8.7 KiB

Raw Blame History

Examples

There are a million ways to use ntfy, but here are some inspirations. I try to collect examples on GitHub, so be sure to check those out, too.

A long process is done: backups, copying data, pipelines, ...

I started adding notifications pretty much all of my scripts. Typically, I just chain the curl call directly to the command I'm running. The following example will either send Laptop backup succeeded or ⚠️ Laptop backup failed directly to my phone:

rsync -a root@laptop /backups/laptop \
  && zfs snapshot ... \
  && curl -H prio:low -d "Laptop backup succeeded" ntfy.sh/backups \
  || curl -H tags:warning -H prio:high -d "Laptop backup failed" ntfy.sh/backups

Low disk space alerts

Here's a simple cronjob that I use to alert me when the disk space on the root disk is running low. It's simple, but effective.

#!/bin/bash

mingigs=10
avail=$(df | awk '$6 == "/" && $4 < '$mingigs' * 1024*1024 { print $4/1024/1024 }')
topicurl=https://ntfy.sh/mytopic

if [ -n "$avail" ]; then
  curl \
    -d "Only $avail GB available on the root disk. Better clean that up." \
    -H "Title: Low disk space alert on $(hostname)" \
    -H "Priority: high" \
    -H "Tags: warning,cd" \
    $topicurl
fi

Server-sent messages in your web app

Just as you can subscribe to topics in the Web UI, you can use ntfy in your own web application. Check out the live example or just look the source of this page.

Years ago my home server was broken into. That shook me hard, so every time someone logs into any machine that I own, I now message myself. Here's an example of how to use PAM to notify yourself on SSH login.

=== "/etc/pam.d/sshd" # at the end of the file session optional pam_exec.so /usr/bin/ntfy-ssh-login.sh

=== "/usr/bin/ntfy-ssh-login.sh" bash #!/bin/bash if [ "${PAM_TYPE}" = "open_session" ]; then curl \ -H prio:high \ -H tags:warning \ -d "SSH login: ${PAM_USER} from ${PAM_RHOST}" \ ntfy.sh/alerts fi

Collect data from multiple machines

The other day I was running tasks on 20 servers, and I wanted to collect the interim results as a CSV in one place. Each of the servers was publishing to a topic as the results completed (publish-result.sh), and I had one central collector to grab the results as they came in (collect-results.sh).

It looked something like this:

=== "collect-results.sh" bash while read result; do [ -n "$result" ] && echo "$result" >> results.csv done < <(stdbuf -i0 -o0 curl -s ntfy.sh/results/raw) === "publish-result.sh" ```bash // This script was run on each of the 20 servers. It was doing heavy processing ...

// Publish script results
curl -d "$(hostname),$count,$time" ntfy.sh/results
```

Ansible, Salt and Puppet

You can easily integrate ntfy into Ansible, Salt, or Puppet to notify you when runs are done or are highstated. One of my co-workers uses the following Ansible task to let him know when things are done:

- name: Send ntfy.sh update
  uri:
    url: "https://ntfy.sh/{{ ntfy_channel }}"
    method: POST
    body: "{{ inventory_hostname }} reseeding complete"

Watchtower notifications (shoutrrr)

You can use shoutrrr generic webhook support to send watchtower notifications to your ntfy topic.

Example docker-compose.yml:

services:
  watchtower:
    image: containrrr/watchtower
    environment:
      - WATCHTOWER_NOTIFICATIONS=shoutrrr
      - WATCHTOWER_NOTIFICATION_URL=generic+https://ntfy.sh/my_watchtower_topic?title=WatchtowerUpdates

Or, if you only want to send notifications using shoutrrr:

shoutrrr send -u "generic+https://ntfy.sh/my_watchtower_topic?title=WatchtowerUpdates" -m "testMessage"

Random cronjobs

Alright, here's one for the history books. I desperately want the github.com/ntfy organization, but all my tickets with GitHub have been hopeless. In case it ever becomes available, I want to know immediately.

# Check github/ntfy user
*/6 * * * * if curl -s https://api.github.com/users/ntfy | grep "Not Found"; then curl -d "github.com/ntfy is available" -H "Tags: tada" -H "Prio: high" ntfy.sh/my-alerts; fi
~

Download notifications (Sonarr, Radarr, Lidarr, Readarr, Prowlarr, SABnzbd)

It's possible to use custom scripts for all the *arr services, plus SABnzbd. Notifications for downloads, warnings, grabs etc. Some simple bash scripts to achieve this are kindly provided in nickexyz's repository.

Node-RED

You can use the HTTP request node to send messages with Node-RED, some examples:

Example: Send a message

[
  {
    "id": "8f09d37dd5773f88",
    "type": "http request",
    "z": "ff3ad4e1.d3415",
    "name": "ntfy",
    "method": "POST",
    "ret": "txt",
    "paytoqs": "ignore",
    "url": "https://example.com/topic",
    "tls": "",
    "persist": false,
    "proxy": "",
    "authType": "",
    "senderr": false,
    "credentials": {},
    "x": 1410,
    "y": 740,
    "wires": [
      []
    ]
  },
  {
    "id": "2603f296b25fe351",
    "type": "function",
    "z": "ff3ad4e1.d3415",
    "name": "data",
    "func": "msg.payload = \"Something happened\";\nmsg.headers = {};\nmsg.headers['tags'] = 'house';\nmsg.headers['X-Title'] = 'Home Assistant';\n\nreturn msg;",
    "outputs": 1,
    "noerr": 0,
    "initialize": "",
    "finalize": "",
    "libs": [],
    "x": 1290,
    "y": 740,
    "wires": [
      [
        "8f09d37dd5773f88"
      ]
    ]
  },
  {
    "id": "d2351ed0720a239f",
    "type": "inject",
    "z": "ff3ad4e1.d3415",
    "name": "Manual start",
    "props": [
      {
        "p": "payload"
      },
      {
        "p": "topic",
        "vt": "str"
      }
    ],
    "repeat": "",
    "crontab": "",
    "once": false,
    "onceDelay": "20",
    "topic": "",
    "payload": "",
    "payloadType": "date",
    "x": 1150,
    "y": 740,
    "wires": [
      [
        "2603f296b25fe351"
      ]
    ]
  }
]

Example: Send a picture

[
  {
    "id": "726d0d75d6c0f70e",
    "type": "http request",
    "z": "ff3ad4e1.d3415",
    "name": "Download jpeg",
    "method": "GET",
    "ret": "bin",
    "paytoqs": "ignore",
    "url": "https://www.google.com/images/branding/googlelogo/1x/googlelogo_color_272x92dp.png",
    "tls": "",
    "persist": false,
    "proxy": "",
    "authType": "",
    "senderr": false,
    "credentials": {},
    "x": 1320,
    "y": 780,
    "wires": [
      [
        "730dbbc9dbf1ed8a"
      ]
    ]
  },
  {
    "id": "730dbbc9dbf1ed8a",
    "type": "function",
    "z": "ff3ad4e1.d3415",
    "name": "data",
    "func": "msg.payload = msg.payload;\nmsg.headers = {};\nmsg.headers['tags'] = 'house';\nmsg.headers['X-Title'] = 'Home Assistant - Picture';\n\nreturn msg;",
    "outputs": 1,
    "noerr": 0,
    "initialize": "",
    "finalize": "",
    "libs": [],
    "x": 1470,
    "y": 780,
    "wires": [
      [
        "592f424b37f76f5c"
      ]
    ]
  },
  {
    "id": "592f424b37f76f5c",
    "type": "http request",
    "z": "ff3ad4e1.d3415",
    "name": "ntfy",
    "method": "PUT",
    "ret": "bin",
    "paytoqs": "ignore",
    "url": "https://example.com/topic",
    "tls": "",
    "persist": false,
    "proxy": "",
    "authType": "",
    "senderr": false,
    "x": 1590,
    "y": 780,
    "wires": [
      []
    ]
  },
  {
    "id": "8aa06dda3c902f6a",
    "type": "inject",
    "z": "ff3ad4e1.d3415",
    "name": "Manual start",
    "props": [
      {
        "p": "payload"
      },
      {
        "p": "topic",
        "vt": "str"
      }
    ],
    "repeat": "",
    "crontab": "",
    "once": false,
    "onceDelay": "20",
    "topic": "",
    "payload": "",
    "payloadType": "date",
    "x": 1150,
    "y": 780,
    "wires": [
      [
        "726d0d75d6c0f70e"
      ]
    ]
  }
]

Gatus service health check

An example for a custom alert with Gatus

alerting:
  custom:
    url: "https://ntfy.sh"
    method: "POST"
    body: |
      {
        "topic": "mytopic",
        "message": "[ENDPOINT_NAME] - [ALERT_DESCRIPTION]",
        "title": "Gatus",
        "tags": ["[ALERT_TRIGGERED_OR_RESOLVED]"],
        "priority": 3
      }
    default-alert:
      enabled: true
      description: "health check failed"
      send-on-resolved: true
      failure-threshold: 3
      success-threshold: 3
    placeholders:
      ALERT_TRIGGERED_OR_RESOLVED:
        TRIGGERED: "warning"
        RESOLVED: "white_check_mark"

8.7 KiB Raw Blame History