@Jan-Macenka yeah, there's some bug in the fs module we use . We are replacing it next release, so maybe this bug goes away.

Blocking /metrics and /health endpoints for Apps by default

I'd also like to track Jellyfin which I love to run as an cloudron App. It comes with pre installed Prometheus endpoint according to the documentation which I'd like to use. The default Cloudron-location for this file is /app/data/jellyfin/config/system.xml
However I dont want to expose this endpoint to the public Internet. Any Ideas how I could prevent this?

If we find a good solution, I think it would be best to have the /metrics and /health endpoints blocked by default because Cloudron-Admins most likely dont want to expose these to the Internet anyway.

@girish would you have an Idea where I could apply such a config change?

Best regards and thanks for the answer in advance 🙂

PS: Will gladly contribute to the Documentation as we found a good solution. I just fell down the rabbit hole to a certain degree in advancing my home-lab more towards a real data-center 🙂

@Jan-Macenka said in Prometheus + Grafana actually working:

just had the same issue and wanted to add something here as a documentation:

If both Grafana and Prometheus are running as a cloudron-app, you can add Prometheus as a data-source in Grafana by inputting http://<PROMETHEUS-SERVICE-UUID>:9090 in the URL-field where PROMETHEUS-SERVICE-UUID can be found form the prometheus container either by looking up its installation location in the Storage-Tab or by connecting via terminal and copying the hostname.

Cheers,
Jan

@staff sounds like something that belongs in the docs 🙂

Thanks @girish and sorry for the late reply

@cortex said in Adding local targets to Prometheus:

How do I add the other Cloudron apps to the targets to monitor them?

I don't have much idea here, maybe @eriktad or @BrutalBirdie knows. In general, what is needed to add a target to Prometheus? IP Address?

@brutalbirdie thanks, I merged it and made a new package release.

@fbartels it works, thanks!

@christophermag This should be fixed now.