Hi all:
I feel that LibreNMS poses a rather high barrier to entry. I wouldn’t underestimate ease of installation as a factor for success.
Let me tell you about my experience trying to get started. I have a distant brother-in-law who knows a friend that runs a small business. There is also this nephew who helps in a small charity (a local volunteer library). And so on. Anyway, they all have the odd server and a few clients. There is also a printer. Maybe a consumer-level NAS.
When something goes wrong, they call me and complain that the printer isn’t printing. The paper just does not come out printed, as it should. Yes, there was some error message, but he/she did not write it down. And by the way, a hard disk suddenly died. Or is full. Or whatever.
So I thought I should just install a small network monitoring system that sends me an e-mail whenever the printer is not reachable anymore. Or when S.M.A.R.T. warns of an imminent disk failure. Etc.
I am no sysadmin, so I just want an easy, “set and forget” monitoring system that sends me alert e-mails. In this day and age of AI and self-driving cars, how hard can it be???
So I heard of LibreNMS, and then I looked at the installation instructions. PHP dependencies, file permissions, MariaDB, Web server configuration, cron logrotate… Mmmm, I don’t think that’s for me. I’m actually in a hurry. 
So I found a VM image:
But it’s in OVA format, built with VirtualBox. I only know the “standard” libvirt/KVM/QEMU that comes with Ubuntu. There is no easy, automatic conversion from the OVA format to libvirt/KVM/QEMU.
And a complete VM is not that easy to keep up to date. It’s actually a separate system to maintain.
Besides, there is no mention that the VM comes with any ready-to-use e-mail notification feature. So I would have to research how to install and integrate it. Did I mention that I am on a hurry???
Then I found a LibreNMS Docker image:
A Docker instance would probably be easier to manage. I have used the odd container instance and it’s a reasonable system: you place all your configuration information into a separate volume, and then you can replace the whole Docker image with the next version and re-use your old configuration. There is no extra Linux system to manage.
But it turns out that the official LibreNMS project provides no single Docker image after all. It’s a whole collection of them with “sidecar” containers. That is why “Docker Compose” is the recommended way to run them. I definitely do not want to learn how to manage multiple, interdependent Docker containers for such a small monitoring project.
The main LibreNMS container does not even have a database server, which is always required. And it looks like you need a separate container just to send notification e-mails. This is definitely not something for the mere mortal.
May I suggest that the LibreNMS project provides a single Docker image suitable for a small organisation?
If you have a few computers to watch, you do not need a “dispatcher” with “distributed polling” or anything fancy.
A “reverse proxy” only adds complexity. “Let’s Encrypt certificates” is probably something for servers exposed to the Internet, which most people will not want anyway.
Memcached or any other caching “for better scalability” is just silly for something that a 10-year-old laptop should handle as a side task. An OPCache is likely over the top too.
However, the ability to send an alert e-mail is a must and should be integrated in the one and only Docker image.
It does not really matter if the single Docker image has all features, hopefully disabled by default. In this day and age, it is OK if you need a few gigabytes more of wasted disk space. The vast majority of small organisations do not need to optimise the individual sizes of a swarm of virtual servers. But the Docker image needs to be easy to start with.
Regards,
rdiez