Comment by andydd

Can you confirm that it's impossible to avoid downtime, even with HA, because of forced updates?

Surely that's what HA is? no downtime as you update each node one at a time?

If it's impossible then it's a dealbreaker.