Hello,
I’m currently moving my single-node PostgreSQL app (running the flex variant) to a HA one, and the process I did was the following:
- Clone my
mad
instance intocdg
region. - Clone my
mad
instance intoams
region.
Then waited until they become healthy, and tried a failover. It failed without any error.
Then I learnt about regional failovers not being done automatically and about this steps to do it manually, but after changing my PRIMARY_REGION
env var, nothing happened and mad
was still the leader.
Am I right thinking that regional failover is not supported at all?
What is the best practices, running HA just in your primary region?
To be honest, I just want HA to protect from Out Of Memory errors and so, regional failovers was just a bonus tip for me.
Btw, the experience with postgresql-flex
has been much better than with the Stolon variant, which failed for no reason for me several times. Great work!