Re-running terraform apply after initial Error will error out with "domain [...] already exists"

[raoulbhatia]

Related to #920 and also #423

System Information etc.

See #920

---

Description of Issue/Question

Setup

See #920

Steps to Reproduce Issue

1. terraform apply until you hit Automatically configure app-armor for non-standard paths (to work around upstream bug) #920
2. Manually fix by editing /etc/apparmor.d/local/abstractions/libvirt-qemu
3. terraform apply

libvirt_domain.domain-ubuntu: Creating...
╷
│ Error: error defining libvirt domain: operation failed: domain 'terraform-kvm-ansible' already exists with uuid 7454585d-3a7c-42a9-a822-835b9ca3c1b6
│
│   with libvirt_domain.domain-ubuntu,
│   on main.tf line 33, in resource "libvirt_domain" "domain-ubuntu":
│   33: resource "libvirt_domain" "domain-ubuntu" {
│

Solution

1. Run virsh undefine terraform-kvm-ansible
2. terraform apply
3. terraform-kvm-ansible will come up successfully

---

Additional information:

Apparmor as articulated in #920

[alancaldelas]

This also is happening to me, any updates?

[tovoro]

Same here while deploying Centos 8 guests on a Fedora35 Host. Additionally, the domain does not get destroyed when running terraform destroy

[alancaldelas]

Yeah, I'm also on a F35 host. Additionally, I'm doing a UEFI deployment and running the destroy also removes the /usr/share/edk2/OVMF_VARS.fd file so when I got re-deploy the file is now missing.