Deleted monitors still show in /metrics

[hippsabq]

Is it a duplicate question?
I looked and did not find anything

Describe the bug
If you remove a monitor from the dashboard, the monitor and its last status still show up in the metrics page. This also occurs if you modify the monitor_url. Changing the monitor url also creates a redundant line in the metrics output.

I am scraping these into Prometheus and using the Grafana Dashboard, but all deleted or modified monitors will show up and remain in my dashboard.

To Reproduce
Steps to reproduce the behavior:
the example I use in the screenshot is Plex.

1. I first setup a Plex monitor to look at the "ipaddress:port" of my plex server
2. The monitor came back down since the plex web ui is under "ipaddress:port/web/"
3. I modified the Plex monitor to include the "/web" ending
4. Navigated to /metrics of my uptime-kuma server
5. Noticed that a duplicate "Plex" monitor was created and the first incorrect one was still there showing down
6. Deleted the "Plex" monitor altogether
7. Navigated to /metrics of my uptime-kuma server
8. Noticed that both "Plex" monitors were still output to the /metrics

Expected behavior

1. Modifications to a monitor do not create a duplicate /metrics output
2. Deleted monitors no longer show up in /metrics output

Info
Uptime Kuma Version: 1.7.3
Using Docker?: Yes
Docker Version: 20.10.9
Node.js Version (Without Docker only):
OS: Win 10
Browser: Brave

Screenshots

Error Log

[deefdragon]

This appears to also happen if you change the name of the monitor.

[RisedSky]

This appears to also happen if you change the name of the monitor.

Seems like the URL is changed, maybe it duplicates because of it ?

[louislam]

https://github.com/louislam/uptime-kuma/blob/master/server/prometheus.js

It seems that there is no "remove logic" in the current implement. May need to remove the monitor from all gauges if the monitor is stopped.

[x4FF3]

as i am not familiar with JS, is this going to be fixed anytime soon?

[gruessung]

+1

[Coliinnn]

Also experiencing this issue. A fix would be great, unfortunately I am also not experienced enough with JS. Alternatively, does anyone know of a way to filter out a specific monitor from within Prometheus or Grafana?

[chakflying]

Took a shot at it in #1136, please check.

[Proxymiity]

A quick and easy hack is to restart the container. Restarting seems to have cleared every duplicated monitor.
It's as easy as docker-compose restart or docker restart uptime-kuma :)

[kyuuk]

Little workaround:
To clean monitors that have been deleted/renamed i discovered that you need to clear all statistics from database in the settings, like this

[nmcc1212]

Still an issue, and clearing all statistics didn't work

[louislam]

#1136

May need someone to test this pr, as I dont using Prometheus.

[nmcc1212]

i am testing it with now, but you don't need prometheus to see if it works, just change/delete an endpoint and check uptime-kuma:3001/metrics and see if the value still exists/ has changed

[nmcc1212]

just tested it with prometheus, it works well and is pretty much instant (less than 1 second delay between edit/delete and updating prometheus endpoint