Schema migrations are in the critical path and need to block reboots so new code is not accessing old database schemas. Data migrations can be run at any time without downtime. Separating the two allow deployment to go more smoothly, as well as help with implementing zero downtime deployments.
Figure the best way to do this for now is just to point -d to two different directories and have two different scripts. I'll reopen if someone has a better idea.
Schema migrations are in the critical path and need to block reboots so new code is not accessing old database schemas. Data migrations can be run at any time without downtime. Separating the two allow deployment to go more smoothly, as well as help with implementing zero downtime deployments.