Need to figure how how to best run more than one Docker discourse instance to withstand EC2 instance failures. * run more than one instance at all times * configure auto-scaling/ELB to ensure there's always at least one healthy node
For this we just need to up the minimum instances in the auto-scaling config, but I'd like confirmation from jp first just in case he has any objections.
I suppose the question is how fault tolerant it needs to be. I'm fine with it running a second AS node, but if you can take 1-2 min of downtime, it will save a bunch of $$ to keep the AS group scaled to one.
We'll stick with one instance and ensure it spins a new one up if it ever goes down.