When creating a host metric monitoring rule, you can select the following four items to be monitored:

• CPU %
• Memory %
• Swap %
• Filesystem %

These monitoring rules fall under the "metric rate monitor" category. In this category, thresholds can be specified as a percentage (%) based on the utilization rate calculated from the system metrics sent by mackerel-agent.

The detailed specifications for the four items are as follows:

• CPU %
  • Regardless of the number of cores, this monitors the overall CPU usage calculated from cpu.idle.percentage (against a maximum of 100%).
• Memory %
  • Monitors the ratio of memory.used to memory.total.
• Swap %
  • Monitors the ratio of memory.swap_used to memory.swap_total.
• Filesystem %
  • Monitors the ratio of filesystem.*.used to filesystem.*.size.
  • If there are multiple filesystems, an alert is triggered if any of them exceeds or falls below the threshold.
  • If there are filesystems that do not need to be monitored, exclude them from metric collection by following the steps below:
    • Excluding specific file systems from metric collection

Note that monitoring rules are not applied to hosts where the utilization rate of the monitored item is always 0.
For example, on systems without a swap area, the swap utilization rate is always 0, so the Swap % monitoring rule will not be applied.