Over time I have found myself (helping to) monitor and administrate a number of Yunohosts.
So far it is a mix-and-match of scripts that I fire blindly to for example apt-upgrade hosts and manual pings and check to see if things run and are up to date, or logging in to run the diagnosis and checking if certificates got updated.
There is a number of monitoring packages available, some directly on Yunohost. I remember (pre-Yunohost) trying out a few, but getting lost in options and parameters to set without actually getting useful monitoring.
Can you recommend a monitoring app that I could give a number of hostnames to check the availability of, perhaps supported by an active component on those servers? Looking though the apps in the catalog, I found:
PHP server monitor
YunoMonitor may be just what I need, I will be trying that one for sure. I hope for some reactions/experiences with the other systems!
– edit – I installed Yunomonitor on a couple of the servers, letting one monitor the others, but I have to check whether the extended readme is intended only for non-Yuno-installations or also for Yunohosts. So far I didn’t receive an email or other notification.
– edit – @frittro just mentioned Prometheus, added to the list, and removed Grafana
From my understanding, Grafana is not a monitoring app in and of itself. It is just a data visualisation tool to display the aggregated data from a monitoring app, such as Netdata or Prometheus, etc. I intend to add Grafana to my project at a later date.
This is something that I would like to know more about. Does running a monitoring app on the same device that you are monitoring add to the load? Would I be better off just installing a monitoring app on my laptop / client device, rather than on the YunoHost device itself?
Something like a lightweight agent, which just exposes the raw data, so that it can be consumed by the monitoring app itself, remotely? Yes, that is what I’m looking for as well.
Yunomonitor is quite experimental. I am the person who wrote the code of yunomonitor and i have included some checks of yunomonitor (mail checks) directly in yunohost diagnosis.
Currently, if you don’t use the SMS features of yunomonitor, i recommend a light solution just to check a service is reachable, and let all internel checks done by the yunohost itself.
Do you have that installed on a raspi, under YunoHost? I don’t see that as an option in the Application Catalog for YNH. I am trying to avoid installing anything that isn’t supported from the YunoHost-Apps repo, so that my future upgrade path isn’t encumbered with dependency issues.
Ah, I wondered about the overlap between the periodic self-diagnosis and Yunomonitor.
The servers all run stable, as long as they are not touched, so the mail twice a day is enough in case of trouble. My provider does not offer a SMS bridge, and it’s not urgent enough for me to pay extra for SMS somewhere else.
Emails from Yunomonitor have the advantage that I will receive an email besides the ‘main’ admins (my childrens classmates), without having to create a dummy forwarding account to split root@-mails over their and my email address.
Another thing I hoped monitoring would assist in, is keeping track of installations and their properties in a visual way. Mostly I hoped I’d find I had a solved problem
I’ll ponder over my actual needs, and keep monitoring using my own eyes
I forgot about it mainly because when I installed it, it was not what I needed but it gives good results.
You may ‘give it’ a log file and it will analyse it and give you a lot of graphics. I found a script that concatenates the logs of nginx and sends the full log to goaccess to get a html file. You can even have a real-time view.