This is a fork of @alexbakker plugin.
alertmanager-ntfy is a service that forwards Prometheus Alertmanager notifications to ntfy.sh. It is highly customizable, allowing you to dynamically configure the ntfy notification based on the properties of the alert that is being handled.
Example of receiving a Blackbox exporter notifcation on an Android device:
The primary way to configure the service is through a YAML configuration file. An example configuration is shown below.
Some of the ntfy notification options also accept a gval expression, rather than just a hardcoded value. This allows you to dynamically decide the value of certain notification properties based on the alert that the service is currently handling. For example, if an alert is resolved, you may want to show a different icon than when an alert is currently firing.
The title and description of the notification can be customized using Go's
templating engine. Refer to the definition of
the Alert
struct in format.go for the
data structure that is passed to the templates.
http:
addr: :8000
auth:
username: "alertmanager"
password: "verysecure"
ntfy:
baseurl: https://ntfy.sh
auth:
username: "admin"
password: "verysecure"
notification:
# The topic can either be a hardcoded string or a gval expression that evaluates to a string
topic: alertmanager
# Priority reference: https://docs.ntfy.sh/publish/#message-priority
# Can either be a hardcoded string or a gval expression that evaluates to a string
priority: |
status == "firing" ? "urgent" : "default"
tags:
# Tags with no condition are always included
# Emoji shortcode reference: https://docs.ntfy.sh/emojis/
- tag: "+1"
condition: status == "resolved"
- tag: rotating_light
condition: status == "firing"
templates:
title: |
{{ if eq .Status "resolved" }}Resolved: {{ end }}{{ index .Annotations "summary" }}
description: |
{{ index .Annotations "description" }}
There are a couple of command line options as well that can be used to override settings in the config file.
It is possible to split the configuration into multiple files. This allows you to to keep a separate encrypted file for all of the sensitive credential settings, while keeping the main configuration file in plain text. Simply pass a comma-separated list of filenames:
alertmanager-ntfy --configs config.yml,auth.yml
The configuration files will be merged automatically in-memory on startup of the service.
Example of a separate auth.yaml file:
http:
auth:
username: "alertmanager"
password: "verysecure"
ntfy:
auth:
username: "admin"
password: "verysecure"
Assuming you've already set up Alertmanager for your Prometheus deployment,
configuring it to send notifications to alertmanager-ntfy is fairly
straightforward. Add a webhook_configs
section to the receiver in your
Alertmanager configuration file:
webhook_configs:
- url: http://127.0.0.1:8111/hook
http_config:
basic_auth:
username: alertmanager
password: verysecure
Replace the URL and username/password with values appropriate for your environment.
For a full overview of the available configuration options, refer to the Prometheus Alertmanager documentation.
This repository comes with a NixOS module. To use it, simply add this repository as an input to your Nix flake and import the module into your NixOS configuration.
To replicate the configuration example above:
{
services.alertmanager-ntfy = {
enable = true;
settings = {
http = {
addr = "127.0.0.1:8111";
};
ntfy = {
baseurl = "https://ntfy.sh";
notification = {
topic = "alertmanager";
priority = ''
status == "firing" ? "high" : "default"
'';
tags = [
{
tag = "+1";
condition = ''status == "resolved"'';
}
{
tag = "rotating_light";
condition = ''status == "firing"'';
}
];
templates = {
title = ''{{ if eq .Status "resolved" }}Resolved: {{ end }}{{ index .Annotations "summary" }}'';
description = ''{{ index .Annotations "description" }}'';
};
};
};
};
extraSettingsPath = "/path/to/your/auth.yml";
};
}