Monitoring

The monitor module allows the eBox administrator to know the state of the resources of the eBox machine. This information is essential to both troubleshoot and plan in advance the necessary resources.

Monitoring implies knowing how to interpret some system values in order to decide if these values fall into an expected range, or otherwise, they are too high or too low. The main issue of monitoring is the selection of these ranges. As every machine can have different values depending on the kind of use. For example, in a file sharing server, the free storage space is a very important value that can change very quickly. However, in a router with an enabled content filter, free memory and CPU load are more interesting values. You should avoid fetching values that are useless for your scenario.

This is the reason why eBox monitors only a few system metrics in its current version. These are: system load, CPU usage, memory usage, and file system usage.

The monitor module displays the fetched data using graphs. This allows the user to easily visualize the evolution of the resources during time. To access these graphs you have to click on the menu entry labeled as Monitor. You can place the mouse pointer over any graph point to know the exact value at that point.

You can see different time scales of the registered data: hourly, daily, monthly or yearly. You just need to click on the relevant tab.

Metrics

System load

The system load tries to measure the rate of pending work over the completed work. This value is computed using the number of active process in the CPU.

This metric is the capacity of the used CPU over a given time. This means that a load of 1 represents a CPU working at full capacity. A load of 0.5 means that the CPU could take twice as much. Conversely, a load of 2 means that it would need another CPU to fullfill the requirements of the current work load.

You have to take into account that those processes that are waiting for read/write operations in disk also contribute to this value.

CPU usage

This graph shows detailed information of the CPU usage. In the case of having a multi-core or multi-cpu machine you will see one graph for each one of the cores.

This graph represents the amount of time that the CPU spends in each of its states: running user code, system code, inactive, input/output wait, and so on. This measure is not a percentage, but scheduling units known as jiffies. In most Linux systems this value is 100 per second, but it can be different.

Memory usage

This graphs shows the memory usage. The following variables are monitored:

Free memory:

Amount of memory not used

Page cache:

Amount of memory that is cached in disk swap

Buffer cache:

Amount of memory that is cached for input/output operations

Memory used:

Amount of memory that is not included in any of the above

File system usage

This graph displays the used and free space of every mounting point.

Temperature

This graph allows you to know the system temperature in degrees Celsius by using the ACPI system [1]. You need to have data available in these directories: /sys/class/thermal or /proc/acpi/thermal_zone.

[1]	Advanced Configuration and Power Interface (ACPI) is an open standard to configure devices focused on operating systems and power management. http://www.acpi.info/

Alerts

These graphs are not very helpful if in case of unexpected behaviour the administrator is not properly notified. By using alerts, you can know when the machine has reached an unusual system load or is approaching its full capacity.

You can configure monitor alerts in Events ‣ Configure Events. The relevant alert is called monitor.

You can access the configuration page by clicking on the configuration cell. In this page you can pick any monitored metric and set the threshold that will trigger an event.

There are two different thresholds, warning and failure, this allows the user to filter based on the event severity. You can use the option reverse: to swap the values that are considered right and wrong. Other important option is persistent:. Depending on the metric we can also set other parameters.

Once you have configured and enabled the event you will need to configure, at least, one observer. The observer configuration is the same as the configuration of any other event. Check the Events and alerts chapter for further information.