[General Usage]: How to migrate from older provider version to latest without recreating tables

[piyushbhojgude]

Terraform CLI Version

No response

Terraform Provider Version

No response

Company Name

No response

We are trying to upgrade snowflake terraform provider from 0.69 to 1.0.1 version.
Would like to know if there are any basic guidelines on how we can successfully upgrade.
I have made the necessary code changes, to make sure every resource is modified as per latest release, and verified that fresh deployment works fine.
But now I need to take care of use-case where we already have all snowflake resources deployed and maintained using 0.69 version.
When I try to run terraform plan against existing deployment, I see below errors for certain resources:

1. For views:
   Warning: Failed to decode resource from state
   │
   │ Error decoding "module.services["workflow_name"].snowflake_view.wf_views["VIEW_NAME"]" from prior state: unsupported attribute "created_on"

2. For tables:
   Warning: Failed to decode resource from state
   │
   │ Error decoding "module.workflows["workflow_name"].snowflake_table.wf_tables["TABLE_NAME"]" from prior state: unsupported attribute "data_retention_days"
   ╵

3. For tasks:
   │ Warning: Failed to decode resource from state
   │
   │ Error decoding "module.workflows["workflow_name"].snowflake_task.wf_tasks["TASK_NAME"]" from prior state: unsupported attribute "enabled"
   ╵

Seeing below error for procedures, functions, file_format and all types of grants

│ Warning: Missing resource schema from provider
│
│ No resource schema found for snowflake_procedure when decoding prior state

We are okay with removing the older deprecated grant resources manually and then have them created via the new terraform provider.
But we would like to keep the tables as is and not re-create those.

I was going through this documentation: https://github.com/Snowflake-Labs/terraform-provider-snowflake/blob/main/docs/technical-documentation/resource_migration.md

Do we need to import all existing table resources into new one, in order to avoid recreation?

Category

category:resource

Object type(s)

resource:table

Expected Behavior

NA

Actual Behavior

NA

Steps to Reproduce

Upgraded tf provider version from 0.69 to 1.0.1, and modified all resources accordingly.
Tried running terraform plan.

How much impact is this issue causing?

High

Logs

No response

Additional Information

No response

[sfc-gh-asawicki]

Hey @piyushbhojgude. Thanks for reaching out to us.

Yes, basically to avoid recreation, the easiest way is to remove resources from the state and reimport them as described in the resource migration you linked.

Also, you are taking a huge leap from 0.69.0 to 1.0.1. These last 1.5 years were busy and there were lots of changes in the provider. Please make sure that you not only adjust the resources according to the newest docs but also that you went through the migration guide where all the important provider changes were listed version-by-version (starting at around 0.73 version).

[piyushbhojgude]

Hi @sfc-gh-asawicki , thanks for the update and for the detailed info in migration doc, we were able to successfully migrate to 0.100 version. Didn't move to 1.0 since we wanted to still use the older function and procedure resources. Looks like COPY GRANTS aren't supported currently for procs/UDFs, we will lose all grants each deployment, which we didn't want to.
And we are also facing similar issue: #3297