Is it possible to add a hook which identifies if the plugin is installed (for self-hosted) and loads it on the dashboard? In addition those preloaded queries would be great to include the queries for those graphs
Important to note that I considered doing that while writing it, but couldn’t come up with a good list of what to pre-load.
What I am realy missing for my self hosted instance are some hardware information, like memory, diskspace and cpu usage. That would be awesome.
All good, that’s my area of expertise. Although TBH we used to measure pretty different things only a couple of years ago. The things we used to KPI ourselves on are now considered vanity metrics.
That’s outside of our scope I’m afraid. We don’t have a reliable way of providing that information across all instances.
That’s where monitoring with Icinga or Prometheus kicks in. You’d also want to ensure that Docker is running, and many other things in the background. I do understand that CPU, disk usage and memory is more easy to find out about but always consider that the application is running inside a container. Which runs isolated and may not necessarily fetch host system information that easy.
Here’s an example of which additional services I do get notifications for. One may consider adding such as a custom URL into the Discourse dashboard, maybe.
Maybe one could collect some metrics from inside the running container, I am not that deep into the Discourse backend yet. Would be interesting though, also to have them exported via the Prometheus exporter plugin.
Not what you’re asking for, but Digital Ocean will show that on their dashboard.
That would recommend using Digital Ocean as hoster, that no solution for me.
What I dont want is complex monitoring, just the possibility to take a quick look and see that my system is or will not out of ressources. But as it is not at the scope of discourse, its ok for me. Was just an idea.
I like that idea, some sort of self-monitoring. Do you know Graylog (for logs and event collecting) … this also has a memory usage indicator. Differs from the stack though, Java has access to these metrics. I’m toying with the idea to look what’s possible with Ruby here, but it will take some time.
I’ve seen that this is now in Git master (experimental, cannot get it to work in prod right now).
Since I’ve also read about PostgreSQL 10 as backend, I was wondering if the inner components versions could also be displayed in the admin dashboard. Like PostgreSQL 10.x.y, Redis 4.x, Ruby 2.5.0 and so on. This would allow to identify specific build versions better, and help with troubleshooting. Although I do understand that this requires queries to many backends, and likely should be cached somewhere in the database during upgrade/installation. Just an idea