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.
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
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" \
$topicurl
fi
Just as you can subscribe to topics in the Web UI, you can use ntfy in your own web application. Check out the live example or just look the source of this page.
Years ago my home server was broken into. That shook me hard, so every time someone logs into any machine that I own, I now message myself. Here's an example of how to use PAM to notify yourself on SSH login.
=== "/etc/pam.d/sshd"
```
# at the end of the file
session optional pam_exec.so /usr/bin/ntfy-ssh-login.sh
```
=== "/usr/bin/ntfy-ssh-login.sh"
```bash
#!/bin/bash
if [ "${PAM_TYPE}" = "open_session" ]; then
curl \
-H prio:high \
-H tags:warning \
-d "SSH login: ${PAM_USER} from ${PAM_RHOST}" \
ntfy.sh/alerts
fi
```
The other day I was running tasks on 20 servers, and I wanted to collect the interim results
as a CSV in one place. Each of the servers was publishing to a topic as the results completed (publish-result.sh
),
and I had one central collector to grab the results as they came in (collect-results.sh
).
It looked something like this:
=== "collect-results.sh"
```bash
while read result; do
[ -n "$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
```
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"
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"
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
~