I found configuring an HA setup easier to do with MySQL. The ability to configure a master <-> master setup simply is really helpful. Of course, having writes on both sides is generally a bad idea, but it's quite simple to have master - hot standby setup, and not go through the slave promotion step.

An HA setup can be a master <-> master setup with a VIP using VRRP and a check script ( keepalived). Of course, you have to remain cautious about network partition. Another thing that might be interesting in some use cases is the ability to skip some replication errors. This is specially interesting in cases where consistency is not critical.

I actually did some setup like that with a replication ring (4 full masters), and an additional daemon re-configuring the ring dynamically when a node was down. It also monitored the transaction log, trimming them if they were about to fill up the disk and setting the GTIDs to the new values. I added some skip error to not block replication. However it was for a very simple DB (session DB containing just one table, but an SQL db was required). Basically I switched from a CA DB to an AP DB, and it's nice to be able to do these kind of things.

I know those are too simplistic setups not taking into account all the failure modes. But it also make them easier to understand and to debug.