Amplify push fails with UPDATE_ROLLBACK_COMPLETE shown as an error status.

I was trying to make some Amplify schema.graphql updates and the CloudFormation stack ended up showing an error status (red x) next to UPDATE_ROLLBACK_COMPLETE. The specific nested stack showing that error is an update to an API DynamoDB table (which isn't what I modified in the schema). Cloud Formation doesn't allow the Continue update rollback apparently because the status is UPDATE_ROLLBACK_COMPLETE rather than UPDATE_ROLLBACK_FAILED.

I have tried reverting the schema.graphql to the prior version with just some comment changes and running "amplify push", but no luck. It shows UPDATE_COMPLETE_CLEANUP_IN_PROGRESS on the erroring nested stack, but then that fails and it shows the same error again.

The Amplify CLI output only gives the following error: Rolled back (2 of 1) 🛑 ["Index: 0 State: {"deploy":"waitingForDeployment"} Message: Resource is not in the state stackUpdateComplete"]

I haven't been able to find anything in the Troubleshooting CloudFormation document about this particular state.

Note, that I was trying to make a change to a secondary index on a particular table when this happened. I omitted the sortKeyField, but it sounds like you can't actually make that sort of change to an existing GSI.