Skip to main content

Agent alert notifications

This is reference documentation for Netdata's Agent alert notification feature, which supports dozens of endpoints, user roles, and more.

The script to execute on alarm line in netdata.conf defines the external script that will be called once the alert is triggered.

The default script is alarm-notify.sh.

Info

This file mentions editing configuration files.

  • To edit configuration files safely, we provide the edit config scriptlocated in your Netdata config directory (typically is /etc/netdata) that creates the proper file and opens it in an editor automatically.
    Note that to run the script you need to be inside your Netdata config directory.

  • Please also note that after most configuration changes, you will need to restart the Agent for the changes to take effect.

It is recommended to use this way for configuring Netdata.

You can change the default script globally by editing netdata.conf and changing the script to execute on alarm in the [health] section.

alarm-notify.sh is capable of sending notifications:

  • to multiple recipients
  • using multiple notification methods
  • filtering severity per recipient

It uses roles. For example sysadmin, webmaster, dba, etc.

Each alert is assigned to one or more roles, using the to line of the alert configuration. For example, here is the alert configuration for ram.conf that defaults to the role sysadmin:

    alarm: ram_in_use
on: system.ram
class: Utilization
type: System
component: Memory
os: linux
hosts: *
calc: $used * 100 / ($used + $cached + $free + $buffers)
units: %
every: 10s
warn: $this > (($status >= $WARNING) ? (80) : (90))
crit: $this > (($status == $CRITICAL) ? (90) : (98))
delay: down 15m multiplier 1.5 max 1h
info: system memory utilization
to: sysadmin

Then alarm-notify.sh uses its own configuration file health_alarm_notify.conf, which at the bottom of the file stores the recipients per role, for all notification methods.

Here is an example, of the sysadmin's role recipients for the email notification.
You can send the notification to multiple recipients by separating the emails with a space.


###############################################################################
# RECIPIENTS PER ROLE

# -----------------------------------------------------------------------------
# generic system alerts
# CPU, disks, network interfaces, entropy, etc

role_recipients_email[sysadmin]="[email protected] [email protected]"

Each role may have one or more destinations and one or more notification methods.

So, for example, the sysadmin role may send:

  1. emails to [email protected] and [email protected]
  2. pushover.net notifications to USERTOKENS A, B and C.
  3. pushbullet.com push notifications to [email protected] and [email protected]
  4. messages to the #alerts and #systems channels of a Slack workspace.
  5. messages to Discord channels #alerts and #systems.

Configuration

You can edit health_alarm_notify.conf using the edit-config script to configure:

  • Settings per notification method:

    All notification methods, except email, require some configuration (i.e., API keys, tokens, destination rooms, channels, etc.). Please check this section's content to find the configuration guides for your notification option of choice

  • Recipients per role per notification method

    role_recipients_email[sysadmin]="${DEFAULT_RECIPIENT_EMAIL}"
    role_recipients_pushover[sysadmin]="${DEFAULT_RECIPIENT_PUSHOVER}"
    role_recipients_pushbullet[sysadmin]="${DEFAULT_RECIPIENT_PUSHBULLET}"
    role_recipients_telegram[sysadmin]="${DEFAULT_RECIPIENT_TELEGRAM}"
    role_recipients_slack[sysadmin]="${DEFAULT_RECIPIENT_SLACK}"
    ...

    Here you can change the ${DEFAULT_...} values to the values of the recipients you want, separated by a space if you have multiple recipients.

Testing Alert Notifications

You can run the following command by hand, to test alerts configuration:

# become user netdata
sudo su -s /bin/bash netdata

# enable debugging info on the console
export NETDATA_ALARM_NOTIFY_DEBUG=1

# send test alerts to sysadmin
/usr/libexec/netdata/plugins.d/alarm-notify.sh test

# send test alerts to any role
/usr/libexec/netdata/plugins.d/alarm-notify.sh test "ROLE"

If you are running your own Registry, add export NETDATA_REGISTRY_URL=[YOUR_URL] before calling alarm-notify.sh.

If you need to dig even deeper, you can trace the execution with bash -x. Note that in test mode, alarm-notify.sh calls itself with many more arguments. So first do:

bash -x /usr/libexec/netdata/plugins.d/alarm-notify.sh test

And then look in the output for the alarm-notify.sh calls and run the one you want to trace with bash -x.

Global configuration options

Notification Filtering

When you define recipients per role for notification methods, you can append |critical to limit the notifications that are sent.

In the following examples, the first recipient receives all the alerts, while the second one receives only notifications for alerts that have at some point become critical. The second user may still receive warning and clear notifications, but only for the event that previously caused a critical alert.

 email      : "[email protected] [email protected]|critical"
pushover : "2987343...9437837 8756278...2362736|critical"
telegram : "111827421 112746832|critical"
slack : "alerts disasters|critical"
alerta : "alerts disasters|critical"
flock : "alerts disasters|critical"
discord : "alerts disasters|critical"
twilio : "+15555555555 +17777777777|critical"
messagebird: "+15555555555 +17777777777|critical"
kavenegar : "09155555555 09177777777|critical"
pd : "<pd_service_key_1> <pd_service_key_2>|critical"
irc : "<irc_channel_1> <irc_channel_2>|critical"

If a per role recipient is set to an empty string, the default recipient of the given notification method (email, pushover, telegram, slack, alerta, etc.) will be used.

To disable a notification, use the recipient called: disabled This works for all notification methods (including the default recipients).

Proxy configuration

If you need to send curl-based notifications (pushover, pushbullet, slack, alerta, flock, discord, telegram) via a proxy, you should set these variables to your proxy address:

export http_proxy="http://10.0.0.1:3128/"
export https_proxy="http://10.0.0.1:3128/"

Notification images

Images in notifications need to be downloaded from an Internet facing site.

To allow notification providers to fetch the icons/images, by default we set the URL of the global public Netdata Registry.

If you have an Internet facing netdata (or you have copied the images/ folder of Netdata to your web server), set its URL here, to fetch the notification images from it.

images_base_url="http://my.public.netdata.server:19999"

Date handling

You can configure netdata alerts to send dates in any format you want via editing the date_format variable.

This uses standard date command format strings. See man date for more info on what formats are supported.

Note that this has to start with a '+'; otherwise it won't work.

  • For ISO 8601 dates, use +%FT%T%z
  • For RFC 5322 dates, use +%a, %d %b %Y %H:%M:%S %z
  • For RFC 3339 dates, use +%F %T%:z
  • For RFC 1123 dates, use +%a, %d %b %Y %H:%M:%S %Z
  • For RFC 1036 dates, use +%A, %d-%b-%y %H:%M:%S %Z
  • For a reasonably local date and time (in that order), use +%x %X
  • For the old default behavior (compatible with ANSI C's asctime() function), leave the date_format field empty.

Hostname handling

By default, Netdata will use the simple hostname for the system (the hostname with everything after the first . removed) when displaying the hostname in alert notifications.

If you instead prefer to have Netdata use the host's fully qualified domain name, you can set use_fdqn to YES.

This setting does not account for child systems for which the system you are configuring is a parent.

Note

If the system's host name is overridden in /etc/netdata.conf with the hostname option, that name will be used unconditionally.


Do you have any feedback for this page? If so, you can open a new issue on our netdata/learn repository.