In the rush to bring up the site, I inadvertently installed several NT 3.51
workstations with the wrong time zone configured. Software distribution
applications were hard to run because of this mistake. So I used KiXtart to
determine which workstations had the wrong configuration. The script, which is shown in Listing 3, performs three steps:
1. Scan the Registry for the time zone.
2. Log the machine name and the time zone setting if incorrect.
3. Log all machine names to a file that will ensure that incorrect machines
are logged only once.
Since fixing the time zone problem, I have successfully used KiXtart for
various other projects. KiXtart is a very useful tool for administering NT
networks.
—Marty Scher
mscher@advanta.com
Save EffortUse a Tab
I found a neat shortcut for people who often use the command prompt. In the
Registry, go to HKEY_CURRENT_USER\Software\Microsoft\Command Processor and
add or modify the value CompletionChar of data type REG_DWORD with a data value of 9. The data value (9) is the ASCII value for the Tab key (but you can easily change it to any other key on the keyboard).
Now, when you enter a command, you just need to type in part of the file or directory name and then press the Tab key. Windows NT will automatically
complete the file or directory name. For example, if you enter CD C:\ Prog and then press Tab, NT will complete the case directory name to CD C:\Program Files.
If more than one file or directory matches the partial name, you can browse
the retrieved list by pressing Tab several times. For example, suppose you type
in back and the filename C:\backup appears, but you want the file
C:\backup1 instead. Just keep pressing Tab until that filename appears. This procedure is especially useful for long filenames and filenames with spaces.
—Sascha Frassini
sascha_frassini@ccmail.gb.swissbank.com
Don't Lose the Online Election
I'm impressed with the way NT 4.0 handles WANs. Marine Forces Pacific and
its subordinate headquarters chose NT 4.0 as the network operating system for a
recent joint exercise with the Republic
of South Korea. I was part of the team that built and then tore down the
large network.
The WAN topology for this two-week exercise consisted of 20 Cisco Systems
routers spanning the South Korean peninsula. About 20 class-C IP networks
connected the routers. With the assistance of the LMHOSTS file and Windows
Internet Name Service (WINS), the team connected six NT domains and established
trusts among them. Liaison officers in two remote sites were able to connect to
the domain and read their Microsoft Exchange email five routers away.
But then the trouble began. The initial symptoms appeared when I was trying
to access the User Manager for Domains program from my Primary Domain Controller
(PDC). The error read "Cannot access this domain, unable to locate domain
controller." My first reaction was immediate panic. I shut the system down
and then brought it back up again. When I logged on as the administrator, I
noticed that some services didn't start. The severity of my problems became
clear when I looked at the Server Manager program. In a single online election,
my PDC had been demoted to a workstation.
I found the reason for the crisis in the Event Viewer: Another computer had
the same name as my PDC. A computer hundreds of miles away on the other side of
four routers was seriously affecting my PDC.
Fortunately, earlier in the week, I had convinced a fellow technician to
stand his machine up as a classified Backup Domain Controller (BDC) and to load
the last copy of Exchange onto it. With that BDC, the team managed to bring up
the PDC's Microsoft Exchange System Attendant and Information Store services.
The team then used these services to migrate the mailboxes from the bridgehead
server to the backup server. With the mailboxes intact, the team promoted the
BDC to the PDC.
To bring the downed server back up, the team needed to rename it and reload
the operating system to stand up another BDC. The Help desk immediately
mobilized the client workstations for a quick reconfiguring of the Exchange
server's new location. With the machine renamed, the team reloaded WINS and
added several other WINS servers. The team then began Exchange Directory Service
replication. All was well.
The team's experience illustrates that details matter when planning the
architecture of multidomain WAN systems. The team failed to plan naming
conventions down to the node. The team now knows that a site-unique computer
name must also be WAN-unique.
—Jeffrey Fukushima
fukushimaj@mfp.usmc.mil