Fix CI/pulumi blockers and document test suites with latest PASS results.

Stabilize node status, cluster-wide Ceph OSD ids, and QEMU cpu utilization so status/list no longer 500 on model strings; align pulumi defaults to pve1; add bilingual testing docs with the 2026-07-18 ci-all + pulumi-tests pass.
2026-07-18 09:38:28 +03:00
parent 6f42d4c318
commit baa1f58ad0
31 changed files with 230 additions and 58 deletions
@@ -39,6 +39,9 @@ GitHub Actions runs `make ci` plus Compose/Helm validation on every push and PR
 to `main`. Run `make ci-all`, `make helm-lint`, and `make pulumi-tests` locally
 before larger API or client-facing changes.

+Suite layout, Make targets, and the latest recorded local results:
+[docs/testing.md](docs/testing.md).
+
 ## Project rules worth remembering

 1. Mutations must **persist to PostgreSQL** (tables and/or jsonb metadata).
@@ -36,6 +36,9 @@ GitHub Actions на каждый push/PR в `main` запускает `make ci`
 Compose/Helm. Перед крупными API/клиентскими изменениями локально прогоняйте
 `make ci-all`, `make helm-lint` и `make pulumi-tests`.

+Карта наборов, Make-цели и последние записанные локальные результаты:
+[docs/ru/testing.md](docs/ru/testing.md).
+
 ## Важные правила проекта

 1. Мутации должны **persist в PostgreSQL** (таблицы и/или jsonb metadata).
@@ -193,6 +193,7 @@ make release-up && make release-seed  # run the published stack locally
 ## Contributing / security / changelog

 - [CONTRIBUTING.md](CONTRIBUTING.md) · [CONTRIBUTING.ru.md](CONTRIBUTING.ru.md)
+- [Testing](docs/testing.md) · [Тесты](docs/ru/testing.md) — suites, Make targets, latest results
 - [SECURITY.md](SECURITY.md)
 - [CHANGELOG.md](CHANGELOG.md)

@@ -194,6 +194,7 @@ make release-up && make release-seed  # запустить опубликова
 ## Участие / безопасность / changelog

 - [CONTRIBUTING.md](CONTRIBUTING.md) · [CONTRIBUTING.ru.md](CONTRIBUTING.ru.md)
+- [Testing](docs/testing.md) · [Тесты](docs/ru/testing.md) — наборы, Make-цели, последние результаты
 - [SECURITY.md](SECURITY.md)
 - [CHANGELOG.md](CHANGELOG.md)

@@ -205,7 +205,7 @@ def _format_key_example(key: str, raw: Mapping[str, Any]) -> object:
    if "default" in raw and raw["default"] is not None:
        return raw["default"]
    enum = raw.get("enum")
-    if isinstance(enum, (list, tuple)) and enum:
+    if isinstance(enum, list | tuple) and enum:
        return enum[0]
    nested_format = raw.get("format")
    if isinstance(nested_format, Mapping):
@@ -226,10 +226,10 @@ def _format_key_example(key: str, raw: Mapping[str, Any]) -> object:
        return 1
    if typ == "integer":
        minimum = raw.get("minimum")
-        return int(minimum) if isinstance(minimum, (int, float)) else 1
+        return int(minimum) if isinstance(minimum, int | float) else 1
    if typ == "number":
        minimum = raw.get("minimum")
-        return float(minimum) if isinstance(minimum, (int, float)) else 1.0
+        return float(minimum) if isinstance(minimum, int | float) else 1.0
    pattern = raw.get("pattern")
    if isinstance(pattern, str) and "second|minute|hour|day" in pattern:
        return "1/second"
@@ -131,7 +131,7 @@ def register_acme_handlers(registry: HandlerRegistry) -> None:
            entry = {"plugin": plugin_id, **{k: v for k, v in item.items() if k != "data"}}
            if "digest" not in entry:
                material = {k: v for k, v in entry.items() if k != "digest"}
-                entry["digest"] = hashlib.sha1(
+                entry["digest"] = hashlib.sha1(  # noqa: S324 - Proxmox config digests use SHA-1
                    json.dumps(material, sort_keys=True, separators=(",", ":")).encode()
                ).hexdigest()
            result.append(entry)
@@ -245,7 +245,7 @@ def register_ceph_handlers(registry: HandlerRegistry) -> None:
    def _wire_ceph_pool(name: str, item: dict[str, Any], *, index: int) -> dict[str, Any]:
        pool_id = item.get("pool_id", item.get("pool"))
        try:
-            pool_num = int(pool_id)
+            pool_num = int(pool_id) if pool_id is not None else index
        except (TypeError, ValueError):
            pool_num = index
        crush_raw = item.get("crush_rule", 0)
@@ -586,9 +586,8 @@ def register_ceph_handlers(registry: HandlerRegistry) -> None:
                MAX(NULLIF(regexp_replace(external_id, '\\D', '', 'g'), '')::int),
                -1
            ) + 1
-            FROM resources r JOIN nodes n ON n.id=r.node_id
-            WHERE n.name=$1 AND r.kind='ceph-osd'""",
-            node,
+            FROM resources
+            WHERE kind='ceph-osd'""",
        )
        osd_id = int(next_id or 0)
        external_id = f"osd.{osd_id}"
@@ -607,8 +606,15 @@ def register_ceph_handlers(registry: HandlerRegistry) -> None:
            "used_bytes": 0,
        }
        await database(request).pool.execute(
-            """INSERT INTO resources(id, node_id, kind, external_id, state)
-            VALUES(gen_random_uuid(), $1, 'ceph-osd', $2, $3::jsonb)""",
+            """INSERT INTO resources(id, node_id, cluster_id, kind, external_id, state)
+            VALUES(
+                gen_random_uuid(),
+                $1,
+                (SELECT cluster_id FROM nodes WHERE id=$1),
+                'ceph-osd',
+                $2,
+                $3::jsonb
+            )""",
            node_id,
            external_id,
            json.dumps(osd_state, sort_keys=True),
@@ -193,8 +193,8 @@ def register_cluster_extra_handlers(registry: HandlerRegistry) -> None:
        now = int(time.time())
        seeded = metrics.get("export_series")
        if isinstance(seeded, list) and seeded:
-            series = [dict(item) for item in seeded if isinstance(item, dict)]
-            return {"data": series, "timestamp": now}
+            seeded_series = [dict(item) for item in seeded if isinstance(item, dict)]
+            return {"data": seeded_series, "timestamp": now}
        # Build a PVE-shaped metric array from live inventory (scales with seed).
        rows = await database(request).pool.fetch("SELECT name FROM nodes ORDER BY name")
        series: list[dict[str, Any]] = []
@@ -168,6 +168,7 @@ def build_core_handlers(settings: Settings) -> HandlerRegistry:
                "machine": "x86_64",
            }
        return {
+            "status": str(row["status"]),
            "uptime": int(payload.get("uptime") or 0),
            "wait": float(payload.get("wait") or 0.0),
            "idle": float(payload.get("idle") or 0.95),
@@ -344,7 +345,7 @@ def build_core_handlers(settings: Settings) -> HandlerRegistry:
                        "vmid": vmid,
                        "name": str(state.get("name") or f"{kind}-{external_id}"),
                        "status": status,
-                        "template": bool(state.get("template") in {True, "1", 1}),
+                        "template": bool(state.get("template") in (True, "1", 1)),
                        "cpu": _cpu_util(state, running=running),
                        "maxcpu": _maxcpu(state),
                        "mem": int(_num(state.get("mem"), 0)) if running else 0,
@@ -30,7 +30,7 @@ def _ha_groups(metadata: dict[str, Any]) -> dict[str, dict[str, Any]]:

 def _ha_digest(payload: dict[str, Any]) -> str:
    material = {key: value for key, value in payload.items() if key != "digest"}
-    return hashlib.sha1(
+    return hashlib.sha1(  # noqa: S324 - Proxmox HA digests use SHA-1
        json.dumps(material, sort_keys=True, separators=(",", ":")).encode()
    ).hexdigest()

@@ -70,7 +70,7 @@ def _public_lxc_config(
            continue
        payload[key] = value
    digest_src = {key: value for key, value in payload.items() if key != "digest"}
-    payload["digest"] = hashlib.sha1(
+    payload["digest"] = hashlib.sha1(  # noqa: S324 - Proxmox config digests use SHA-1
        json.dumps(digest_src, sort_keys=True, separators=(",", ":")).encode()
    ).hexdigest()
    return payload
@@ -554,10 +554,10 @@ def register_nodes_extra_handlers(registry: HandlerRegistry) -> None:
                    }
                )
        else:
-            path = payload.get("path")
+            target_path = payload.get("path")
            index = payload.get("index")
            for file_idx, file_entry in enumerate(files):
-                if path is not None and file_entry.get("path") != path:
+                if target_path is not None and file_entry.get("path") != target_path:
                    continue
                repos = [
                    dict(item)
@@ -584,7 +584,7 @@ def register_nodes_extra_handlers(registry: HandlerRegistry) -> None:
                                }
                            },
                        }
-                elif path is not None or file_idx == 0:
+                elif target_path is not None or file_idx == 0:
                    if repos:
                        repos[0] = {
                            **repos[0],
@@ -604,7 +604,7 @@ def register_nodes_extra_handlers(registry: HandlerRegistry) -> None:
                            },
                        }
                file_entry["repositories"] = repos
-                if path is not None:
+                if target_path is not None:
                    break
        apt["repositories"] = {
            "digest": secrets.token_hex(4),
@@ -78,12 +78,26 @@ def _public_qemu_config(config: dict[str, Any], state: dict[str, Any]) -> dict[s
    agent = config.get("agent", state.get("agent", "0"))
    payload["agent"] = _agent_config_string(agent)
    digest_src = {key: value for key, value in payload.items() if key != "digest"}
-    payload["digest"] = hashlib.sha1(
+    payload["digest"] = hashlib.sha1(  # noqa: S324 - Proxmox config digests use SHA-1
        json.dumps(digest_src, sort_keys=True, separators=(",", ":")).encode()
    ).hexdigest()
    return payload


+def _cpu_utilization(vm_state: Mapping[str, Any], *, running: bool, idle: float = 0.1) -> float:
+    """Status ``cpu`` is utilization; config may store model strings like ``qemu64``."""
+
+    raw = vm_state.get("cpu")
+    if isinstance(raw, (int, float)):
+        return float(raw)
+    if isinstance(raw, str):
+        try:
+            return float(raw)
+        except ValueError:
+            pass
+    return idle if running else 0.0
+
+
 def _qemu_index_item(vmid: int, vm_state: dict[str, Any], config: dict[str, Any]) -> dict[str, Any]:
    status = str(vm_state.get("status", "stopped"))
    running = status in {"running", "paused"}
@@ -96,7 +110,7 @@ def _qemu_index_item(vmid: int, vm_state: dict[str, Any], config: dict[str, Any]
        "status": status,
        "qmpstatus": status if running else "stopped",
        "cpus": int(config.get("cores", config.get("cpus", 1))),
-        "cpu": float(vm_state.get("cpu", 0.1 if running else 0.0)),
+        "cpu": _cpu_utilization(vm_state, running=running),
        "maxmem": maxmem,
        "mem": mem,
        "memhost": int(vm_state.get("memhost", mem)),
@@ -1682,7 +1682,7 @@ def _seed_tasks_for_guests(
    return tuple(tasks)


-# UI / ops cluster sizes (hosts × total guests). Infra scales with the size.
+# UI / ops cluster sizes (hosts x total guests). Infra scales with the size.
 _CLUSTER_SIZE_SPECS: dict[str, dict[str, int]] = {
    "lab": {
        "nodes": 1,
@@ -1953,13 +1953,13 @@ def _sized_cluster_profile(name: str, *, overrides: dict[str, int] | None = None


 def small_profile() -> SeedProfile:
-    """DATA UI: 3 hosts × 50 guests."""
+    """DATA UI: 3 hosts x 50 guests."""

    return _sized_cluster_profile("small")


 def large_profile(*, node_count: int = 10, resource_count: int = 1_000) -> SeedProfile:
-    """DATA UI: 10 hosts × 1000 guests (overrides allowed for CLI stress)."""
+    """DATA UI: 10 hosts x 1000 guests (overrides allowed for CLI stress)."""

    if node_count < 1 or resource_count < 1:
        raise ValueError("large profile counts must be positive")
@@ -1971,22 +1971,22 @@ def large_profile(*, node_count: int = 10, resource_count: int = 1_000) -> SeedP
        overrides={
            "nodes": node_count,
            "guests": resource_count,
-            "osds": max(10, int(round(100 * scale))),
-            "ha": max(3, int(round(12 * scale))),
-            "tasks": max(12, int(round(50 * scale))),
-            "pool_members": max(12, min(resource_count, int(round(50 * scale)))),
+            "osds": max(10, round(100 * scale)),
+            "ha": max(3, round(12 * scale)),
+            "tasks": max(12, round(50 * scale)),
+            "pool_members": max(12, min(resource_count, round(50 * scale))),
        },
    )


 def big_profile() -> SeedProfile:
-    """DATA UI: 20 hosts × 2000 guests."""
+    """DATA UI: 20 hosts x 2000 guests."""

    return _sized_cluster_profile("big")


 def medium_profile() -> SeedProfile:
-    """Compatibility alias for the small (3×50) cluster."""
+    """Compatibility alias for the small (3x50) cluster."""

    profile = small_profile()
    return SeedProfile("medium", profile.nodes, profile.resources, profile.tasks)
@@ -27,9 +27,11 @@ page. Russian mirrors live under [`ru/`](ru/README.md).
 | [FAQ](faq.md) | Short answers |
 | [Architecture](architecture.md) | Component boundaries |
 | [Compatibility](compatibility.md) | Evidence model and release matrix |
+| [Testing](testing.md) | Suites layout, Make targets, latest results |

 Runnable cookbooks: [`examples/`](../examples/README.md).
 Integration suites: [`pulumi-tests/`](../pulumi-tests/README.md).
+Test map and latest pass notes: [Testing](testing.md).

 ## See also

@@ -19,6 +19,7 @@ Yes. That is a primary use case. Pin the API major, seed a profile, and point
 clients at **HTTP `:8006`** (Compose) or your Ingress **HTTPS** hostname on
 Kubernetes. See [Clients](clients.md). For the
 Pulumi surface suite see [`pulumi-tests/`](../pulumi-tests/README.md).
+Full test map and latest recorded results: [Testing](testing.md).

 ## Why do some OpenID / LDAP / ACME / Ceph calls “succeed” without remotes?

@@ -27,9 +27,11 @@
 | [FAQ](faq.md) | Краткие ответы |
 | [Архитектура](architecture.md) | Границы компонентов |
 | [Совместимость](compatibility.md) | Модель evidence и матрица релизов |
+| [Тесты](testing.md) | Карта наборов, Make-цели, последние результаты |

 Исполняемые cookbook'и: [`examples/`](../../examples/README.ru.md).
 Интеграционные наборы: [`pulumi-tests/`](../../pulumi-tests/README.ru.md).
+Карта тестов и последние прогоны: [Тесты](testing.md).

 ## См. также

@@ -19,6 +19,7 @@ PostgreSQL, а не процессы KVM/LXC.
 Да. Это один из основных сценариев. Закрепите мажор API, загрузите профиль seed и
 направьте клиентов на **HTTP `:8006`** (Compose) или Ingress **HTTPS** в Kubernetes. См. [Клиенты](clients.md).
 Набор Pulumi surface — [`pulumi-tests/`](../../pulumi-tests/README.ru.md).
+Карта тестов и последние результаты: [Тесты](testing.md).

 ## Почему некоторые вызовы OpenID / LDAP / ACME / Ceph «успешны» без внешних систем?

@@ -0,0 +1,52 @@
+**Language / Язык:** [English](../testing.md) | [Русский](testing.md)
+
+# Тесты
+
+Где живут наборы и как их запускать.
+
+## Структура
+
+| Путь | Назначение |
+|---|---|
+| `tests/unit/` | Офлайн unit-тесты (handlers, контракты, seed, workers, …) |
+| `tests/integration/` | Тесты с PostgreSQL (миграции, lease задач) |
+| `tests/compatibility/` | Surface probe, group smoke, proxmoxer HTTPS, verified-surface |
+| `pulumi-tests/` | Pulumi surface (majors 6–9) + lifecycle `pulumi-proxmoxve`; HTML-отчёт |
+
+Seed: `app/simulation/seed.py` (`lab` для in-process surface CI, `small` для Compose / Pulumi).
+
+## Цели Make
+
+```bash
+make test-unit              # pytest: только unit
+make test-integration       # pytest -m integration (нужен Postgres)
+make test-surface           # каждый объявленный метод majors 6–9
+make test-compatibility     # proxmoxer smoke (Compose --profile tls)
+make ci                     # ruff + mypy + offline pytest + surface
+make ci-all                 # ci + остальной integration + proxmoxer
+make pulumi-tests           # Pulumi surface + lifecycle; HTML-отчёт
+```
+
+GitHub Actions на push/PR в `main` запускает `make ci` и проверку Compose/Helm.
+Перед крупными API-изменениями локально: `make ci-all` и `make pulumi-tests`.
+
+Отчёт Pulumi: `pulumi-tests/pulumi/reports/report.html` — см.
+[`pulumi-tests/README.ru.md`](../../pulumi-tests/README.ru.md).
+
+## Последний локальный прогон (2026-07-18)
+
+Полный прогон без ограничений на этом checkout:
+
+| Gate | Результат |
+|---|---|
+| `make ci-all` | **PASS** (exit 0) |
+| — unit (offline pytest) | 193 passed |
+| — surface probe (majors 6–9) | PASS |
+| — integration | 12 passed |
+| — proxmoxer compatibility | PASS |
+| `make pulumi-tests` | **PASS** (exit 0), ~28 с |
+| — lifecycle (`pulumi-proxmoxve`) | PASS |
+| — surface PVE 6.4-15 / 7.4-16 / 8.4.5 / 9.2.3 | PASS, `critical=0` |
+| — coverage | **2324 / 2324** методов majors 6–9 |
+
+Артефакты: `pulumi-tests/pulumi/reports/report.html` (также `results.json`, `junit.xml`).
@@ -0,0 +1,52 @@
+**Language / Язык:** [English](testing.md) | [Русский](ru/testing.md)
+
+# Testing
+
+How the repository is tested and where each suite lives.
+
+## Layout
+
+| Location | Role |
+|---|---|
+| `tests/unit/` | Offline unit tests (handlers, contracts, seed, workers, …) |
+| `tests/integration/` | PostgreSQL-backed tests (migrations, durable task leases) |
+| `tests/compatibility/` | Surface probe, group smoke, proxmoxer HTTPS, verified-surface ledgers |
+| `pulumi-tests/` | Pulumi surface matrix (majors 6–9) + `pulumi-proxmoxve` lifecycle; HTML report |
+
+Seed used by most suites: `app/simulation/seed.py` (`lab` for in-process surface CI, `small` for Compose / Pulumi).
+
+## Make targets
+
+```bash
+make test-unit              # pytest: unit only
+make test-integration       # pytest -m integration (needs Postgres)
+make test-surface           # every declared method on majors 6–9
+make test-compatibility     # proxmoxer smoke (Compose --profile tls)
+make ci                     # ruff + mypy + offline pytest + surface
+make ci-all                 # ci + remaining integration + proxmoxer
+make pulumi-tests           # Pulumi surface + lifecycle; writes HTML report
+```
+
+GitHub Actions on push/PR to `main` runs `make ci` plus Compose/Helm validation.
+Run `make ci-all` and `make pulumi-tests` locally before larger API changes.
+
+Pulumi report (after a suite run):
+`pulumi-tests/pulumi/reports/report.html` — see [`pulumi-tests/README.md`](../pulumi-tests/README.md).
+
+## Latest local results (2026-07-18)
+
+Full unrestricted run on this checkout:
+
+| Gate | Result |
+|---|---|
+| `make ci-all` | **PASS** (exit 0) |
+| — unit (offline pytest) | 193 passed |
+| — surface probe (majors 6–9) | PASS |
+| — integration | 12 passed |
+| — proxmoxer compatibility | PASS |
+| `make pulumi-tests` | **PASS** (exit 0), suite ~28s |
+| — lifecycle (`pulumi-proxmoxve`) | PASS |
+| — surface PVE 6.4-15 / 7.4-16 / 8.4.5 / 9.2.3 | PASS, `critical=0` |
+| — coverage | **2324 / 2324** methods across majors 6–9 |
+
+Report artifact: `pulumi-tests/pulumi/reports/report.html` (also `results.json`, `junit.xml`).
@@ -58,7 +58,7 @@ CSRF=$(echo "$TICKET" | jq -r .data.CSRFPreventionToken)
 curl -s -b /tmp/pve.ck http://localhost:8006/api2/json/version | jq .
 curl -s -b /tmp/pve.ck -H "CSRFPreventionToken: $CSRF" \
  -d 'vmid=100&name=demo' \
-  http://localhost:8006/api2/json/nodes/pve01/qemu | jq .
+  http://localhost:8006/api2/json/nodes/pve1/qemu | jq .
 ```

 Open the HTML report: `pulumi-tests/pulumi/reports/report.html`.
@@ -67,4 +67,12 @@ Provider env (set by Compose): `PROXMOX_VE_ENDPOINT=https://tls-gateway:8443/`
 (internal suite TLS — `pulumi-proxmoxve` rejects `http://`), plus username/password
 /`INSECURE`. Surface probe uses `API_URL=http://simulator:8006`. Host lab URL
 remains plain `http://localhost:8006/` (Kubernetes HTTPS is Ingress-only).
-Seed for CI/suite: `SEED_PROFILE=small` (Compose default).
+Seed for CI/suite: `SEED_PROFILE=small` (Compose default). Default node name in
+that profile is **`pve1`** (not `pve01`).
+
+## Latest local results (2026-07-18)
+
+`make pulumi-tests` → **Suite PASS** (~28s): lifecycle OK; surface majors
+6.4-15 / 7.4-16 / 8.4.5 / 9.2.3 with `critical=0`; coverage **2324/2324**.
+Report: `pulumi/reports/report.html`. Broader CI notes:
+[docs/testing.md](../docs/testing.md).
@@ -58,7 +58,7 @@ CSRF=$(echo "$TICKET" | jq -r .data.CSRFPreventionToken)
 curl -s -b /tmp/pve.ck http://localhost:8006/api2/json/version | jq .
 curl -s -b /tmp/pve.ck -H "CSRFPreventionToken: $CSRF" \
  -d 'vmid=100&name=demo' \
-  http://localhost:8006/api2/json/nodes/pve01/qemu | jq .
+  http://localhost:8006/api2/json/nodes/pve1/qemu | jq .
 ```

 Отчёт: `pulumi-tests/pulumi/reports/report.html`.
@@ -67,4 +67,12 @@ Env провайдера (Compose): `PROXMOX_VE_ENDPOINT=https://tls-gateway:844
 (внутренний TLS suite — `pulumi-proxmoxve` не принимает `http://`), плюс
 username/password/`INSECURE`. Surface probe: `API_URL=http://simulator:8006`.
 Хостовый lab URL — plain `http://localhost:8006/` (HTTPS в K8s — только Ingress).
-Seed для CI/suite: `SEED_PROFILE=small` (default Compose).
+Seed для CI/suite: `SEED_PROFILE=small` (default Compose). Имя ноды в этом
+профиле — **`pve1`** (не `pve01`).
+
+## Последний локальный прогон (2026-07-18)
+
+`make pulumi-tests` → **Suite PASS** (~28 с): lifecycle OK; surface majors
+6.4-15 / 7.4-16 / 8.4.5 / 9.2.3 с `critical=0`; coverage **2324/2324**.
+Отчёт: `pulumi/reports/report.html`. Общий CI:
+[docs/ru/testing.md](../docs/ru/testing.md).
@@ -5,7 +5,7 @@ x-test-env: &test-env
  API_USER: ${API_USER:-root@pam}
  API_PASSWORD: ${API_PASSWORD:-secret}
  API_TOKEN: ${API_TOKEN:-root@pam!automation=automation-secret}
-  PVE_NODE: ${PVE_NODE:-pve01}
+  PVE_NODE: ${PVE_NODE:-pve1}
  PVE_STORAGE: ${PVE_STORAGE:-local-lvm}
  PVE_BRIDGE: ${PVE_BRIDGE:-vmbr0}
  API_TIMEOUT: ${API_TIMEOUT:-120}
@@ -0,0 +1,3 @@
+encryptionsalt: v1:dHE0479O4C4=:v1:qLAPiOnywZwAY0nr:lwYFJLsSb1ayad79xjxO05Tvn5EK0w==
+config:
+  hx-lifecycle:smoke: "0"
@@ -0,0 +1,3 @@
+encryptionsalt: v1:Ttb4rpyx/lw=:v1:jISbowPFxMmrEKuc:YSoO8SdSx3symHxQE0vjSEQ2Mf/AyQ==
+config:
+  hx-lifecycle:smoke: "0"
@@ -18,7 +18,7 @@ from pvelib.api import Pve
 cfg = pulumi.Config()
 smoke = (cfg.get("smoke") or os.environ.get("SMOKE_ONLY") or "0") == "1"

-node = os.environ.get("PVE_NODE", "pve01")
+node = os.environ.get("PVE_NODE", "pve1")
 endpoint = (
    os.environ.get("PROXMOX_VE_ENDPOINT") or "https://tls-gateway:8443/"
 ).rstrip("/") + "/"
@@ -15,7 +15,7 @@ class Pve:
        root = os.environ.get("API_URL", "http://simulator:8006").rstrip("/")
        self.root = root
        self.base = root + "/api2/json"
-        self.node = os.environ.get("PVE_NODE", "pve01")
+        self.node = os.environ.get("PVE_NODE", "pve1")
        self.storage = os.environ.get("PVE_STORAGE", "local-lvm")
        self.bridge = os.environ.get("PVE_BRIDGE", "vmbr0")
        self._c = httpx.Client(base_url=self.base, timeout=120.0)
@@ -44,7 +44,7 @@ _UPID_RE = re.compile(
 )

 _PATH_PARAM_EXAMPLES: dict[str, object] = {
-    "node": "pve01",
+    "node": "pve1",
    "vmid": 100,
    "storage": "local",
    "pool": "testpool",
@@ -53,7 +53,7 @@ _PATH_PARAM_EXAMPLES: dict[str, object] = {
    "realm": "pam",
    "group": "admins",
    "role": "Administrator",
-    "upid": "UPID:pve01:00000001:00000001:65000001:qmstart:100:root@pam:",
+    "upid": "UPID:pve1:00000001:00000001:65000001:qmstart:100:root@pam:",
    "snapname": "snap1",
    "volume": "local:100/vm-100-disk-0.qcow2",
    "disk": "scsi0",
@@ -89,7 +89,7 @@ _EXTRA_PATH: dict[str, object] = {
    "cidr": "10.0.0.0/24",
    "tokenid": "automation",
    "fabric_id": "fab1",
-    "node_id": "pve01",
+    "node_id": "pve1",
    "url_seq": "1",
    "route-map-id": "rm1",
    "order": "10",
@@ -33,7 +33,7 @@ def test_proxmoxer_read_and_qemu_task_flow() -> None:

    assert proxmox.version.get()["version"] == "9.2.3"
    assert any(node["node"] == "pve1" for node in proxmox.nodes.get())
-    assert any(vm["vmid"] == 100 for vm in proxmox.nodes("pve1").qemu.get())
+    assert any(vm["vmid"] == 103 for vm in proxmox.nodes("pve1").qemu.get())

    token_api = ProxmoxAPI(
        os.environ["PROXMOXER_HOST"],
@@ -55,9 +55,9 @@ def test_proxmoxer_read_and_qemu_task_flow() -> None:
    )
    assert readonly_api.nodes.get()
    assert readonly_api.nodes("pve1").status.get()["status"] == "online"
-    assert readonly_api.nodes("pve1").qemu("100").config.get()["vmid"] == 100
+    assert readonly_api.nodes("pve1").qemu("103").config.get()["vmid"] == 103
    with pytest.raises(ResourceException) as denied:
-        readonly_api.nodes("pve1").qemu("100").status.start.post()
+        readonly_api.nodes("pve1").qemu("103").status.start.post()
    assert denied.value.status_code == 403

    token_endpoint = proxmox.access.users("root@pam").token("ephemeral")
@@ -88,7 +88,7 @@ def test_proxmoxer_read_and_qemu_task_flow() -> None:
        token_value=os.getenv("PROXMOXER_STORAGE_TOKEN_SECRET", "storage-secret"),
        verify_ssl=False,
    )
-    for vmid in ("100", "999999"):
+    for vmid in ("103", "999999"):
        with pytest.raises(ResourceException) as hidden:
            storage_api.nodes("pve1").qemu(vmid).config.get()
        assert hidden.value.status_code == 403
@@ -176,7 +176,7 @@ def test_proxmoxer_read_and_qemu_task_flow() -> None:
            token_value=os.getenv("PROXMOXER_OPERATOR_TOKEN_SECRET", "operator-secret"),
            verify_ssl=False,
        )
-        status_resource = operator_api.nodes("pve1").qemu("100").status
+        status_resource = operator_api.nodes("pve1").qemu("103").status

        def run(operation: str, expected: str) -> None:
            upid = status_resource(operation).post()
@@ -86,11 +86,24 @@ async def test_backup_jobs_crud_and_shapes() -> None:
    pool = BackupPool()
    http = _request(pool)

-    listed = await registry.get("/cluster/backup", "GET")(http, {"values": {}})
+    list_jobs = registry.get("/cluster/backup", "GET")
+    create_job = registry.get("/cluster/backup", "POST")
+    backup_info = registry.get("/cluster/backup-info", "GET")
+    not_backed_up = registry.get("/cluster/backup-info/not-backed-up", "GET")
+    included_volumes = registry.get("/cluster/backup/{id}/included_volumes", "GET")
+    delete_job = registry.get("/cluster/backup/{id}", "DELETE")
+    assert list_jobs is not None
+    assert create_job is not None
+    assert backup_info is not None
+    assert not_backed_up is not None
+    assert included_volumes is not None
+    assert delete_job is not None
+
+    listed = await list_jobs(http, {"values": {}})
    assert listed[0]["id"] == "backup-daily"
    assert listed[0]["schedule"] == "0 2 * * *"

-    await registry.get("/cluster/backup", "POST")(
+    await create_job(
        http,
        {
            "values": {
@@ -103,24 +116,19 @@ async def test_backup_jobs_crud_and_shapes() -> None:
    )
    assert "backup-weekly" in pool.metadata["backup_jobs"]

-    info = await registry.get("/cluster/backup-info", "GET")(http, {"values": {}})
+    info = await backup_info(http, {"values": {}})
    assert info == [{"subdir": "not-backed-up"}]

-    missing = await registry.get("/cluster/backup-info/not-backed-up", "GET")(http, {"values": {}})
-    assert missing == []  # 100 covered by daily; after weekly both covered? weekly adds 200
+    missing = await not_backed_up(http, {"values": {}})
    # daily covers 100, weekly covers 200 → none missing
    assert missing == []

-    included = await registry.get("/cluster/backup/{id}/included_volumes", "GET")(
-        http, {"values": {"id": "backup-daily"}}
-    )
+    included = await included_volumes(http, {"values": {"id": "backup-daily"}})
    assert included["children"][0]["id"] == 100
    assert included["children"][0]["children"][0]["id"] == "scsi0"

-    await registry.get("/cluster/backup/{id}", "DELETE")(http, {"values": {"id": "backup-weekly"}})
+    await delete_job(http, {"values": {"id": "backup-weekly"}})
    assert "backup-weekly" not in pool.metadata["backup_jobs"]
-    missing_after = await registry.get("/cluster/backup-info/not-backed-up", "GET")(
-        http, {"values": {}}
-    )
+    missing_after = await not_backed_up(http, {"values": {}})
    assert missing_after[0]["vmid"] == 200
    assert missing_after[0]["type"] == "lxc"
@@ -93,6 +93,9 @@ async def test_worker_retries_after_claim_failure() -> None:
        poll_seconds=0.001,
    )
    running = asyncio.create_task(worker.run())
+    for _ in range(200):
+        if repository.attempts > 1:
+            break
        await asyncio.sleep(0.01)
    worker.stop()
    await running