I’m going to get started on testing Zabbix and see if it will help me keep on eye on my system at work. This is 3 XCP-NG hosts, 4 Truenas, 5 Windows Servers 2022, and a few Linux servers. I will probably (eventually) add in network switches if I ever get my new Extreme switches (going on 1.7 years of backorder) and they will probably be SNMP.
I’m looking at the Zabbix Appliance to get going quickly, but it says not to use this for production. Do I just need more processor, ram, and disk to bring the appliance up from meager resources to production level, or is there a different reason? It runs Alma Linux and the default they suggest is 2 processors, like 1.5gb ram and some incredibly small hard drive. 8 cores, 8GB ram, and 60gb disk is easy to give a VM.
Yes I know the passwords are all public, but I assume I can change those without breaking things. If changing the passwords is the issue, then I’ll have to build it from scratch, and probably on Debian 12 (just can’t bring myself to Alma, Rocky, and certainly not Oracle right now). I would maybe consider Suse since they forked Redhat and have committed to keeping it open, even if no longer bug for bug the same (but binary compatible) as Redhat. I just can not trust Oracle after buying and closing OpenSolaris and eventually just shutting down the Solaris project completely (they could have opened it and given it to another group).
I use Zabbix at home for my homelab and pretty much every other device on my network.
Setting it up initially wasn’t difficult at all, but only a little time consuming (an hour or 2), there are some concepts you will generally want to familiarize yourself with (templates, triggers, etc.), but if you have used monitoring software before its not too bad, I initially struggled with the concept of a template, but it does make sense.
I would recommend a “build it from scratch”… but I also understand the “it needs to run NOW”, I have my zabbix server setup on an old atom based machine with 16GB of ram and 250GB of disk space (way more than it needs). I have about 20 hosts reporting via all different methods, though most are using the zabbix agent.
The configuration of a host (a monitored device) does not take long, and there are only a few edits that need to happen on the host side, everything else can be done on the server side.
If possible, try it out in your own “lab” or non-production side first to get the hang of things. I’ve never messed with the Zabbix appliance, but I would say I generally agree with “not for production”.
I think those appliances are all-in-one builds, db, webserver all on the same host, that will just work. You can also build it on a headless debian with mariaDB and Apache all on the same host too.
It doesn’t need that much in terms of resources, in the past it ran fine on a Pi.
A few times I did zabbix upgrades and it broke the system, I had no idea how to fix it, so building in a vm, then having the ability to roll back might be handy.
I think they offer it as a docker container too. The appliance wanted 1.5gb ram, 8gb of disk, and 2 cores for a VM. The installer was a net version, so you need to have DHCP and internet available, I made a DHCP reservation so things don’t get lost.
The VM is up on my lab, but that’s probably as far as I’ll get today. The web page is running and firewall allowed me to see the log in screen, so some of it is working. Waiting for a book to arrive on Zabbix 6.0 (most recent available).
The part about breaking it with updates does reinforce that I should know how to build it from scratch. I’m just not much of a Linux person. Probably Debian 12 will be the production version.
Everything will just work if you install it all on a single vm, you need to make a few tweaks to install in a db on another host.
Not too much to it when it comes to installing. However, things break frequently with Zabbix but it’s more to do with the host OS or SQL or versions, just a quick look in their forums will bring it up.
This is the book I bought, should be here in a few days, I just like having books around for reference when I’m working through a problem.
Protip buying directly from Packt gives you the option to get an ebook with the printed book (pdf or epub), but some books have a pdf available to download in the first few pages of the printed book (the above book, so don’t spend the extra on this unless all you have is a Kindle).
Note on the book… It is set up for Zabbix 6.0 LTS and a different version of MariaDB than you get from the standard apt install command. Passage of time and all that.
Also there are some typos in the commands to install Zabbix that I had to weed out, at least for version 6.4 that I installed. But up and running on Debian 12 (latest) on my lab. The commands were all correct on the Zabbix website and mostly what I ended up using this time around.
I found an old post regarding installing the agent on XCP-NG hosts and Tom said to just install the agent. Can I assume that means just “yum install zabbix-agent” or something similar.
Trying to get a small grip on this in my lab so I can start rolling it out to production either next week or in the coming weeks. I need an easier way to see “more stuff” on my servers, WAC doesn’t really let me see all of them without a bunch of clicking and waiting around for things to finally load. Also tells me nothing about the XCP hosts that they are living on.
Any windows agent pitfalls I should know about before I get to them? All Server 2022 at this point, and fully patched for critical patches.
I can tell you that getting the agent installed on XCP-NG Beta 8.3 is not simple, missing some libraries that are not in the standard XCP repo, I’ll have to solve this. See image for specific error, this happens on both Zabbix 6.0LTS and 6.4.
