HA pair across a 10G WAN link? And... MySQL replication for DR?

[jtnfoley]

I'm trying to imagine an architecture where an HA pair can also support disaster failover, even if manually.
I believe Zabbix HA is two (or more?) active Zabbix hosts connecting to the same database engine. I have 10gig fiber between my primary and backup datacenters, I wonder if Zabbix will be happy with Zabbix A and MySQL in the primary datacenter and Zabbix B in the backup.
Then, for DR purposes, MySQL replication tools to replicate the database to the Backup DC. In the event of genuine disaster, Zabbix B (still in my imagination) could be re-pointed to MySQL B?

So, two questions (three, including "is anyone already doing this???") : Will Zabbix be happy with an HA secondary server being separated from the primary and database by 10Gbit? And can the Zabbix secondary be made primary (or standalone) and manually connected to a replicated copy of the database?

[cyber]

You should have somekind of DB HA, not just simple replication to other site.. Server HA expects to have connections from (both) servers to same DB. So if you just replicate, then, in case of disaster (and server decides to switch) you still need to go and point your secondary host to replica DB and start it up .. no point having a HA server, could just ha astandby, which is always pointedto backup DB and start it manually ... Switching back will be as big of a headache...

[jtnfoley]

Understood - I'm looking at A/A options for MySQL.
Meanwhile I'm going to slow-walk into the 7.2.2 upgrade (the end goal of this effort is to get HA/DR while effecting a version upgrade from 6.0.x.)

[jtnfoley]

Slow walking so far: I stood up a 8c/48G VM running a modern Linux in an offline environment (production is airgapped) with MySQL 8.4 and all the supporting Perl/Python/Zabbix agent to become a dedicated MySQL server. Migrated Zabbix 6.0.x to a split frontend/backend then used https://gitlab.com/jhboricua/zabbix-mysql-partitioning python partitioning to get my history/housekeeping under control. (I need to submit a bug report to jhboricua - there was an issue with the systemd files I had to fix.)
I tuned MySQL to take advantage of all that RAM before partitioning. It took about four hours to partition 77gb history_uint and 23gb history_text (I think??) and I'm thrilled to be off of the housekeeper. (<=== one of my goals for this upgrade/migration!)

So, now I have a stable and working partitioned MySQL 8.4, separate from the Zabbix app and GUI server. Next steps will be 6.0.x->7.2.current migration in the primary DC, then set up a proxy and abandon the standalone Zabbix in the backup DC, followed by Zabbix HA, finally MySQL HA.
If anyone has been down this road before... does this sound like a plan?

[cyber]

<insert a smug face of John "Hannibal" Smith here> with his most quoted quote ever...

[jtnfoley]

I LOVE it when a plan comes together!