From the docs it shows one must step through a lot of steps when migrating from one release to another. Here’s the doc for the 0.14 upgrade. However, we have an instance that a bug found in 0.13 was not fixed until 0.15. With customers currently using 0.13 with their deployments, upgrading to 0.15 would be ideal. There is no need for them to use 0.14 at all. What are the drawbacks using these steps:
- Install 0.13
- tf init
- tf apply . . . # Initial 0.13 deployment
- Install 0.15
- tf init
- tf apply . . . # Upgrade to 0.15 adjusting the tfstate file