tepichord/milner

Fork 0

IaC Worker Node - Naina #29

Merged

tepichord merged 13 commits from feat/17-iac-worker-node into main

2026-09-08 15:24:34 -07:00

tepichord commented

2026-09-08 13:32:51 -07:00

Owner

Closes #17

Automated PR created by forgejo-done adapter.

Closes #17 Automated PR created by forgejo-done adapter.

tepichord added 12 commits

2026-09-08 13:32:51 -07:00

feat(control-plane): expose cluster identity outputs and worker DHCP reservation 165a8e3664

Add worker_dhcp_hosts variable and 6 new outputs (machine_secrets,
client_configuration, cluster_name, cluster_endpoint, talos_version,
kubernetes_version) to control-plane root so the worker root can
read cluster identity via terraform_remote_state.

Append worker DHCP reservation to libvirt_network DHCP hosts via concat().
Add test assertions for worker reservation and output presence.

Refs: #17

feat(worker): add Terraform root for naina worker-large node 48c229a9ab

Add worker/main.tf: remote-state read from control-plane, worker
volumes/domain (20480 MiB, 6 vCPU, 40 GiB qcow2, MAC 52:54:00:10:00:04
on control-plane-net), and talos_machine_configuration_apply.worker
(joining via shared machine secrets, no bootstrap/kubeconfig).

Add worker/terraform.tfvars with repo defaults matching ARCHITECTURE.md
sizing. Add mock-provider test suite (plan.tftest.hcl) with fixture CP
state for offline testing.

Refs: #17

feat(worker): add cleanup script and runbook 9042293312

Add worker/cleanup.sh: confirm-gated terraform destroy of worker
resources only (no pool sweep — pool is CP-owned). Mirrors
control-plane/cleanup.sh style.

Add worker/README.md: runbook covering deploy, ISO eject, talosctl
usage, VM replacement, full teardown, apply-order constraints, and
cross-root resource ownership.

Refs: #17

chore: remove superseded shell provisioning artifacts 15c86f19bd

Delete worker/vm-setup/create-worker-vm.sh, worker/tests/
test_create_worker_vm.bats, worker/patches/worker-patch-1.yaml, and
tests/test_shell.py. These are superseded by the worker Terraform root
(main.tf + tests/plan.tftest.hcl). Shell config patches are now inline
config_patches in the Talos machine config apply.

Refs: #17

test: extend test harness for worker root 53ab734394

Add test_terraform_test_worker to run terraform test in worker/ alongside
the existing CP test. Extend terraform fmt -check to include worker/.
Update shellcheck list to replace deleted bats script with worker/cleanup.sh.

Refs: #17

docs: update repo-map, runbooks, and PLAN.md for worker root 182e1cb78a

Add worker node runbook to root README.md. Update AGENTS.md repo-map
(worker/ = Terraform, not shell scripts) and testing commands. Mark
PLAN.md Phase 4 as complete (worker-large node is Terraform-managed).

Refs: #17

docs(repo): fix stale BATS reference and worker description in docs ade2c8a964

Refs: #17

fix(test): correct worker test fixture types for talos provider 3074305e98

The worker test fixture cp.tfstate provided machine_secrets and
client_configuration as plain strings, but the talos provider expects
nested objects. Update the fixture to use proper Terraform state v4
complex type encoding with all required fields.

Also add the worker .terraform.lock.hcl (previously untracked) to
match the control-plane convention.

fix(worker): source talos/kubernetes versions from control-plane state 975e190c88

The worker duplicated talos_version/kubernetes_version as local knobs that
had to be manually kept in sync, while the control-plane root publishes
them as remote-state outputs specifically for the worker and the worker's
own tests and README document reading them via terraform_remote_state.
Read the versions from the CP state (single source of truth) for the
machine config, installer ISO URL, and install image, and drop the
now-unused variables.

test(worker): replace vacuous no-network/pool assertion with join-secret checks 4fd335d1e6

The no_network_or_pool_resources run claimed to verify the worker declares
no network or pool resources but actually asserted an unrelated remote-state
output count, so it passed regardless. Absence of network/pool resources is
structural (nothing undeclared can be planned), so drop the run and instead
assert the sensitive machine_secrets/client_configuration outputs the worker
actually requires are published by the control-plane state.

docs: correct installer-ISO ownership in worker runbooks 081566a2fd

Both runbooks claimed the installer-ISO volume is owned by the
control-plane root and never re-declared by the worker, but worker/main.tf
declares its own talos-metal-arm64-worker.iso volume (destroyed at eject).
Describe the actual ownership split: network and pool are CP-owned and
referenced by name; the worker owns its own installer ISO.

chore(flake): drop bats dev-shell dependency 502cba764f

This branch removed the only bats consumers (tests/test_shell.py harness
and worker/tests/*.bats), leaving bats in the dev shell as a stale
dependency.

tepichord added the

reviewing

label

2026-09-08 13:32:52 -07:00

tepichord added this to the MVP project

2026-09-08 14:10:30 -07:00

tepichord added 1 commit

2026-09-08 15:23:11 -07:00

docs(worker): correct network-replacement behavior, pre-reserve worker DHCP slots 9b1d7a4f34

- worker/README.md: a DHCP-reservation edit replaces the libvirt network
  (previously claimed to be an in-place dnsmasq restart); the recreation
  strands every attached VM until it is hard-restarted (ACPI reboot does not
  re-create the host-side tap). Add an 'Adding a worker node' section that
  prescribes batching reservations to keep network churn near zero.
- control-plane/terraform.tfvars: pre-reserve workers 2-4 (:05 -> 10.5.0.21,
  :06 -> 10.5.0.22, :07 -> 10.5.0.23) extending worker-1's .20, so future
  workers add domains only, never network changes; document the MAC scheme.
- worker/terraform.tfvars: point the identity comment at the CP table.
- worker/tests/plan.tftest.hcl: pin memory_mb/vcpu_count alongside
  installer_iso so plan tests stay deterministic against local *.auto.tfvars
  overrides.

tepichord merged commit e7521360b7 into main

2026-09-08 15:24:34 -07:00

tepichord deleted branch feat/17-iac-worker-node

2026-09-08 15:24:34 -07:00

tepichord referenced this pull request from a commit

2026-09-08 15:24:34 -07:00

Create Worker Node

tepichord referenced this pull request

2026-09-08 15:45:41 -07:00

Add kubectl bootstrap script #30