Monitoring the server allows you to
MONITOR is the key NetWare® tool for performance monitoring at the server level. It runs in the NetWare kernel so that it can make calls directly to the operating system. Most statistics are updated every second.
You can also use the Java-based ConsoleOne for monitoring. See Contents > Server Operating System > ConsoleOne.
[LOAD] [path] MONITOR
The LOAD command is necessary only if you have a batch file with the same name as the NLM. (Using the LOAD command indicates that MONITOR is a loadable module.)
The path is specified only if you moved the utility from its default directory sys:system to a directory that does not have a search path.
Press Alt + Esc. To cycle through other console screens, continue to hold Alt down and tap Esc until the current MONITOR screen reappears. Use the bar at the top of the screen for navigation.
Or press Ctrl + Esc. A numbered list of available console screens appears. Enter the number for the NetWare 5 Console Monitor.
Repeatedly press Esc until the Exit confirmation box appears. Then press Enter to return to the console prompt.
Or to bring up the Exit confirmation box immediately, press Alt + F10. Then press Enter to return to the console prompt.
Monitoring utilities can be overwhelming in the volume of information and statistics they provide. To simplify your access to key parameters, “MONITOR's General Information” includes most of them of one screen, including performance indicators that pertain to memory utilization and disk I/O.
See “Monitoring Allocated Services” for an explanation of the interaction of three types of configurable parameters that control dynamically allocated services. The guidelines allow you to be proactive in maintaining the level of service your users require.
“Other Information and Statistics to Monitor” include error logs, available free disk space, and users accounts. To be prepared for power supply interruptions, test your UPS (uninterruptable power system) periodically.
Becoming familiar with your server's day-to-day performance and its characteristic response to its unique load will help you to better interpret the information and statistics the monitoring utilities provide.
Most of the important performance indicators that can be used to track, diagnose, and resolve server problems appear in the “General Information Screen in MONITOR.” This window displays when you load MONITOR.
General Information Screen in MONITOR
The onscreen performance indicators are described following.
Does utilization go to 100 percent? Does it reach a plateau or does it peak for a few seconds and then drop? What time and how frequently does it peak?
Because file caching has a dramatic impact on server performance, you want this number to be as high as possible.
Monitor this number to see whether it is increasing. If it is, you might have a bottleneck in the file system.
Use this percentage to assess overall disk cache utilization. If this value falls below 90%, the disk thrashes and server performance degrades. The solution is to install more physical memory (RAM).
For another MONITOR statistic that is useful in assessing RAM, see Understanding LRU Sitting Time.
Monitor this number to see whether it is increasing. If it is, a slow hard disk might be your bottleneck.
This number is the result of the interaction of three configurable parameters. Monitor the current value to ensure that it does not reach the maximum value.
Monitor the current service processes to ensure that it does not reach this value.
Use this number to determine whether the server has enough free task handlers to service client requests. If the number of current service processes reaches the maximum, the value of the Maximum Service Processes parameter must be increased. (See MONITOR's Available Options > Server Parameters > Communications Parameters.)
There are limits to resources in every computing environment. NetWare allocates resources according to need and availability. When a request is received, the operating system doesn't immediately allocate new resources. It waits a specified amount of time to see if existing resources become available to service the demand.
Dynamically allocated services are controlled by the interaction of three types of settable parameters. If you understand how they work together, you can interpret the statistics that you monitor and know how to configure the system's response to demand. The three types of parameters that interact to control allocation are
Low minimum limits will slow the growth of a particular service. High minimum limits allow rapid growth.
If resources become available, no new resources are allocated. If they don't become available within the time limit, new resources are allocated. The time limit ensures that sudden, infrequent peaks of server activity don't permanently allocate unneeded resources.
Consider, for example, the allocation of directory cache buffers. If the minimum number of directory cache buffers is set to 20, the system allocates another buffer resource as soon as a request is made---until 20 cache buffers have been allocated.
When 20 directory cache buffers are allocated, the system waits 2.2 seconds (default) when a request comes in, and then allocates another buffer if the request is still active.
However, if the minimum number of directory cache buffers is set to 40, the system allocates 40 directory cache buffers before it starts slowing the growth by waiting 2.2 seconds after each request.
Common maintenance tasks also include monitoring: