Closed donovanmuller closed 5 years ago
Thank you for reporting this, @donovanmuller. It seems that the original test for the error code might have been too clever. It seems that the test did what it should have done, it failed with the expected code and it even said that. However, the code handling did not handle the magic value correctly.
Steps to reproduce
When using a remote bastion, if an Ansible play has been run before but it did not complete successfully on the target host, then any subsequent
terraform apply
will fail after Ansible connects with this error message on the bastion instance:Deleting the directory
/tmp/tf-ansible-bootstrap/
on the bastion is a workaround.Expected behavior
Ansible play is re-run as it was the first time
Actual behavior
Fails after first failed apply with:
Configuration
Terraform version: Terraform v0.12.9
terraform-provisioner-ansible version/SHA: 2.3.1/1ea0e20
Terraform file / provisioner configuration:
Terraform run log: