While upgrading our internal service discovery system, we accidentally broke standby machines. We’ve fixed that now.
As a reminder, when you deploy an app with no services, we create a second standby machine for you by default (unless you specify --ha=false). The standby will automatically be started if the primary machine’s underlying host has a problem. Apps with a service & a port don’t need a standby as our load balancer, fly-proxy, will start machines on demand.
If you run fly status, you can see any standby machines:
