UNIX Monitor anomalies
The UnixMonitor collector allows to insert the following types of anomalies in an Anomalies Profile:
CPU - %cpu total
%cpu total is the sum of cpu user, cpu system and cpu waiting.
CPU - %cpu Wait I/O
System - Processes per CPU
A high value can be the result of a disk-intensive workload or can indicate that the disk subsystem might not be balanced properly.
System - Processes in Wait Queue
If the Processes per CPU is greater than 1, it indicates that there is at least one processes waiting for CPU. A high number of processes per CPU value will slow down system.
Processes waiting for IO or suspended.
System - Context Switches
CPU context switches.
Paging - Paging in
Paging - Pages Scanned
The number of pages paged in from paging space, frequent and excessive paging is usually accompanied by poor system performance.
Disk - Disk Use
Pages scanned by clock algorithm per second. If the scan rate is over 200 pages/second, it can indicate that the system is short of memory.
Disk - Disk Service Time
Percent of time a disk is active. If it exceeds 70%, processes will wait for IO to complete.
Disk Service Time in milliseconds. A value greater than 30 ms can indicate a performance problem.
Disk - Queue Length
Number of transactions waiting for service.
Network - Input Errors Rate
Percentage of Input Errors to Input Packets. If this rate is > 1%, it can indicate a lack of memory (use 'netstat -m' for more information).
Network - Output Errors Rate
Percentage of Output Errors to Output Packets. If this rate is > 1%, it can indicate that the queue size is too small.
Network - Collisions Rate
Percentage of Collisions to Output Packets. If this rate is > 10%, it indicate a high network utilization.
Copyright © AgileLoad. All rights reserved.