Confirmed working against a real VK Cloud PROD deployment (3 routers,
19 resources, apply succeeded end to end). Fixes found along the way:
- provider "vkcs" was never configured (versions.tf) - username/password/
project_id/region were declared but wired to nothing; added auth_url and
user_domain_name to complete it.
- Nova keypairs are per-user, not per-project - added an optional
vkcs_compute_keypair resource (var.ssh_public_key) so Terraform can
register a keypair under the deploying service account itself.
- router_priv_port used a hand-computed fixed_ip offset that collided with
VKCS's own auto-created service ports on each network (observed: a
"network:dns" port) - now left unset so Neutron's IPAM auto-assigns,
which is collision-free by construction.
- vkcs_compute_instance set image_id at the top level while also booting
from a volume via block_device - the provider docs say not to do this;
Nova echoes back a sentinel string for image_id on a volume-booted
server, which Terraform read as drift on a ForceNew attribute and
wanted to destroy+recreate every already-created instance on every
subsequent plan.
- private_network_cidrs bumped from /29 to /28 - too tight once the
platform's own reserved ports are accounted for.
Also removed the priv_srv_01/02/03 demo instances and the LAN network/
security group only they used - this deployment provisions router VMs
only, confirmed with the user.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011hXR2ftXZZhJ4Y3XuSoR8r
The "default" security group VK Cloud auto-creates in every project has
a UUID unique to that project, but main.tf hardcoded one project's UUID
across all 4 instance resources (router, priv_srv_01/02/03) - not
portable to another project. Replaced with data.vkcs_networking_secgroup
(matched by name = "default", verified against the real provider schema
via terraform providers schema -json on the local provider mirror) behind
local.default_security_group_id. The new default_security_group_id
variable is a last-resort manual override via coalesce() for projects
where the lookup doesn't fit (non-standard name/SDN) - not the normal path.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011hXR2ftXZZhJ4Y3XuSoR8r
private_supernet/private_subnet_prefix_length/private_interface_count
(which auto-derived per-router-per-role micro-subnets via cidrsubnet())
are replaced by a single required variable, private_network_cidrs: one
CIDR per shared private network that router VMs get an interface into,
supplied explicitly by the admin - no auto-carving. Each router gets its
own port/IP inside every listed network (cidrhost(cidr, router_index+2)),
closer to the original lan_net design but generalized to N networks and
N routers. network-init.sh.tpl needed no changes - it already matches
interfaces by CIDR membership regardless of whether the CIDR is shared.
Also fixes a testing gap found along the way: `terraform validate` does
not enforce variable validation{} blocks for externally-supplied values
in this terraform version - only `plan`/`apply` do. The test suite now
exercises those validations for real via `terraform plan` against an
isolated, provider-free copy of variables.tf.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011hXR2ftXZZhJ4Y3XuSoR8r
Terraform now provisions router_count IaaS Router VMs (default 2, no
longer hardcoded to router1/router2), each with 1 public +
private_interface_count isolated private interfaces (no shared LAN or
VRRP between routers). Both counts scale via Terraform variables and
TF_VAR_* environment variables. The post-install script became a
Terraform template that matches interfaces to their expected subnet by
CIDR instead of a fragile "first private IP" heuristic.
Added an offline pytest suite (terraform/tests/) that checks the
delivery's internal consistency and runs real terraform init/validate
against the actual vkcs provider schema via a project-local filesystem
mirror (provider binary fetched from its GitHub releases, bypassing the
region-blocked HashiCorp registry) - no cloud credentials or API calls
involved. terraform/versions.tf now declares the previously-missing
required_providers block.
Ansible (inventory.ini, base/frr_router/keepalived roles) still assumes
the old 2-router/2-NIC/VRRP topology and is not yet adapted - documented
as a follow-up, not addressed here.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011hXR2ftXZZhJ4Y3XuSoR8r
ayurishchevandClaude Sonnet 5