**Language / Язык:** [English](seed-profiles.md) | [Русский](ru/seed-profiles.md)

# Seed profiles

Seeds replace the vSphere inventory **atomically** using deterministic MOIDs
so labs are reproducible. Definitions live in
[`app/vsphere/profiles.py`](../app/vsphere/profiles.py).

```bash
make seed                          # default: large (10 hosts / 1000 VMs)
VSPHERE_PROFILE=small make seed
```

## Profiles

| Profile | Contents |
|---|---|
| `small` | 3 ESXi hosts, 2 datastores, 2 networks, one datacenter/cluster/resource-pool, and five named VMs: `web-01`, `web-02`, `db-01`, `app-01`, `jumpbox` (mixed power states). Used by unit/integration tests. |
| `large` (default) | Configurable hosts/VMs (`SEED_VSPHERE_LARGE_HOSTS` default 10, `SEED_VSPHERE_LARGE_VMS` default 1000), 4 datastores, 4 networks/portgroups, a `VmwareDistributedVirtualSwitch`, production/staging/templates VM folders. The first five VMs match the `small` names for cookbook stability; the rest are generated (`web-`, `app-`, `db-`, `cache-`, `batch-`, `jump-`, `ci-`, `mon-`, `log-`, `ml-` role prefixes). |
| `demo-cluster` | `large` with 20 hosts / 1000 VMs — an enterprise-shaped dataset for UI demos. |

Every profile also seeds the four lab credentials, role-scoped permissions
(see [Authorization](domains/authz.md)), and — where the platform tables
exist — a starter content library, tag categories/tags, and datastore file
metadata (`seed_platform_extras`).

## Examples

```bash
make seed                                   # large, 10 hosts / 1000 VMs
VSPHERE_PROFILE=small make seed
VSPHERE_PROFILE=demo-cluster make seed
VSPHERE_PROFILE=large VSPHERE_HOSTS=20 VSPHERE_VMS=5000 make seed
```

Or run the seed CLI directly with the underlying environment variables (for
example from a non-`make` script or CI step):

```bash
SEED_VSPHERE_PROFILE=small \
  docker compose run --rm --entrypoint python simulator -m app.simulation.seed_cli
```

## Topology shape

Every profile builds the same skeleton (`Datacenters` folder → `Datacenter` →
host/vm/datastore/network sub-folders → one `ClusterComputeResource` +
`ResourcePool`), then scales hosts, datastores, portgroups, and VMs. VM MOIDs
are `vm-{100+n}`; host MOIDs are `host-{10+n}`; each VM carries the same
hardware shape used by both REST (`hardware/*`) and SOAP (`VirtualMachineConfigInfo`)
responses — NICs, disks, CD-ROM, boot order, and a synthetic guest IP/filesystem.

## Demo cluster via UI

The interactive console can load the demo dataset and reseed on demand:

- `POST /ui/api/demo/load` — loads `demo-cluster`
- `POST /ui/api/demo/unload` — wipes API-created state, then loads `small`
- `GET /ui/api/demo/state`
- `POST /ui/api/vsphere/seed?profile=small|large|demo-cluster` — reseed any profile

These UI helper endpoints are development-oriented and are not separately
authenticated today. Treat them as lab controls only.

## Reseed vs client state

Terraform, Pulumi, and Ansible may still hold resource state after a reseed
(VM MOIDs and names can change). Refresh or destroy/recreate external state
after replacing the PostgreSQL inventory. See [Operations](operations.md) and
[Clients](clients.md).