Author: Neftaly Malatjie

On completion of this section you will be able to maintain a multi-user networked operating system. 

1. The maintenance ensures that any configurations made to the multi-user networked operating system are in accordance with organisation requirements. . 
2. The maintenance ensures that changes to the multi-user networked operating system maintain the integrity and security of the software. . 
3. The maintenance ensures that monitoring and tuning of the system maximises operation and performance, and meets manufacturer and organisation specifications. 

Once the problem with the operating system has been identified and resolved, the final step is to document it.

The following are reasons for documenting problem causes and resolutions

• Future references
• As proof that the problem was rectified

The following is a template that can be used;

Various diagnostic tools are available to identify multi-user operating system. We shall look at some of them below;

Task Manager

The information in the Task Manager can be misleading if you don’t have a good understanding of the Windows memory architecture. A more accurate assessment of the desktop’s memory is at the bottom of the window, which in this example says Physical Memory: 65%. This means the computer is using 65% of physical memory, a fact verified in the Memory bar chart.

But what about the Physical Memory being Free = 0? Well, in this case, free refers to free memory pages that are not in cache and have not been written to.

The CPU usage data is more straightforward. In Figure 1 above, there are two graphs in the CPU Usage History. This indicates two processors — or cores. The bar chart matches the CPU Usage: 31% in the bottom of the Task Manager window.

To determine which processes are the top hitters for memory, CPU and other resources, click the Processes tab. Or you can click Resource Monitor in Performance tab. The resulting console shows dynamic CPU, disk, network and memory graphs, as well as top processes in the drop-down bars, as shown in Figure

this example, the memory information has been expanded. Each column can be sorted for additional troubleshooting data. Note that svchost.exe is a generic process that many system resources and applications can run under. In the Processes tab, we see only svchost.exe and a process ID to look up. But in Resource Monitor, we see the actual service or app in parentheses (see the red circles in Figure comparing the two).

PerfMon

If you want to use Performance Monitor (PerfMon), you can launch it from Monitoring System Activity with Performance Monitor, which is under Learn More on this screen. Additional data can be extracted for experienced personnel by setting specific counters and then analyzing the data in PerfMon.

PerfMon is beyond the scope of this article, but for those familiar with Performance Monitor in earlier versions of Windows, note that Windows Vista and Windows Server 2008’s Resource Monitor have very powerful and convenient options. Fortunately, performance logs taken on Windows XP or Windows Server 2003 can be fed into Resource Monitor on Vista or 2008, and they can use many of the features in the new user interface. 

Symptom: The user cannot access the internet or key applications. The network appears to be down.

Cause: Domain Name Services may be to blame. The client PC cannot resolve the name of the server with the IP address of that server, so it will not send a connection request. This is often caused by having the wrong DNS server configured on the client, sending DNS requests that the server does not have in its database, or packet loss on the network. DNS is a UDP-based protocol, so packets that are lost will not be retransmitted, causing DNS to fail.

Resolution: Check the client configuration to see which DNS server it is setup to use. If this is the wrong server, adjust this setting in the client or in the DHCP server which provided it. Test the DNS server from the client connection repeatedly to determine if there is delay in response due to packet loss. If packets are lost, look for Ethernet errors between the client and server. Capture failing DNS requests to determine if there is any response at all from the server. Ideally, setup a tool that will constantly test the DNS server and alert when a problem occurs.

Symptom: If the client PC is able to link to the network, performance may be poor. The PC may not be able to connect at all.

Cause: In networks today, Gigabit to the desktop is common. Gigabit requires four pairs of cable, so anything lower than Category 5 will not work for Gig. In older buildings this must be taken into consideration. In addition, any amount of untwisting of the cable (often near the RJ-45 termination or patch panel) can cause signal loss. This will result in FCS Errors on the switch ports or network adapters.

Resolution: In most cases with cable problems, the cable simply needs to be replaced. If the problem is because the cable has become untwisted, re-terminating the cable may resolve the errors. When supporting new technologies such as Gigabit or Power over Ethernet, the cable must be Cat 5 or better.