Configure health alarms

Netdata's health watchdog is highly configurable, with support for dynamic thresholds, hysteresis, alarm templates, and more. You can tweak any of the existing alarms based on your infrastructure's topology or specific monitoring needs, or create new entities.

You can use health alarms in conjunction with any of Netdata's collectors (see the supported collector list) to monitor the health of your systems, containers, and applications in real time.

While you can see active alarms both on the local dashboard and Netdata Cloud, all health alarms are configured per node via individual Netdata Agents. If you want to deploy a new alarm across your infrastructure, you must configure each node with the same health configuration files.

Edit health configuration files#

All of Netdata's health configuration files are in Netdata's config directory, inside the health.d/ directory. Navigate to your Netdata config directory and use edit-config to make changes to any of these files.

For example, to edit the cpu.conf health configuration file, run:

sudo ./edit-config health.d/cpu.conf

Each health configuration file contains one or more health entities, which always begin with alarm: or template:. For example, here is the first health entity in health.d/cpu.conf:

template: 10min_cpu_usage
on: system.cpu
os: linux
hosts: *
lookup: average -10m unaligned of user,system,softirq,irq,guest
units: %
every: 1m
warn: $this > (($status >= $WARNING) ? (75) : (85))
crit: $this > (($status == $CRITICAL) ? (85) : (95))
delay: down 15m multiplier 1.5 max 1h
info: average cpu utilization for the last 10 minutes (excluding iowait, nice and steal)
to: sysadmin

To tune this alarm to trigger warning and critical alarms at a lower CPU utilization, change the warn and crit lines to the values of your choosing. For example:

warn: $this > (($status >= $WARNING) ? (60) : (75))
crit: $this > (($status == $CRITICAL) ? (75) : (85))

Save the file and reload Netdata's health configuration to make your changes live.

Silence an individual alarm#

Instead of disabling an alarm altogether, or even disabling all alarms, you can silence individual alarms by changing one line in a given health entity. To silence any single alarm, change the to: line in its entity to silent.

to: silent

Write a new health entity#

While tuning existing alarms may work in some cases, you may need to write entirely new health entities based on how your systems, containers, and applications work.

Read Netdata's health reference for a full listing of the format, syntax, and functionality of health entities.

To write a new health entity into a new file, navigate to your Netdata config directory, then use touch to create a new file in the health.d/ directory. Use edit-config to start editing the file.

As an example, let's create a ram-usage.conf file.

sudo touch health.d/ram-usage.conf
sudo ./edit-config health.d/ram-usage.conf

For example, here is a health entity that triggers a warning alarm when a node's RAM usage rises above 80%, and a critical alarm above 90%:

alarm: ram_usage
on: system.ram
lookup: average -1m percentage of used
units: %
every: 1m
warn: $this > 80
crit: $this > 90
info: The percentage of RAM being used by the system.

Let's look into each of the lines to see how they create a working health entity.

  • alarm: The name for your new entity. The name needs to follow these requirements:
    • Any alphabet letter or number.
    • The symbols . and _.
    • Cannot be chart name, dimension name, family name, or chart variable names.
  • on: Which chart the entity listens to.
  • lookup: Which metrics the alarm monitors, the duration of time to monitor, and how to process the metrics into a usable format.
    • average: Calculate the average of all the metrics collected.
    • -1m: Use metrics from 1 minute ago until now to calculate that average.
    • percentage: Clarify that we're calculating a percentage of RAM usage.
    • of used: Specify which dimension (used) on the system.ram chart you want to monitor with this entity.
  • units: Use percentages rather than absolute units.
  • every: How often to perform the lookup calculation to decide whether or not to trigger this alarm.
  • warn/crit: The value at which Netdata should trigger a warning or critical alarm. This example uses simple syntax, but most pre-configured health entities use hysteresis to avoid superfluous notifications.
  • info: A description of the alarm, which will appear in the dashboard and notifications.

In human-readable format:

This health entity, named ram_usage, watches the system.ram chart. It looks up the last 1 minute of metrics from the used dimension and calculates the average of all those metrics in a percentage format, using a % unit. The entity performs this lookup every minute.

If the average RAM usage percentage over the last 1 minute is more than 80%, the entity triggers a warning alarm. If the usage is more than 90%, the entity triggers a critical alarm.

When you finish writing this new health entity, reload Netdata's health configuration to see it live on the local dashboard or Netdata Cloud.

Reload health configuration#

To make any changes to your health configuration live, you must reload Netdata's health monitoring system. To do that without restarting all of Netdata, run netdatacli reload-health or killall -USR2 netdata.

What's next?#

With your health entities configured properly, it's time to enable notifications to get notified whenever a node reaches a warning or critical state.

To build complex, dynamic alarms, read our guide on dimension templates.

Last updated on

Monitor everything in real time โ€“ for free

Troubleshoot slowdowns and anomalies in your infrastructure with thousands of per-second metrics, meaningful visualizations, and insightful health alarms with zero configuration.

Get Netdata