitions |
Creates a new file resembling the old file,
and deletes the old file. Inherits the target
folders
permissions. |
Auditing can be enabled in the User Manager.
The Event Viewer is used to view audited
events.
When using Event Viewer, only local
administrators can see the security log, but anyone
(by default) can view other logs.
Only administrators have the right to share
folders on NT Server ( and Power Users on NT
Workstation)
-
Groups and Account Management
-
- Global groups - Groups which contain users with
similar rights and requirements. Can only be created
on Domain Controllers, and can only contain users in
that specific domain.
- Local groups - Groups used to allow members to
access resources in the local computer/domain. Can be
created on any NT system. Should only contain global
groups from the computer's domain or a trusted domain
but can also include members (not recommended).
Creating new accounts requires one piece of
information: username
(Password by default is blank and this is
acceptable.)
Disabling an account is typically used when
someone else will take the user's place, or when the
user might return.
Delete an account only when absolutely
necessary for space or organization
purposes.
When copying a user account, the new user will
stay in the same groups that the old user was a member
of. The user will keep all group rights that were
granted through groups, but lose all individual rights
that were granted specifically for that
user.
NT Default Accounts
Backup Operators |
Group designated for members to backup and
restore computers from tape. Backup Operators
can only backup and restore from tape when
logged in locally to the computer. This
group is found on all NT Servers. |
Account Operators |
Group designated for members to manage user
and group accounts. This group is found only on
Domain Controllers. |
Server Operators |
Group designated for members to manage
resources, but cannot manage user accounts. Can
backup and restore from tape. This group is
found only on Domain Controllers. |
Replicator |
Group designated for NT computers to perform
directory replication. This group is found on
all NT
Servers. |
-
RAS (Remote Access Services)
- RAS is capable of using the following
connection protocols:
- SLIP - Has less overhead than PPP, but cannot
automatically assign an IP address, and only uses
TCP/IP.
- PPP - Can automatically assign IP addresses,
supports encryption and other protocols besides
TCP/IP.
- RAS - Used by Windows 3.x and Windows NT 3.x
clients.
RAS supports call back security to either the
calling number or to a specified, non-changing
number.
RAS for NT 4.0 supports multilink (the use of
more than one modem to achieve higher transmission
speeds). Multilink cannot be used with callback
security unless there are two (or more) ISDN
modems configured on the same phone number.
RAS uses NetBEUI as the default network
protocol, but can also use TCP/IP and IPX/SPX. TCP/IP
will need to be used if you are using programs that
utilize the Windows Sockets (Winsock) interface over
the RAS services.
RAS will default to the first network protocol
on each side of the connection. Thus, if NetBEUI is
the first protocol that is in common, Winsock
applications (such as a web browser) will not be
available to the client.
To speed up NetBIOS resolution on RAS clients,
put an LMHOSTS file on each client locally.
RAS encryption settings
Allow any authentication including clear
text |
This will allow RAS to use a number of
password authentication protocols including the
Password Authentication Protocol (PAP) which
uses a plain-text password authentication. This
option is useful if you have a number of
different types of RAS clients, or to support
third-party RAS clients. |
Require encrypted authentication |
This option will support any authentication
used by RAS except PAP. |
Require Microsoft encrypted
authentication |
This option will only make use of
Microsoft's CHAP (Challenge Handshake
Authentication Protocol). All Microsoft
operating systems use MS-CHAP by default. |
Require data encryption |
This option will enable the encryption of
all data sent to and from the RAS
server. |
RAS will write to a log file which can be used
for troubleshooting RAS services. In order to enable
RAS to write to the log, you have to enable it in the
Registry.
-
Netware
- NWLink (MS's version of the IPX/SPX protocol)
is the protocol used by NT to allow Netware systems to
access its resources.
NWLink is all that you need to run in order to
allow an NT system to run applications off of a
NetWare server.
To allow file and print sharing between NT and
a NetWare server, CSNW (Client Services for NetWare)
must be installed on the NT system. Both NWLink and
CSNW are automatically installed when Gateway Services
for Netware is installed.
Gateway Services for Netware can be implemented
on your NT Server to provide a MS client system access
to your Netware server by using the NT Server as a
gateway. You must have a group account setup on the
Netware server called NTGATEWAY. In this Netware group
you add user accounts of all the NT accounts needing
access to the Netware server. However, a single
account is all that is needed on the NT Server to
access Netware resources for all NT users.
NWLink is automatically installed when Gateway
Services for Netware is installed.
Frame types for the NWLink protocol must match
the computer that the NT system is trying to connect
with. Unmatching frame types will cause connectivity
problems between the two systems. If multiple frame
types are in use, you should manually specify each
frame type. If NT is set to auto sense it will only
detect one frame type and in the following order:
802.2, 802.3, Ethernet_II and 802.5 (token
ring).
If you decide to convert a Netware server to an
NT Server, you will first need to implement the NWLink
and Gateway Services for Netware on the NT Server.
Once the conversion has completed, you will need to
make sure all Netware workstations have had the
Microsoft (SMB) redirector installed on their systems
to access the NT Server.
Netware 3 servers uses Bindery emulation
(Preferred Server in CSNW). Netware 4 servers use NDS
(Default Tree and Context.)
There are two ways to change a password on a
netware server - SETPASS.EXE and the Change Password
option (from the CTRL-ALT-DEL dialog box). The Change
Password option is only available to Netware 4.x
servers using NDS.
-
Networking
- Computer Name Resolution:
- DNS (Domain Name Services) - Used to resolve
DNS host name to an IP address.
- WINS (Windows Internet Naming Service) - Used
to resolve NetBIOS computer name to an IP
address.
- HOSTS - File which contains mappings between
DNS host names and their IP addresses. Must be
maintained manually.
- LMHOSTS - File which contains mappings
between NetBIOS computer names and their IP
addresses. Must be maintained manually.
TCP/IP is an internet protocol currently used
for most networking situations. Each computer using
TCP/IP will contain a unique address in a
x.x.x.x format (where each x equals a
number between 0 and 255) and a subnet
mask.
Subnet mask - A value that is used to
distinguish the network ID portion of the IP address
from the host ID.
Default gateway - A TCP/IP address for the host
which you would send packets to, to be sent elsewhere
on the network (typically a bridge or a
router).
Common TCP/IP problems are caused by incorrect
subnet masks and default gateways.
Install a WINS server in addition to a DNS
server to alleviate traffic due to b node
broacasts.
If bandwidth is hogged by a particular group of
users on a TCP/IP network, create a separate physical
subnet by installing a 2nd NIC on the server,
installing a new hub, and putting the problem users on
this hub.
UNIX computers use the TCP/IP
protocol.
NetBEUI is a non-routable protocol that is used
solely by Microsoft O/S's.
Universal Naming Convention (UNC) - Universal
network pathname which is integrated into Microsoft
systems. Named as \\computername\sharename, where
computername = the NetBIOS name of the computer, and
sharename = the share name of the folder.
Trap messages are sent using SNMP (Simple
Network Management Protocol).
-
Profiles
- Profiles are the user settings which are loaded
when a user logs in. They can contain desktop and
start menu preferences. These files can be located
either locally or on a server which has been mapped in
the User Manager.
NTUser.dat and *.dat files are the typical,
user-configurable profiles used.
NTUser.man and *.man files are read-only. If
the user attempts to configure their desktop, the
*.man file will not be updated. When the user logs in
again, it will restore the original
profile.
You may copy profiles using the User Profiles
menu located under CONTROL PANEL | SYSTEM
PROPERTIES.
-
Policies
- Policies take precedence over profiles.
Individual policies take precendence over group
policies.
Machine policies take precedence over all
policies.
-
Printing
- Microsoft uses the terminology "Print Device"
to refer to the physical piece of hardware, whereas a
"Printer" is a conceptual idea describing the icon in
the Control Panel.
NT 4.0 has the option to maintain drivers for
different operating systems on the server. Each
operating system uses different drivers. For example,
NT 4.0, NT 3.51 and Win95 systems cannot use the same
print drivers. By installing the drivers for each of
these types of system on the print server, each of
these types of clients can automatically download the
driver they need without manual
installation.
NT clients (3.51 and 4.0) automatically
download updated drivers from the server. Win95
machines will initially download print drivers but
will not automatically update to a newer version of
the driver. Win 3.1x and DOS clients must have the
drivers installed on each client manually.
- Print Pooling - Consists of two or more
identical print devices associated with one printer.
- Availability - This option allows you to
specify which hours the printer can be printed to.
- Priority - This option specifies which virtual
printer should print first if other virtual printers
are trying to print to the same physical printer at
the same time. Priorities range from 1 - 99 with 1
being the lowest and 99 the highest.
You can select Restart in the Document Menu of
the printer to reprint a document from the beginning.
This is useful when a document is printing and the
printer jams. Resume can be selected to start printing
where you left off.
You can change the directory containing the
print spooler in the advanced server properties for
the printer.
To remedy a stalled spooler, you will need to
stop and restart the spooler services in the Services
applet of control panel.
Printing to a TCP/IP printer requires you to
know the IP address and printer name.
The DLC protocol needs to be installed in order
to connect to a HP print server.
The AppleTalk protocol needs to be installed to
communicate with Apple printers.
Use the PCL.SEP separator to switch from
PostScript to PCL.
Use PSCRIPT.SEP separator to switch from PCL to
PostScript.
-
Troubleshooting
- To create a boot disk, format the diskette from
the NT system you want a boot disk for (Win 95 and DOS
will not work), and copy over the following files:
NTLDR, NTDETECT.COM, BOOT.INI and NTBOOTDD.SYS (SCSI
only).
To create an Emergency Repair diskette, you can
choose to do so either during the installation of NT,
or you can run RDISK.EXE. When RDISK.EXE is run with
the /S option, the utility backs up user accounts and
file security.
To use the Emergency Repair diskette, you will
need to boot the server with the NT installation boot
diskettes, and choose to repair NT with the Emergency
Repair disk that was created.
The Emergency Repair Process can a) inspect the
registry files and return them to the state on the
repair disk, b) inspect the startup environment, c)
verify the system files and d) inspect the boot
sector.
To troubleshoot bootup problems, you can edit
the Boot.Ini file and add the /SOS switch to the end
of the Windows NT entries in the [Operating Systems]
section of the Boot.Ini file to display driver names
while they are being loaded. The VGA startup option
has /SOS added by default.
Use the Last Known Good option on bootup to
restore the system to a bootable state if problems
arise from switching video drivers or changing
registry settings.
Common error codes:
- No system or boot disk message when trying to
dual-boot = BOOTSECT.DOS is corrupt
- Copy single file non-critical error - could
not copy file = Occurs when you install Windows NT
from an unsupported CD-ROM or network drive.
- Server stop errors - In the System Properties
-> Startup/Shutdown tab, there are options to
configure where you would like the Server stop errors
to be written. The errors are written to a .DMP file
which is readable by the program DUMPEXAM.EXE. You
must have free space in a swapfile on your boot drive
equal to or larger than the amount of physical RAM in
your system in order to generate a dumpfile.
-
PDCs and BDCs
- To upgrade from a member server to a BDC or
PDC, NT Server must be reinstalled.
To downgrade from a PDC or BDC to a member
server, NT Server must be reinstalled.
To change a PDC to a BDC, or a BDC to a PDC,
you must promote a BDC to a PDC in the Server Manager.
There is no "Demote" option here, only Promote a BDC.
NT will disconnect the current PDC if online and
handle everything automatically.
Only when a PDC goes offline unexpectedly, will
there be an option to Demote to a Backup Domain
Controller. This will only be seen when the original
PDC comes back online.
A BDC cannot automatically promote itself when
the PDC becomes disconnected from the network. A BDC
will continue to service login requests during the
time that the PDC is unavailable.
-
Browser Services
- All NT Servers have browser services available.
The Master browser will maintain a browse list which
contains a list of all workstations, servers and
domains on the network. There can be only one master
browser per subnet.
The PDC will always be the domain master
browser. All BDC'S will be backup domain master
browsers and are capable of becoming domain master
browsers in the event of a PDC failure. All member
servers are capable of becoming master browers or
backup browsers.
You can disable the ability of a server to
become a master browser by making the proper changes
in the registry.
-
ARC Naming Convention
- The Advanced Risc Computing (ARC) path is
located in the BOOT.INI and is used by NTLDR to
determine which disk contains the operating system.
multi(x) |
Specifies SCSI controller with the BIOS
enabled, or non-SCSI controller. x=ordinal
number of controller. |
scsi(x) |
Defines SCSI controller with the BIOS
disabled. x=ordinal number of
controller. |
disk(x) |
Defines SCSI disk which the OS resides
on. When multi is used, x=0. When
scsi is used, x= the SCSI ID number of
the disk with the OS. |
rdisk(x) |
Defines disk which the OS resides on. Used
when OS does not reside on a SCSI disk. x=0-1
if on primary controller. x=2-3 if on
multi-channel EIDE controller. |
partition(x) |
Specifies partition number which the OS
resides on. x=cardinal number of partition,
and the lowest possible value is
1. |
multi(0)disk(0)rdisk(0)partition(1). These are
the lowest numbers that an ARC path can
have.
-
Performance Monitor
-
- Memory - add more RAM if you detect problems
with the following:
- Pages/sec - excessive disk paging. Should not
be above 20.
- Available bytes - virtual memory available.
Should not be below 4MB.
- Commited bytes - memory being used by
applications. Should be less than RAM in
computer.
- CPU - upgrade the processor if you detect
problems with the following.
- %Processor time - amount of time the
processor is in use. Upgrade if constantly over
80%.
- System Object: Processor Queue Length - should
not be over 2.
- Disks - upgrade hard disk or controller, add
another hdd controller to balance the load, or
implement disk striping for multiple I/O channels if
receiving inadequate disk performance.
- %Disk Time Counter - amount of time the disk
is in use. Should not be over 90%.
- Current Disk Queue Length - files in disk
queue. Should not be over 2.
Must run DISKPERF -Y to enable disk performance
counters.
Alert view allows alerts to be made when the
counters surpass the threshold you set.
Log view allows the tracked objects to be
written to a log file. Used to create a baseline for
future reference.
Report view gives the ability to present a
concise report of current
statistics.
-
UPS
- Wrong polarity setting (e.g. positive instead
of negative) can make the UPS alert service not work.
Instead of getting warnings and a clean shutdown,
you'll get an abrupt power off when the juice is gone
from the UPS.
Using a standard RS-232 cable instead of a
special UPS cable can cause unpredictable results,
including BSOD.
Special thanks to Joe
Seeley for providing updates to this
material. |
| |
|
|
|
Sponsor |
|
|
|
|