Don't fail deleting an already deleted bucket

[nhoughto]

Currently when terraform destroying a terraform managed bucket, the provider fails with an error if the bucket has already been deleted.

Error is : Error: No such bucket: 629955074482c...

Since we are trying to delete the bucket, and the bucket is already deleted a more user-friendly behaviour would be to just complete successfully, this is the default behaviour for many providers including the AWS S3 bucket resource. At the moment the workaround is to terraform state rm.. the resource manually to fix the problem, this is obviously not ideal.

The reason the bucket is deleted outside of terraform is related to: #22 so a solution to either would work =)

[nhoughto]

Hmm just tested this with 0.7.0 (we had to wait for TF 1.0 upgrade), and i still see this failing?

│ Error: Traceback (most recent call last):
│   File "b2_terraform/provider_tool.py", line 497, in run_command
│   File "b2_terraform/provider_tool.py", line 120, in run
│   File "b2_terraform/provider_tool.py", line 242, in resource_read
│   File "b2_terraform/api_wrapper.py", line 37, in get_bucket_by_id
│ b2sdk.exception.NonExistentBucket: No such bucket: b2d935b784127f137acc0e13

[nhoughto]

maybe the exception handling of BadRequest doesn't match what actually happens, says its throwing a NonExistentBucket exception?

[mszumocki-reef]

No, there's something else at play here - it's an error in resource_read, not resource_delete.
Could you check what terraform plan shows in this situation for you?

[nhoughto]

Ah ok, so the situation is im running a destroy (after apply) and in between the two commands the bucket has been deleted manually. I guess resource read should return an empty state rather than an exception? I can’t remember what the terraform interface for this scenario expects. The reason it’s doing a read before destroy is to figure out what to destroy I suppose.

[nhoughto]

Looking at what the aws s3 resource terraform provider does, it logs a warning in Read if bucket doesn't exist and then removes the bucket from state, so no error is thrown, I believe this is what terraform expects in this situation, code:

https://github.com/hashicorp/terraform-provider-aws/blob/main/aws/resource_aws_s3_bucket.go#L835

[mszumocki-reef]

I'll be releasing that change in 0.7.1

[mszumocki-reef]

Released, @nhoughto please let me know if that works for you.