Microsoft KB Archive/139634

= Tips for Troubleshooting LAN Manager Problems =

Article ID: 139634

Article Last Modified on 9/30/2003

-

APPLIES TO


 * Microsoft LAN Manager 2.2 Standard Edition

-



This article was previously published under Q139634



SUMMARY
This article lists questions to ask and things to check when troubleshooting Microsoft OS/2 LAN Manager problems. This article applies to Microsoft LAN Manager running on Microsoft OS/2 versions 1.x. If you have a UNIX implementation of Microsoft LAN Manager, contact the manufacturer (Hewlett-Packard, AT&T, etc.) of the UNIX implementation for support.



What Version of the LAN Manager Server Are You Running?
A large number of articles exist in the Microsoft Knowledge Base that detail LAN Manager issues and resolutions. Many of these resolutions involve upgrades to later versions. Therefore, you need to know what version you are working with. Be sure to compare article dates with the version date of your server. Your LAN Manager 2.2 server may not work properly if you install a file from LAN Manager version 2.1.

The last released server version of LAN Manager is 2.2b, dated 12/93. LAN Manager version 2.2a is dated 6/93, and version 2.2 is dated 11/92. Check the file dates in the C:\LANMAN\NETPROG subdirectory to determine the customer's version.

What Hardware Is LAN Manager Running On?
LAN Manager, and more specifically OS/2, is limited to what hardware it runs on. Please see the following article in the Microsoft Knowledge Base:

102628 Microsoft OS/2 Version 1.3 Hardware Compatibility List

The Hardware Compatibility List (HCL) is also included with the LAN Manager 2.2 server documentation.

How Much Memory Does the Server Have?
OS/2 can address a maximum of 16 megabytes (MB) of RAM, and typically LAN Manager servers require this much memory. Some computers encounter trouble if more than 16 MB of RAM is installed. However, reducing the installed amount to 16 MB of RAM can be helpful.

What Kind of Hard Drive Controller Are You Using?
The hard drive controller requires a Microsoft OS/2 specific driver, and the driver must be specifically written to run on the OS/2 version you are using. Therefore, do not attempt to use the IBM OS/2 2.11 driver for your controller (or any device) on a Microsoft OS/2 1.3 server. Many drivers are included with Microsoft OS/2, but if a new system cannot run FDISK.EXE or a new controller does not seem to work, it may be due to the system requiring a driver (usually a Bus Interface Driver [*.BID] for a hard drive controller). This can be a problem since most companies have not been developing drivers for OS/2 1.3 for some time. Therefore, if you buy a new controller and the manufacturer does not provide an appropriate *.BID file, get another controller.

However, keep in mind that even if a *.BID file is available, Microsoft only provides support for the equipment listed on the HCL. Also, LAN Manager was never certified to run RAID arrays and to span partitions across drives.

Microsoft does not support or recommend these configurations.

What Type and Size of Hard Drives Are You Using?
When you use boot drives that exceed boot drive geometry specifications, problems can arise. Please see the following articles in the Microsoft Knowledge Base:

86863 Maximum Hard Drive Partitions with MS OS/2 1.3 for Server

87135 Large Drive Translation Under Microsoft OS/2 1.3

Basically, boot drives should not exceed any of the following values:

512 bytes/sector, 63 heads, 255 sectors, 1024 cylinders

Data drives do not have these limitations.

What Is the Size of Your Hard Drive Cache in CONFIG.SYS?
For additional information on setting cache size, please see the following article in the Microsoft Knowledge Base:

129669 HPFS Cache Size CACHEMEM and IFS Do Not Match

By default, if you use HPFS, the first line of CONFIG.SYS is as follows:   IFS=C:\LANMAN\NETPROG\HPFS386.IFS /I:C:\LANMAN /CACHE:AUTO If the cache is set to AUTO, the system sets aside all but 7 MB of system RAM for drive caching. Therefore, a typical server set at CACHE:AUTO has 9 MB cache. Under most circumstances, all LAN Manager servers should be set at no more than 4 MB. The CACHE parameter works only with values specified kilo byte (KB) numbers, therefore use CACHE:4096. If the server runs a lot of services that service a lot of clients simultaneously (that is, the server is "stressed"), it uses a lot of RAM. For that, you can set the CACHE value to 2048 or 1024.

NOTE: If you set the cache size in CONFIG.SYS from AUTO to a specific value, and you then run LAN Manager Setup, Setup reads C:\LANMAN\SETUP.INI and resets the CACHE parameter to AUTO. Therefore, the best way to change the cache size is through the Setup program.

How Big is Your SWAPPER.DAT File?
SWAPPER.DAT is the OS/2 dynamic memory paging file. It starts out at around 500 KB by default and grows as memory runs out. With the exception of SQL Server, which can use a 20 MB SWAPPER.DAT, values above a few MB are generally a sign that the server is overloaded. In this case you should remove services or move them to other servers.

Are there Any Printers or Printer Queues Controlled by the Server?
OS/2 can control up to 6 printers. Six printers can seriously reduce system resources. Typically, if someone has 6 printers, they are remoteboot printers (for example, HP JetDirect), although it is possible to hook up a serial board and use COM1 to COM4 and LPT1 and LPT2. If your system is stressed and you have a lot of printers, consider making this server a print server only. Demote it to Member status in the domain or even Standalone status if you can give your users guest access to your printers.

What Services Are You Running?
Type NET START at a command prompt. This lists the services that are running. The more services, the more likely the server is stressed. Knowing what services are running also gives you a better idea of what might be causing your problem.

What Protocols Are You Running?
More protocols means more possibilities for problems. To check what protocols you are using, check what protocols are listed in the CONFIG.SYS file. To check how the current protocols are configured check the C:\LANMAN\PROTOCOL.INI file. For example, many older servers are still loading XNS even though it is not used by the clients. Also, many server traps are protocol related. Knowing the protocols can sometimes tell you what is causing the trap.

What Role Does this Server Play in Your Domain?
LAN Manager has four server roles: primary domain controller (PDC), backup domain controller (BDC), Member server, and Standalone server. If you know what role the server is configured for, you know what functions that server performs, and what services it should be running, and so on. For example, PDCs, BDCs, and Member servers run the netlogon service, but only PDCs and BDCs validate domain logons. However, the Member server still runs the netlogon service, so it can get user account replications. Standalone servers do not validate domain logons, and do not replicate the user accounts, therefore they do not run the netlogon service.

If your problem is that netlogon does not run, knowing the server role then tells you if that server should run the netlogon service and why. Also, because standalone servers do not participate in user account replication, account names and passwords changed at the domain level have to be changed manually on the standalone server.

Important Configuration Files to Inspect

 * CONFIG.SYS loads drivers and configures OS/2.
 * STARTUP.CMD is equivalent to the MS-DOS version of AUTOEXEC.BAT. Usually, the only command in STARTUP.CMD is NET START SERVER.
 * LANMAN.INI configures the LAN Manager environment.
 * PROTOCOL.INI configures protocols.

Server Tuning and Server Autotuning
C:\LANMAN\SERVICES\READCFG.EXE displays old and new server values. LAN Manager tries to automatically tune (that is "autotune") certain values, and READCFG.EXE displays what the server tries to tune the next time the server is shut down and restarted. If the values are increasing, the server calculated that it needs more resources. For example, if READCFG.EXE shows numreqbuf has increased from 120 to 150, that does not mean the server has already allocated 150 buffers but that it calculated it needs 150 buffers. If there is a significant increase in values between the old and new server settings, you should shut the server down and restart it to let the new settings take effect.

NOTE: If READCFG.EXE displays an increase, shut down and reboot the server as soon as possible. For example, if you notice the increase on Friday, then leave the server idle over the weekend with the intention of shutting it down Sunday night, the autotuned values will have decreased reflecting the resources the server needs while idle on Saturday and Sunday. If you shut down the server Sunday night, users that start using the server on Monday, find that the server does not have the resources it needs. To avoid this problem, make sure you check the values displayed by READCFG.EXE before you shut the server down. Whatever values are displayed at that moment are the ones the server uses when restarted. If the displayed values show a decrease you may not want to postpone the shutdown.

Overriding Autotuned LANMAN.INI Parameters
You can add some of these values to the LANMAN.INI file, which turns off autotuning for that value. This is usually not recommended. Check the Administrator's Guide and Administrator's Reference for more information on individual values. Some values, if added to the LANMAN.INI, can keep the server from starting. With new installs, it is recommended that you let the server run for a while (2-3 days), then run READCFG.EXE to see if the server calculated increased values, and if so, shut down and restart the server so it can use the new values.

Server Statistics
Run NET ADMIN and choose Server Statistics from the Status menu. This shows you how many times buffers have been exhausted since the last time the server service started and how many network errors and system errors have been encountered.

ERROR LOG
Run NET ADMIN and choose Error Log from the Status menu. The recorded errors have a number associated with them. A NET3190, for example, is a NETWKSTA.EXE internal error. This information is included in the error log. You can then search the Knowledge Base for information on NET3190 and get more information on what can cause the error, the severity, and how it you can resolve it.

The NET HELPMSG Command
While the information provided by this command is often generic, it can give you some information on an error if the error log is unavailable. At a command prompt, type:

net helpmsg

For example, if you are getting a NET3190 error, type:

net helpmsg 3190

This returns an explanation of the error.

Additional query words: prodlm 2.20

Keywords: KB139634

-

[mailto:TECHNET@MICROSOFT.COM Send feedback to Microsoft]

© Microsoft Corporation. All rights reserved.