NetWare's Utilization calculation is based on the amount of time spent polling the Run Queue. The algorithm is based upon a Maximum versus Current ratio, i.e., the maximum amount of time spent polling the process queue during any one second period since the Server was brought up; versus the amount of time spent polling in the last second. The more time NetWare spends polling, the lower the utilization, i.e., 100% polling activity equals 0% utilization.

Individual Process activity can be viewed using the Processor Utilization options within NetWare's MONITOR NLM (a standard menu selection under NetWare v4, optionally activated via the -P command line option when loading the MONITOR NLM on a NetWare v3 Server.) Specific levels to watch for are high Time values with low Counts. Counts indicate the number of times that the thread relinquished control to other processes. The higher the Time and the lower the Count, the more ill-behaved the process.

When an ABEND message indicates that the Interrupted Process is the Polling Process, it is rarely the internal Polling Process logic that is at fault but rather some other NLM process which is dependent upon the Polling Process for its service activity. If the Running Process is an Interrupt Service Routine, chances are great that it is a disk or LAN driver at fault (the higher the nested count, the greater the probability that it is LAN I/O related).

You can debug a Polling/Idle Loop Process ABEND similar to a GPPE ABEND (refer to Abnormal End (ABEND) and Debugger Information or Diagnosing an ABEND for more information).