An insight into interpretation of the Load average graph and metrics on the Smoothwall.
The load average is shown on both the dashboard page and on the Summary Report under the Processor section. The dashboard page is dynamic and will change when the load does. The summary report is static.
The load average graph on the Reports > Reports > Summary page shows an abstract representation of the system workload (Processor section). It is not just a measure of CPU load but rather an abstract that tries to give a general insight into how well the system is coping with the workload.
The load average is an abstract value and uses CPU load, memory load and disk load to create the load average number. This means that a high load does not necessarily mean that the CPU is busy. Often disk load is the main reason for high load average due to web filter logging and reporting database index creation.
System Resource Monitor
The System load average warning level of this monitor will send three numbers. Those numbers are the average sum of the number of processes waiting in the run-queue, plus the number currently executing, over 1, 5, and 15-minute time periods.
A more detailed explanation can be found here: http://www.linuxjournal.com/article/9001
The rule of thumb is that if the load average is consistently significantly higher than the number of cores in the system, the system can be stressed for resources. If the load average is at the same number of cores or close, then the system is doing fine. For example, a dual core system with a load average of 0.5 doesn't have any processes waiting and is handling everything with no delays. A dual core system with a load average of 10 (8 higher than the number of cores) is being taxed the average should show if this is just a spike (in which case it's OK) or a continuous load.
However, as mentioned before, high load can be caused by disk usage due to logging and reporting generating disk IO load. In order to determine if the CPU or disk is causing the high load, the "top" command on the command line will be able to show more information.
Log in to the command line and issue the command:
Once running, top will display an output like this (Yellow highlight added):
The highlighted line shows the resource usage and the most relevant ones are:
us - User processes. Smoothwall engines, like guardian. Mainly CPU usage.
sy - System processes. Kernel and hardware control. Mainly CPU usage.
wa - Wait states. Disk and memory usage indicator. Wait states shows the system waiting to retrieve or write data to disk or memory.
When a high load average is seen, a high "us" and "sy" value indicates high CPU usage, while a high "wa" value indicates high disk load.
NOTE: The system resource monitor alert has parameters that can be set in "Reports - Alerts - Alert settings" - use this to adjust the notification to only show when a load that is relevant to the hardware is reached.