09/26/2022, 10:17 AM
I’m seeing an issue where, if we get an error from upstream (Google Cloud Platform in this case), the resource we’re trying to create isn’t fully added to our stack. There’s a reference to the resource in the stack, but it seems incomplete, presumably because creation didn’t complete. We’re creating Cloud SQL instances, and we add a randomized suffix to the name to avoid collisions, and the effect that we’re seeing is that we create a new Cloud SQL instance on the next attempt at upping the stack. What’s the recommended way to deal with this error mode?