Need suggestion upgrading zabbix

[homestuckfan123]

Hi all,

I'm going to be running our upgrade from 3.2 to 4.0 within the next month and i'd like to ask for peoples experience on timescales.

I understand from documentation that the database schema updates may take a considerable while.

In terms of hardware;

Zabbix is running on a Pacemaker cluster made up of 2x Dell R630s

And our Zabbix Database is on a galera cluster ~600GB
autoclicker software
I'm monitoring ~32000 hosts

How much downtime should I realistically be planing for?

[tim.mooney]

I wish I could offer some estimate for you, but your Zabbix environment is quite different from mine. I'm guessing the only way you're going to be able to get a semi-accurate estimate is if you create a test copy of your database either in the same cluster or on similar hardware and then do the upgrade in your test environment.

If it helps, it only took about 10-12 minutes when I upgraded from 3.2.x to 4.2.x.

My Zabbix server is a single Dell R730 with local MariaDB (not a separate server, and not Galera cluster) on separate fast (but not SSD) disks. My server's database is just under half the size of yours, but I suspect we have more history than you do, but you likely have far more items and triggers than we do. So, more differences that make it harder to compare the two environments for timing.

When I upgraded from 3.2, version 4.2.x was the latest active Zabbix feature branch, so I upgraded from 3.2.x to 4.2.x, not the 3.2 to 4.0 that you're doing.

Here's the start of the database upgrade from my server logs:

13436:20190814:165708.636 current database version (mandatory/optional): 03020000/03020001
13436:20190814:165708.636 required mandatory version: 04020000
13436:20190814:165708.636 starting automatic database upgrade
13436:20190814:165708.754 completed 0% of database upgrade

Here's where it finished about 10 minutes later:

13436:20190814:170108.228 completed 100% of database upgrade
13436:20190814:170108.228 database upgrade fully completed
13436:20190814:170108.337 starting event name update forced by database upgrade
13436:20190814:170108.341 completed 1% of event name update
13436:20190814:170108.341 completed 2% of event name update
13436:20190814:170108.342 completed 3% of event name update
13436:20190814:170108.343 completed 4% of event name update

[ leaving out more messages like this related to event name update ]

13436:20190814:170109.155 completed 99% of event name update
13436:20190814:170109.155 completed 100% of event name update
13436:20190814:170110.185 event name update completed

You may not even have the "event name" stuff; I would have to look back through the versions to find out if that change was for product changes between 3.2 and 4.0, or if it was for a product change between 4.0 and 4.2.

Again, hard to compare your environment with mine.