`terraform apply <planfile>` fails, when `terraform apply` succeeds

[simoncocking]

Terraform Version

Terraform v0.12.1
+ provider.aws v2.13.0
+ provider.external v1.1.2
+ provider.template v2.1.2

Terraform Configuration Files

See this zip file encrypted with your GPG key.

Debug Output

Traces for both terraform plan and terraform apply phases are in the above zip file, as is the generated plan file (which contains our Terraform code).

Expected Behavior

The plan should have been applied without error.

Actual Behavior

Many instances of the following errors:

Error: Resource node has no configuration attached

The graph node for
module.dwh.module.legal_status_table.data.template_file.to_s3 has no
configuration attached to it. This suggests a bug in Terraform's apply graph
builder; please report it!

and

Error: orphan resource module.dwh.module.f_term_info_table.aws_glue_trigger.on_demand still has a non-empty state after apply; this is a bug in Terraform

Steps to Reproduce

1. terraform init
2. terraform plan -out=staging.tfplan
3. terraform apply staging.tfplan

Note that when omitting the plan phase and allowing terraform apply to calculate the plan itself, the apply succeeds without error.

[simonbirtles]

I am also having the same issue on the same version. When using a plan file (cli or PTFE) the apply fails as described. When applying without a plan file, the run successfully completes (cli).

Not a useful comment I know.. I am trying to find a way of providing logs and configs as our security do not permit me to post to any public forum even encrypted. Any ideas of internal hashicorp file upload ? I have PTFE support - can I open a case there and upload to the case ?

[simonbirtles]

0.12.2 Same issue - Error: orphan resource module.xxxxxxx.module.ddi.data.solidserver_dns_zone_list.reverse_zone still has a non-empty state after apply; this is a bug in Terraform

[simonbirtles]

0.12.3 issue persists.

[alexpekurovsky]

Just sharing my 3 days debugging story. I had also huge list of
Error: orphan resource ... still has a non-empty state after apply; this is a bug in Terraform

My problem was that couple of resources in the module had:

lifecycle {
    ignore_changes = ["metadata.0.labels"]
}

This was done for Terraform 0.11, as it couldn't parse some labels (with dots or slashes). Terraform 0.12 successfully parses labels, so this block is not needed. However exactly this block causes problems. Removing the block fixed an apply planfile.

This is definitely a bug, I had to remove all resources and add one after another to find the issue

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.