Simple Implementation Question

[DonPook]

Hello guys,

The scope has changed to a not so simple implementation question.

I am quite new to zabbix and have only installed an "all in one" server for a demo, but now our customer requires the actual solution.
A new implementation of 2 proxy servers, 2 application servers and 2 DB servers.
Now I have some questions.

1. Do I need to have a different dedicated DBs for proxy servers?
2. Are 10050-10051 ports enough for communication between Application - Proxy -agents?
3. Other FW rules I am thinking to ask for are ports: 80 & 443 for Zabbix UI. Anything else you guys believe I should ask for?
4. Can I install the whole solution without root privileges? (I've seen a topic saying that I can but I didn't understand how as I am not very familiar with linux) If yes, is there a guide available or anyone knows how?
5. Any good guide on how to implement the above HA? It seems tricky.

Thanks in advance

[gert.derouck]

Hi DonPook,

i'm curious about the size of your customers environment and the SLA on your monitoring environment.
In my opinion there is nothing wrong with an "all-in-one" server for production in lots of environments. Just make it HA by running it as a VM, so you're proof to hardware failures.


About your questions:
1. Proxy's need there own database which is typically installed on the proxy itselve, since it's 'volatile' data
2. Yes
3. Other ports: APP->DB (since you're running DB on separate servers); proxy<->managed assets (ports depend on all the methods you intend to use for monitoring)
4. You'll need root prvileges during the install, but the solution will run as a normal user by default.
5. For setting up application HA for the application servers, you'll need cluster tools as for instance pacemaker/corosync to implement a failover group of your zabbix application. For the database HA, you could use the same tools, or you can implement database replication. For each of these setups there are lots of tutorials and howto's available on the web.


I must warn you... You're saying that you are not very familiar with Linux. Then it will be a challenge to implement a full blown HA solution as you described...

Regards
Gert

[DonPook]

Hello Gert,

Thank you for your answer. Fortunately, our customer's linux team will handle the HA of the linux servers (and DBs), so what is left for us is the implementation of Zabbix. (I have successfully done it in our lab using pacemaker/corosync, but it seemed a bit tricky)

Our customer is a big telco company, which will give out the Zabbix monitoring solution to its smaller customers (different SLAs per customer, and size will start at 500 hosts and expand with each customer). Hence the HA on multiple servers.

Best Regards.

[DonPook]

Nevermind this

[gert.derouck]

Yep. That's for active and passive checks

[DonPook]

Some more questions regarding general information.

We will basically use Grafana for our dashboarding and report exporting tool.

1) Can we add "notes" or "instructions" on alerts? For example when a server goes down I want a message in the alert console of Grafana, that will give instuctions to a SOC team.
2) Is there event colleration - but - meaning that events won't come as duplicates? Say, alerts for a file check that may come more than once, will there be deduplication or I will get a spam of alerts for every check on the same file?
3) Kinda silly question, but just to be sure, pacemaker/corosync can be used for Proxy servers as well, correct?

Thanks a bunch.