fix(ci): non-interactive drift/plan/apply workflows; surface tofu errors #11
Loading…
Reference in a new issue
No description provided.
Delete branch "fix/drift-check-noninteractive"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
The scheduled drift job hung at 'Refresh-only plan' until the job timeout: init_terraform only passes -backend-config when OBJECT_STORAGE_*/TF_VAR_tf_state_bucket are set (the workflow never sets them; terraform-plan.yml's comment assumes the runner has them), root_pass/user_password have no defaults, and nothing set TF_INPUT=0, so tofu waited for typed input.
Changes: TF_INPUT=0 on drift-check, terraform-plan and terraform-apply; backend settings on drift-check (bucket/region/endpoint non-secret, key pair from repo secrets OBJECT_STORAGE_ACCESS_KEY_ID / OBJECT_STORAGE_SECRET_ACCESS_KEY, which must be added in Settings -> Actions -> Secrets); placeholder root_pass/user_password for the refresh-only plan; the step now fails on tofu exit 1 (previously swallowed, so a failed run looked green).
Not tested: tofu isn't available locally, so TF_INPUT=0 behavior is per OpenTofu's documented semantics, not observed. The ovh and ovh-dr matrix jobs still lack OVH credentials in this workflow and will now fail fast instead of hanging.
🤖 Generated with Claude Code