Merge remote-tracking branch 'origin/main' into fix/half-built-vm-28

# Conflicts:
#	CHANGELOG.md
#	tests/quality_baseline.json
This commit is contained in:
Christian Manivong committed 2026-11-02 00:33:52 +01:00
commit 0f1918f230
28 files changed
+2720 -162

No files matched your search

+66
View File
@@ -12,6 +12,31 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
on 3.12, so 3.10 and 3.11 are no longer tested (NAPALM/driver-standard#6).
### Added
- `get_clusters()` reports quorum `votes` per node (`/cluster/config/nodes`), the
guests HA manages in `ha_resources` (`/cluster/ha/resources`, the CRM state and
node from `/cluster/ha/status/current`, the ids of the PVE 9 HA rules that name
each from `/cluster/ha/rules`, `[]` on PVE 8), the vzdump jobs in `backup_jobs`
(`/cluster/backup`; a pool job lists the pool's guests from `/pools?poolid=`, or
`/pools/{pool}` before PVE 8.1; old `starttime`/`dow` jobs get the schedule PVE
itself derives) and per guest and storage the newest backup in `backups` (each
storage with backup content read once if shared, on each node if local; PBS
included). Each is best-effort: one that cannot be read is None and the rest of
the report stands.
- `get_vm_migration_preflight(name)`: a VM's nodes and blockers from
`GET .../qemu/{vmid}/migrate` in plain words, a local ISO in a CD drive as a
blocking resource, the cloud-init drive left out; a container's candidates are the
online members, its local volumes and unshared bind mounts read from its config.
An offline member is refused as `"node offline"` for both.
- `migrate_vm(name, target_node, online=True, target_storage=None, timeout=3600)`:
`POST .../{qemu|lxc}/{vmid}/migrate` (`online` for a running VM, `with-local-disks`
when it has local disks, restart mode for a running container), followed to the
end; a failure raises with the task log's `ERROR:` lines. A target that is not
another cluster member is refused before the API is called. A guest HA manages is
followed through the HA manager until it is on the target.
- `set_node_maintenance(node, enabled)`: `ha-manager crm-command node-maintenance`
over the driver's SSH session, for a cluster member only and quoted; raises
RuntimeError when SSH is not available. Needs the shapes and contracts of the
next napalm-device-types release for type checking (runtime imports none).
- `get_clusters()`: the cluster the node belongs to, always exactly one, read
from `/cluster/status`, `/cluster/resources`, each online node's status and
network, SDN vnets/zones and the HA manager. Every member is listed, an
@@ -40,6 +65,47 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
- `get_ipv6_neighbors_table()` reports an unknown age as `-1.0`, as `get_arp_table()`
already did (napalm-device-types' `ARP_AGE_UNKNOWN`), instead of `0.0`, which
means "confirmed just now" (NetOrk/netork#1327).
- `set_vlan()` finds the node's switch uplink again. Since 0.8.0 every call failed
with "no LLDP-detected switch uplink found", whatever LLDP reported (#37).
- `configure_snmp_agent()` completes: its "configure" step reported a failure
with empty output although `/etc/snmp/snmpd.conf` was written, so the firewall
and start steps never ran (#30).
- `create_vm_from_cloud_init(ssh_public_keys=...)` installs every key, one per
line, instead of only the first; a key containing `/` or `+` is no longer
refused by Proxmox (#23).
- `get_lldp_neighbors()` lists every neighbour of a port: a second neighbour on
the same port no longer overwrites the first, and a neighbour that sends no
system name is listed with an empty `hostname` instead of being dropped (#22).
- `set_hostname()` renames only the node's own names in `/etc/hosts`: other
hosts whose names start with the old one (`pve-backup`) and comments stay as
they are, and the old FQDN becomes the new one instead of a mix of both (#21).
- `set_hostname()` refuses a node that is part of a cluster (a
`/etc/pve/corosync.conf` or `/etc/corosync/corosync.conf` exists) with a
`CommandErrorException`, before changing anything: it moved
`/etc/pve/nodes/<old>` while the cluster still knew the node as `<old>`. Also
refused when the check cannot be read. A single node renames as before (#21).
- `apply_updates()` reports a failed upgrade as `success: False`, with apt-get's
exit status in `error`; it reported success whatever apt-get did (#20).
- `get_users()` lists the node's root and regular login accounts (uid 1000 and
up) instead of its system accounts (`daemon`, `bin`, …), and gives a Proxmox
user level 15 when the ACL grants it the Administrator role on `/`, directly
or through a group; `root@pam` is always 15. Before, practically every
Proxmox user was level 1 (#19).
- `get_ntp_stats()` reports chrony's sources, Proxmox VE's default, instead of
always returning an empty list: offset in ms, the poll interval in seconds,
the offset's standard deviation as `jitter`; chrony reports no delay, so
`delay` is 0.0. On ntpd, a peer without tally mark is no longer dropped (#18).
- The SSH login to the node takes an Ed25519 or ECDSA key in
`ssh_private_key_str`, not only RSA, and uses the key even when a password is
set too (the password stays the fallback). A key that cannot be read is
logged as a warning and the password used (#17).
- `destroy_vm()` returns once Proxmox has deleted the VM, not as soon as the
delete was queued, and raises when the delete fails (#15).
- `ping()` fills `results` with each reply's address and round-trip time; it
was always empty (#12).
- A failed cloud-image download in `create_vm_from_cloud_init()` says why
(wget's own message, such as "No space left on device") and removes its
partial `.tmp` file instead of leaving it to fill the node's disk (#3).
- A `create_vm_from_cloud_init()` that fails once the VM shell exists (disk import,
storage, NIC or snippet configuration, start) raises `VMProvisionIncomplete`, a
`RuntimeError` with the VMID and node of the VM it left behind and the failure as
+151
View File
@@ -0,0 +1,151 @@
"""get_clusters() for Proxmox VE: the backup jobs, and the backups on the storages.
Read from:
- ``GET /cluster/backup``: each vzdump job. ``enabled`` absent means enabled;
old jobs carry ``starttime``/``dow`` instead of ``schedule``, which PVE itself
turns into ``"<dow> <starttime>"`` (``PVE/API2/Backup.pm``). ``storage`` and
``mode`` are None where the job does not set them: a node's
``/etc/vzdump.conf`` then decides.
- a pool job's guests: ``GET /pools?poolid=<pool>`` (PVE 8.1 and later), else
``GET /pools/<pool>``, which older releases have;
- ``GET /nodes/<node>/storage/<storage>/content?content=backup`` for each storage
that holds backups (PBS answers the same): a shared one once, on the own node
if it sees it, else on the next node that does; a local one on every node
that has it.
Each list is best-effort: when part of it cannot be read it is None, because
half a list would say a guest has no backup, or a job no guest, when it may.
"""
from __future__ import annotations
import logging
from typing import TYPE_CHECKING, Any
from napalm_proxmox.pve_values import split_list, to_flag, to_int, vmid_list
if TYPE_CHECKING:
# Type-only: newer than the napalm_device_types floor in pyproject.toml.
from napalm_device_types import BackupDict, BackupJobDict, ClusterStorageDict
logger = logging.getLogger(__name__)
_JsonDict = dict[str, Any]
_ALL_DAYS = "mon,tue,wed,thu,fri,sat,sun"
_GUEST_TYPES = ("qemu", "lxc")
def _schedule(job: _JsonDict) -> str | None:
if job.get("schedule"):
return str(job["schedule"])
start = job.get("starttime")
if not start:
return None
dow = ",".join(split_list(job.get("dow")))
return str(start) if dow in ("", _ALL_DAYS) else f"{dow} {start}"
def _pool_info(api: Any, pool: str) -> _JsonDict:
try:
for entry in api.pools.get(poolid=pool) or []:
if entry.get("poolid") == pool:
return dict(entry)
except Exception as exc:
logger.debug("GET /pools?poolid=%s failed, trying /pools/%s: %s", pool, pool, exc)
return dict(api.pools(pool).get() or {})
def _pool_members(api: Any, pool: str) -> list[int]:
members = _pool_info(api, pool).get("members") or []
return [
int(m["vmid"])
for m in members
if m.get("type") in _GUEST_TYPES and m.get("vmid") is not None
]
def _job(api: Any, job: _JsonDict) -> BackupJobDict:
pool = job.get("pool") or None
return {
"id": str(job.get("id") or ""),
"enabled": to_flag(job.get("enabled", 1)),
"schedule": _schedule(job),
"storage": job.get("storage") or None,
"mode": job.get("mode") or None,
"all": to_flag(job.get("all")),
"vmids": _pool_members(api, str(pool)) if pool else vmid_list(job.get("vmid")),
"exclude": vmid_list(job.get("exclude")),
"pool": pool,
"node": job.get("node") or None,
}
def read_backup_jobs(api: Any) -> list[BackupJobDict] | None:
"""The cluster's backup jobs; None when they, or a pool's guests, cannot be read."""
try:
return [_job(api, job) for job in api.cluster.backup.get() or []]
except Exception as exc:
logger.debug("Cannot read the backup jobs: %s", exc)
return None
def _content(api: Any, node: str, storage: str) -> list[_JsonDict]:
return list(api.nodes(node).storage(storage).content.get(content="backup") or [])
def _first_answer(api: Any, storage: str, nodes: list[str]) -> list[_JsonDict]:
error: Exception | None = None
for node in nodes:
try:
return _content(api, node, storage)
except Exception as exc:
logger.debug("Cannot read backups on %s from %s: %s", storage, node, exc)
error = exc
if error is not None:
raise error
return []
def _storage_rows(api: Any, storage: ClusterStorageDict, local_node: str) -> list[_JsonDict]:
nodes = storage["nodes"]
if storage["shared"]:
return _first_answer(api, storage["name"], sorted(nodes, key=lambda n: n != local_node))
return [row for node in nodes for row in _content(api, node, storage["name"])]
def _merge(rows: list[_JsonDict], storage: str, merged: dict[tuple[int, str], BackupDict]) -> None:
for row in rows:
if row.get("vmid") is None or row.get("ctime") is None:
continue # not a vzdump/PBS backup of a guest
key, ctime = (int(row["vmid"]), storage), int(row["ctime"])
entry = merged.get(key)
if entry is None:
merged[key] = {
"vmid": key[0],
"storage": storage,
"time": ctime,
"size": to_int(row.get("size")),
"count": 1,
}
continue
entry["count"] += 1
if ctime > entry["time"]:
entry["time"], entry["size"] = ctime, to_int(row.get("size"))
def read_backups(
api: Any, storages: list[ClusterStorageDict], local_node: str
) -> list[BackupDict] | None:
"""One entry per guest and storage with its newest backup; None if a storage
that holds backups cannot be read."""
merged: dict[tuple[int, str], BackupDict] = {}
try:
for storage in storages:
if "backup" in storage["content"]:
_merge(_storage_rows(api, storage, local_node), storage["name"], merged)
except Exception as exc:
logger.debug("Cannot read the backups: %s", exc)
return None
return [merged[key] for key in sorted(merged, key=lambda k: (k[1], k[0]))]
+105
View File
@@ -0,0 +1,105 @@
"""get_clusters() for Proxmox VE: quorum votes and the guests the HA manager keeps.
Read from:
- ``GET /cluster/config/nodes``: corosync's nodelist, ``quorum_votes`` per node;
- ``GET /cluster/ha/resources``: each HA resource's ``sid`` (``vm:100``,
``ct:101``), requested ``state`` and, on PVE 8, ``group``;
- ``GET /cluster/ha/status/current``: per ``service`` entry the ``node`` and the
CRM's ``crm_state`` (only the verbose ``state`` while the CRM has not taken it
up yet);
- ``GET /cluster/ha/rules`` (PVE 9): each rule's ``rule`` id and its
``resources``, a comma-separated list of sids. PVE 8 has no such endpoint.
Each read is best-effort: what cannot be read is None, never a guess.
"""
from __future__ import annotations
import logging
from typing import TYPE_CHECKING, Any
from napalm_proxmox.pve_values import split_list
if TYPE_CHECKING:
# Type-only: newer than the napalm_device_types floor in pyproject.toml.
from napalm_device_types import HAResourceDict
logger = logging.getLogger(__name__)
_JsonDict = dict[str, Any]
#: HA resource type prefix -> guest kind.
_KINDS = {"vm": "qemu", "ct": "lxc"}
#: corosync counts a node without ``quorum_votes`` as one vote.
_DEFAULT_VOTES = 1
#: What HA keeps a resource at when its config names no state.
_DEFAULT_REQUESTED = "started"
def read_votes(api: Any) -> dict[str, int] | None:
"""Quorum votes per node name; None when the nodelist cannot be read."""
try:
rows = api.cluster.config.nodes.get() or []
except Exception as exc:
logger.debug("Cannot read the corosync nodelist: %s", exc)
return None
return {
str(row.get("node") or row.get("name")): int(row.get("quorum_votes", _DEFAULT_VOTES))
for row in rows
if row.get("node") or row.get("name")
}
def _services(api: Any) -> dict[str, _JsonDict] | None:
try:
rows = api.cluster.ha.status.current.get() or []
except Exception as exc:
logger.debug("Cannot read the HA status: %s", exc)
return None
return {str(r["sid"]): r for r in rows if r.get("type") == "service" and r.get("sid")}
def _rules(api: Any) -> dict[str, list[str]]:
"""Rule ids per sid. PVE 8 answers with an error: no rules there."""
try:
rows = api.cluster.ha.rules.get() or []
except Exception as exc:
logger.debug("No HA rules (PVE 8 has none): %s", exc)
return {}
by_sid: dict[str, list[str]] = {}
for rule in rows:
for sid in split_list(rule.get("resources")):
by_sid.setdefault(sid, []).append(str(rule.get("rule")))
return by_sid
def _resource(
row: _JsonDict, services: dict[str, _JsonDict] | None, rules: dict[str, list[str]]
) -> HAResourceDict | None:
sid = str(row.get("sid") or "")
prefix, _, vmid = sid.partition(":")
if prefix not in _KINDS or not vmid.isdigit():
return None
service = (services or {}).get(sid, {})
return {
"vmid": int(vmid),
"kind": _KINDS[prefix],
"requested": str(row.get("state") or _DEFAULT_REQUESTED),
"status": service.get("crm_state") or service.get("state") or None,
"node": service.get("node") or None,
"group": row.get("group") or None,
"rules": rules.get(sid, []),
}
def read_ha_resources(api: Any) -> list[HAResourceDict] | None:
"""Every guest HA manages, by vmid; None when the HA resources cannot be read."""
try:
rows = api.cluster.ha.resources.get() or []
except Exception as exc:
logger.debug("Cannot read the HA resources: %s", exc)
return None
services, rules = _services(api), _rules(api)
resources = [_resource(row, services, rules) for row in rows]
return sorted((r for r in resources if r is not None), key=lambda r: r["vmid"])
+22 -17
View File
@@ -13,7 +13,10 @@ cluster of one, without a name). Read from:
- ``GET /nodes/{node}/network`` of each online node: its bridges;
- ``GET /cluster/sdn/vnets`` and ``/cluster/sdn/zones``: the vnets, on the nodes
of their zone;
- ``GET /cluster/ha/status/manager_status``: which node is in maintenance.
- ``GET /cluster/ha/status/manager_status``: which node is in maintenance;
- quorum votes and HA resources: see ``cluster_ha``; backup jobs and backups:
see ``cluster_backups``. Each of those is best-effort and None when it cannot
be read, without failing the rest of the report.
"""
from __future__ import annotations
@@ -21,7 +24,12 @@ from __future__ import annotations
import logging
from typing import TYPE_CHECKING, Any, NamedTuple
from napalm_proxmox.cluster_backups import read_backup_jobs, read_backups
from napalm_proxmox.cluster_ha import read_ha_resources, read_votes
from napalm_proxmox.network_targets import bridge_targets, vnet_target
from napalm_proxmox.pve_values import split_list as _list
from napalm_proxmox.pve_values import to_flag as _flag
from napalm_proxmox.pve_values import to_int as _int
from napalm_proxmox.utils import cpu_flags
if TYPE_CHECKING:
@@ -51,22 +59,10 @@ class _Cpu(NamedTuple):
flags: list[str] | None = None
def _int(value: Any) -> int | None:
return None if value is None else int(value)
def _float(value: Any) -> float | None:
return None if value is None else float(value)
def _flag(value: Any) -> bool:
return bool(int(value or 0))
def _list(value: Any) -> list[str]:
return [part.strip() for part in str(value or "").split(",") if part.strip()]
def _size(rows: list[_JsonDict], key: str, shared: bool) -> int | None:
"""One storage's size over the nodes in *rows*: summed unless it is shared."""
values = [row.get(key) for row in rows]
@@ -151,14 +147,18 @@ class ProxmoxClusterMixin:
resources = self._api.cluster.resources.get() or []
head = next((s for s in status if s.get("type") == "cluster"), None)
members = self._members(status if head else [])
storages = _storages(resources)
return [
{
"id": None,
"name": (head.get("name") or None) if head else None,
"quorate": _flag(head.get("quorate")) if head else None,
"nodes": self._cluster_nodes(members, resources),
"storages": _storages(resources),
"storages": storages,
"networks": self._networks(members),
"ha_resources": read_ha_resources(self._api),
"backup_jobs": read_backup_jobs(self._api),
"backups": read_backups(self._api, storages, self._node_name),
}
]
@@ -166,8 +166,8 @@ class ProxmoxClusterMixin:
self, members: list[_Member], resources: list[_JsonDict]
) -> list[ClusterNodeDict]:
rows = {r.get("node"): r for r in resources if r.get("type") == "node"}
maintenance = self._maintenance()
return [self._cluster_node(m, rows.get(m.name, {}), maintenance) for m in members]
maintenance, votes = self._maintenance(), read_votes(self._api)
return [self._cluster_node(m, rows.get(m.name, {}), maintenance, votes) for m in members]
def _networks(self, members: list[_Member]) -> list[ClusterNetworkDict]:
online = [m for m in members if m.online]
@@ -225,7 +225,11 @@ class ProxmoxClusterMixin:
)
def _cluster_node(
self, member: _Member, row: _JsonDict, maintenance: dict[str, bool] | None
self,
member: _Member,
row: _JsonDict,
maintenance: dict[str, bool] | None,
votes: dict[str, int] | None,
) -> ClusterNodeDict:
if not member.online:
row, cpu = {}, _Cpu()
@@ -243,6 +247,7 @@ class ProxmoxClusterMixin:
"cpu_usage": _float(row.get("cpu")),
"memory_total": _int(row.get("maxmem")),
"memory_used": _int(row.get("mem")),
"votes": None if votes is None else votes.get(member.name),
}
def _vnets(self, members: list[str]) -> list[ClusterNetworkDict]:
+24 -7
View File
@@ -63,6 +63,7 @@ from napalm_proxmox.routing_mixin import ProxmoxRoutingMixin
from napalm_proxmox.sdn_mixin import ProxmoxSDNMixin
from napalm_proxmox.system_mixin import ProxmoxSystemMixin
from napalm_proxmox.vm_contract_mixin import ProxmoxVMContractMixin
from napalm_proxmox.vm_migrate_mixin import ProxmoxVMMigrateMixin
from napalm_proxmox.vm_mixin import ProxmoxVMMixin
from napalm_proxmox.vm_provision_mixin import ProxmoxVMProvisionMixin
from napalm_proxmox.vm_snapshot_mixin import ProxmoxVMSnapshotMixin
@@ -89,6 +90,7 @@ class ProxmoxDriver(
ProxmoxVMSnapshotMixin,
ProxmoxVMProvisionMixin,
ProxmoxClusterMixin,
ProxmoxVMMigrateMixin,
ProxmoxRoutingMixin,
ProxmoxSystemMixin,
KernelFactsMixin,
@@ -343,17 +345,17 @@ class ProxmoxDriver(
import paramiko
if self._ssh_client is None:
ssh_pass = self._ssh_password or self.password
# A key is tried first; paramiko falls back to the password.
connect_kwargs: _JsonDict = {
"username": self._ssh_username or self.username,
"timeout": self.timeout,
}
if self._ssh_key and not ssh_pass:
from io import StringIO as _StringIO
connect_kwargs["pkey"] = paramiko.RSAKey.from_private_key(_StringIO(self._ssh_key))
else:
connect_kwargs["password"] = ssh_pass
password = self._ssh_password or self.password
if password:
connect_kwargs["password"] = password
pkey = _load_private_key(self._ssh_key) if self._ssh_key else None
if pkey is not None:
connect_kwargs["pkey"] = pkey
client = paramiko.SSHClient()
try:
HostKeyCheck(self.optional_args, self.hostname, 22).connect(
@@ -511,3 +513,18 @@ class ProxmoxDriver(
def cli(self, commands: list[str], encoding: str = "text") -> dict[str, str]:
"""Execute a list of shell commands and return their outputs."""
return {cmd: self._exec_ssh_command(cmd) for cmd in commands}
def _load_private_key(text: str) -> Any: # paramiko.PKey; paramiko is imported lazily
"""The SSH private key in *text*, whatever its type, or None when it cannot be read."""
import io
import paramiko
for key_class in (paramiko.Ed25519Key, paramiko.ECDSAKey, paramiko.RSAKey):
try:
return key_class.from_private_key(io.StringIO(text))
except (paramiko.SSHException, ValueError):
continue
logger.warning("The SSH private key could not be read; logging in with the password")
return None
+12 -12
View File
@@ -26,26 +26,26 @@ class ProxmoxLLDPMixin:
"""Mixin providing LLDP NAPALM methods."""
def get_lldp_neighbors(self) -> dict[str, list[_JsonDict]]:
"""Return LLDP neighbours (requires lldpd on the Proxmox node)."""
"""Return LLDP neighbours (requires lldpd on the Proxmox node).
lldpcli prints one ``Interface:`` block per neighbour, so a port with
several neighbours appears once for each of them.
"""
result: dict[str, list[_JsonDict]] = {}
raw = self._exec_ssh_command("lldpcli show neighbors summary 2>/dev/null || true")
current_iface = ""
current: _JsonDict | None = None
for line in raw.splitlines():
m_iface = re.match(r"^\s*Interface:\s+(\S+?),?\s", line)
if m_iface:
current_iface = m_iface.group(1)
result.setdefault(current_iface, [])
current = {"hostname": "", "port": ""}
result.setdefault(m_iface.group(1), []).append(current)
continue
m_sys = re.match(r"^\s*SysName:\s+(.+)", line)
m_port = re.match(r"^\s*PortID:\s+\S+\s+(.+)", line)
if m_sys and current_iface:
hostname = m_sys.group(1).strip()
if result[current_iface]:
result[current_iface][-1]["hostname"] = hostname
else:
result[current_iface].append({"hostname": hostname, "port": ""})
if m_port and current_iface and result[current_iface]:
result[current_iface][-1]["port"] = m_port.group(1).strip()
if m_sys and current is not None:
current["hostname"] = m_sys.group(1).strip()
if m_port and current is not None:
current["port"] = m_port.group(1).strip()
return result
def get_lldp_neighbors_detail(self, interface: str = "") -> dict[str, list[_JsonDict]]:
+115
View File
@@ -0,0 +1,115 @@
"""Reading Proxmox VE's answers about a guest's migration: refusal reasons, local
volumes and resources, task log errors.
- ``GET /nodes/{node}/qemu/{vmid}/migrate`` (qemu-server's
``migrate_vm_precondition``): ``not_allowed_nodes`` maps a node to its
blockers -- ``unavailable_storages``, ``unavailable-resources`` (mappings the
node lacks), ``blocking-ha-resources`` (PVE 9 HA rules); ``local_disks`` lists
volumes on non-shared storage, CD drives included; ``local_resources``
(``hostpci0``, ``usb0``, ``clipboard=vnc``) block the migration.
- A container has no such endpoint; its config says what it keeps locally
(``PVE/LXC/Migrate.pm``): a bind or device mount point that is not marked
``shared`` stops the migration, a volume on non-shared storage is copied.
- A task log is rows of ``{"n": <line>, "t": <text>}``; a failure writes
``ERROR: ...`` lines.
"""
from __future__ import annotations
import re
from typing import Any
from napalm_proxmox.pve_values import to_flag
_JsonDict = dict[str, Any]
_ERROR = re.compile(r"\bERROR: (.*\S)")
#: The cloud-init drive is a CD drive the migration regenerates on the target.
_CLOUDINIT = re.compile(r"vm-\d+-cloudinit")
_CT_VOLUME_KEY = re.compile(r"^(rootfs|mp\d+|unused\d+)$")
def task_errors(log: list[_JsonDict]) -> list[str]:
"""The ``ERROR:`` lines of a task log, without their timestamps, each once."""
errors: list[str] = []
for row in log:
match = _ERROR.search(str(row.get("t") or ""))
if match and match.group(1) not in errors:
errors.append(match.group(1))
return errors
def upid_field(upid: str, index: int) -> str:
"""A field of ``UPID:node:pid:pstart:starttime:type:id:user:``; "" if absent."""
parts = upid.split(":")
return parts[index] if len(parts) > index else ""
def _names(kind: str, items: list[Any]) -> str:
names = ", ".join(str(item) for item in items)
return f"{kind} {names} is" if len(items) == 1 else f"{kind}s {names} are"
def _ha_block(item: Any) -> str:
if not isinstance(item, dict):
return f"HA resource {item} blocks it"
cause = f" ({item['cause']} rule)" if item.get("cause") else ""
return f"HA resource {item.get('sid')} blocks it{cause}"
def _reason(key: str, value: Any) -> str:
items = value if isinstance(value, list) else [value]
if key == "unavailable_storages":
return f"{_names('storage', items)} not available there"
if key == "unavailable-resources":
return f"{_names('mapped resource', items)} not available there"
if key == "blocking-ha-resources":
return "; ".join(_ha_block(item) for item in items)
return f"{key}: {', '.join(str(item) for item in items)}"
def refusal(blockers: Any) -> str:
"""One node's ``not_allowed_nodes`` entry in plain words."""
if not isinstance(blockers, dict) or not blockers:
return "refused by Proxmox VE"
return "; ".join(_reason(key, blockers[key]) for key in sorted(blockers))
def qemu_local(answer: _JsonDict) -> tuple[list[str], list[str]]:
"""``(local_disks, local_resources)`` of a VM's migrate precondition answer.
A local ISO in a CD drive stops the migration, so it is a resource
(``cdrom=<volid>``, after PVE's own ``clipboard=vnc``), not a disk.
"""
disks: list[str] = []
resources = [str(r) for r in answer.get("local_resources") or []]
for disk in answer.get("local_disks") or []:
volid = str(disk.get("volid") or "")
if not volid or _CLOUDINIT.search(volid):
continue
if to_flag(disk.get("cdrom")):
resources.append(f"cdrom={volid}")
else:
disks.append(volid)
return sorted(disks), resources
def lxc_local(config: _JsonDict, shared: dict[str, bool]) -> tuple[list[str], list[str]]:
"""``(local_disks, local_resources)`` of a container's config.
*shared* says per storage whether it is shared; a storage it does not know is
taken as local, which blocks nothing.
"""
disks: list[str] = []
resources: list[str] = []
for key in sorted(config):
if not _CT_VOLUME_KEY.match(key):
continue
volume, *options = str(config[key]).split(",")
volume = volume.removeprefix("volume=")
if volume.startswith("/"):
if "shared=1" not in options:
resources.append(key)
elif not shared.get(volume.split(":", 1)[0], False):
disks.append(volume)
return sorted(disks), resources
+113
View File
@@ -0,0 +1,113 @@
"""NTP statistics (``get_ntp_stats``) from chrony or ntpd, in NAPALM's NTPStats shape.
Proxmox VE ships chrony; ntpd is read only where chrony is not installed.
chrony is read in its CSV mode (``chronyc -c``): ``sources`` names each source
with its state, ``sourcestats`` adds the spread of its offset, reported as
jitter. chrony reports no delay per source, so ``delay`` is 0.0. Times are in
milliseconds and the poll interval in seconds, as ``ntpq -pn`` gives them.
"""
from __future__ import annotations
import re
from typing import Any
CHRONY_SOURCES_MARK = "CHRONY_SOURCES"
CHRONY_SOURCESTATS_MARK = "CHRONY_SOURCESTATS"
#: chrony when it is installed, ntpd's ntpq otherwise.
NTP_STATS_COMMAND = (
"if command -v chronyc >/dev/null 2>&1; then "
f"echo {CHRONY_SOURCES_MARK}; chronyc -n -c sources; "
f"echo {CHRONY_SOURCESTATS_MARK}; chronyc -n -c sourcestats; "
"else ntpq -pn; fi 2>/dev/null"
)
#: chrony's source mode as ntpq's type: server (unicast), peer (symmetric), refclock (local).
_CHRONY_TYPES = {"^": "u", "=": "s", "#": "l"}
_MS_PER_S = 1000.0
#: ``*192.168.1.1 10.0.0.1 4 u 12 64 377 1.234 0.567 0.123``; the first
#: column is the tally mark, a space for a peer without one.
_NTPQ_LINE = re.compile(
r"^([ *+#ox.-])(\S+)\s+(\S+)\s+(\d+)\s+(\S+)\s+(\S+)\s+(\d+)\s+(\d+)"
r"\s+([-\d.]+)\s+([-\d.]+)\s+([-\d.]+)"
)
def parse_ntp_stats(output: str) -> list[dict[str, Any]]:
"""The NTPStats list in the output of :data:`NTP_STATS_COMMAND`."""
if output.lstrip().startswith(CHRONY_SOURCES_MARK):
return parse_chrony(output)
return parse_ntpq(output)
def parse_chrony(output: str) -> list[dict[str, Any]]:
"""Sources and their statistics from ``chronyc -n -c sources`` / ``sourcestats``."""
sources, _, sourcestats = output.partition(CHRONY_SOURCESTATS_MARK)
jitter = _chrony_jitter(sourcestats)
stats = []
for line in sources.splitlines():
fields = line.strip().split(",")
if len(fields) < 10:
continue
mode, state, name, stratum, poll, reach, last_rx, _, offset, _ = fields[:10]
try:
stats.append(
{
"remote": name,
"referenceid": "",
"synchronized": state == "*",
"stratum": int(stratum),
"type": _CHRONY_TYPES.get(mode, ""),
"when": last_rx,
"hostpoll": 2 ** int(poll),
"reachability": int(reach),
"delay": 0.0,
"offset": float(offset) * _MS_PER_S,
"jitter": jitter.get(name, 0.0),
}
)
except ValueError:
continue
return stats
def _chrony_jitter(sourcestats: str) -> dict[str, float]:
"""Each source's offset standard deviation in ms (``sourcestats``' last column)."""
jitter = {}
for line in sourcestats.splitlines():
fields = line.strip().split(",")
if len(fields) < 8:
continue
try:
jitter[fields[0]] = float(fields[7]) * _MS_PER_S
except ValueError:
continue
return jitter
def parse_ntpq(output: str) -> list[dict[str, Any]]:
"""Peers from ``ntpq -pn``."""
stats = []
for line in output.splitlines():
m = _NTPQ_LINE.match(line)
if not m:
continue
tally, remote, refid, stratum, kind, when, poll, reach, delay, offset, jitter = m.groups()
stats.append(
{
"remote": remote,
"referenceid": refid,
"synchronized": tally == "*",
"stratum": int(stratum),
"type": kind,
"when": when,
"hostpoll": int(poll),
"reachability": int(reach),
"delay": float(delay),
"offset": float(offset),
"jitter": float(jitter),
}
)
return stats
+27
View File
@@ -0,0 +1,27 @@
"""How the Proxmox VE API writes plain values: numbers as strings, flags as 0/1, lists
as comma-separated strings."""
from __future__ import annotations
import re
from typing import Any
_LIST_SEPARATORS = re.compile(r"[\s,;\0]+")
def to_int(value: Any) -> int | None:
return None if value is None else int(value)
def to_flag(value: Any) -> bool:
return bool(int(value or 0))
def split_list(value: Any) -> list[str]:
"""``"a,b"`` -> ``["a", "b"]``; PVE also separates with ``;``, blanks or NUL."""
return [part for part in _LIST_SEPARATORS.split(str(value or "")) if part]
def vmid_list(value: Any) -> list[int]:
"""``"100,101"`` -> ``[100, 101]``; what is not a number is left out."""
return [int(part) for part in split_list(value) if part.isdigit()]
+7 -3
View File
@@ -23,6 +23,10 @@ import napalm.base.constants as C
_JsonDict = dict[str, Any]
#: One reply in iputils' ping output, by address or by name with the address
#: in brackets: "64 bytes from dns.google (8.8.4.4): icmp_seq=1 ttl=118 time=10.5 ms".
_PING_REPLY = re.compile(r"bytes from (?:\S+ \()?([0-9a-fA-F.:]+)\)?: .*?time=([\d.]+) ms")
class ProxmoxRoutingMixin:
"""Mixin providing routing NAPALM methods (get_route_to, ping, traceroute)."""
@@ -90,9 +94,9 @@ class ProxmoxRoutingMixin:
rtt_std = float(m_rtt.group(4)) if m_rtt else 0.0
# Individual probe lines
probes = []
for m_probe in re.finditer(r"icmp_seq=\d+.*?time=([\d.]+) ms.*?from ([\d.a-fA-F:]+)", raw):
probes.append({"ip_address": m_probe.group(2), "rtt": float(m_probe.group(1))})
probes = [
{"ip_address": m.group(1), "rtt": float(m.group(2))} for m in _PING_REPLY.finditer(raw)
]
return {
"success": {
+11
View File
@@ -33,6 +33,17 @@ _JsonDict = dict[str, Any]
class ProxmoxSDNMixin:
"""Mixin providing SDN, VLAN and network-instance NAPALM methods."""
#: Name prefixes of Proxmox-internal and runtime virtual interfaces, never a switch uplink.
_VIRTUAL_IFACE_PREFIXES: tuple[str, ...] = (
"fwpr", # Proxmox firewall proxy veth
"fwln", # Proxmox firewall link veth
"tap", # VM tap devices
"veth", # generic veth pairs (containers)
"virbr", # libvirt bridges
"docker", # Docker virtual interfaces
"lxcbr", # LXC bridges
)
def _get_sdn_zones(self) -> list[_JsonDict]:
try:
return self._api.cluster.sdn.zones.get() or [] # type: ignore[union-attr]
+142 -99
View File
@@ -26,8 +26,11 @@ import re
import shlex
from typing import Any
from napalm.base.exceptions import CommandErrorException
from napalm_device_types import APT_UPGRADABLE_COMMAND, parse_apt_upgradable
from napalm_proxmox.ntp import NTP_STATS_COMMAND, parse_ntp_stats
logger = logging.getLogger(__name__)
_JsonDict = dict[str, Any]
@@ -45,6 +48,71 @@ _SUBSCRIPTION_STATUS = {
_PACKAGE_NAME = re.compile(r"^[a-zA-Z0-9][a-zA-Z0-9_.+:~-]*$")
#: What may stand in an snmpd.conf value: no whitespace, no quote, no comment.
_SNMP_WORD = re.compile(r'^[^\s"#]+$')
#: The net-snmp agent's configuration file.
_SNMPD_CONF = "/etc/snmp/snmpd.conf"
#: Exit status 0 when the node is a cluster member. pvecm itself decides
#: membership by /etc/pve/corosync.conf; /etc/corosync/corosync.conf is the
#: local copy every member keeps, there even when /etc/pve is not mounted.
#: Not ``pvecm status``: it also fails on a member whose corosync is down.
_CLUSTER_MEMBER_CHECK = "[ -e /etc/pve/corosync.conf ] || [ -e /etc/corosync/corosync.conf ]"
def _rename_in_hosts(hosts: str, old_short: str, new_short: str, new_domain: str | None) -> str:
"""*hosts* (an /etc/hosts) with the node's names renamed from *old_short*.
Only whole names change: ``old_short`` becomes ``new_short``, and
``old_short.<domain>`` becomes ``new_short.<new_domain>`` (the old domain
kept when *new_domain* is None). Other names that merely start with the
old name (``pve-backup``) and comments stay as they are.
"""
def rename(match: re.Match[str]) -> str:
name = match.group(0)
if name == old_short:
return new_short
if name.startswith(f"{old_short}."):
return f"{new_short}.{new_domain or name[len(old_short) + 1 :]}"
return name
lines = []
for line in hosts.splitlines(keepends=True):
entry, hash_, comment = line.partition("#")
lines.append(re.sub(r"\S+", rename, entry) + hash_ + comment)
return "".join(lines)
#: Proxmox's superuser, whatever the ACL says.
_PVE_SUPERUSER = "root@pam"
#: The role that makes a Proxmox user an administrator of the node.
_PVE_ADMIN_ROLE = "Administrator"
#: Regular login accounts: Debian gives them uids from 1000 up; 65534 is nobody.
_FIRST_REGULAR_UID = 1000
_NOBODY_UID = 65534
def _words(value: Any, sep: str | None = None) -> list[str]:
return [w.strip() for w in str(value or "").split(sep) if w.strip()]
def _user_entry(level: int) -> _JsonDict:
# NAPALM's get_users shape; the password is never read.
return {"level": level, "password": "", "sshkeys": []} # nosec B105
def _login_accounts(passwd: str) -> list[tuple[str, int]]:
"""``(name, uid)`` of root and every regular account in *passwd*."""
accounts = []
for line in passwd.splitlines():
parts = line.split(":")
if len(parts) < 7:
continue
try:
uid = int(parts[2])
except ValueError:
continue
if uid == 0 or _FIRST_REGULAR_UID <= uid < _NOBODY_UID:
accounts.append((parts[0], uid))
return accounts
def _net(value: str) -> str:
@@ -137,36 +205,8 @@ class ProxmoxSystemMixin:
return servers
def get_ntp_stats(self) -> list[_JsonDict]:
"""Return NTP synchronisation statistics from chronyc/ntpq output."""
raw = self._exec_ssh_command(
"chronyc -n tracking 2>/dev/null || ntpq -pn 2>/dev/null || true"
)
stats: list[_JsonDict] = []
for line in raw.splitlines():
line = line.strip()
# ntpq -pn format: *remote refid st t when poll reach delay offset jitter
m = re.match(
r"^([\*\+\-\s])([\d.]+)\s+([\d.]+)\s+(\d+)\s+\S+\s+(\S+)\s+(\d+)\s+(\d+)\s+([\d.]+)\s+([-\d.]+)\s+([\d.]+)",
line,
)
if m:
synced = m.group(1).strip() == "*"
stats.append(
{
"remote": m.group(2),
"referenceid": m.group(3),
"synchronized": synced,
"stratum": int(m.group(4)),
"type": "",
"when": m.group(5),
"hostpoll": int(m.group(6)),
"reachability": int(m.group(7)),
"delay": float(m.group(8)),
"offset": float(m.group(9)),
"jitter": float(m.group(10)),
}
)
return stats
"""Return NTP synchronisation statistics from chrony, or ntpd where chrony is missing."""
return parse_ntp_stats(self._exec_ssh_command(NTP_STATS_COMMAND))
# ------------------------------------------------------------------ #
# get_snmp_information
@@ -213,55 +253,52 @@ class ProxmoxSystemMixin:
def get_users(self) -> dict[str, _JsonDict]:
"""Return users configured on the Proxmox node.
Reads from both the Proxmox access/users API and local /etc/passwd.
Proxmox's own users (``/access/users``) at level 15 when they hold the
Administrator role on the whole datacenter (``/``), directly or through
a group, 1 otherwise; ``root@pam`` is always 15. Then the node's login
accounts from the passwd database: root (15) and regular users
(uid 1000 and up, without ``nobody``) at 0; system accounts are left out.
"""
result: dict[str, _JsonDict] = {}
try:
pve_users = self._api.access.users.get() or [] # type: ignore[union-attr]
except Exception as exc:
logger.debug("Failed to fetch Proxmox users: %s", exc)
pve_users = []
for user in pve_users:
admins, admin_groups = self._pve_administrators()
for user in self._pve_users():
uid = user.get("userid", "")
if not uid:
continue
# Proxmox roles: Administrator -> 15, otherwise 1
level = 1
try:
roles = self._api.access.users(uid).get() or {} # type: ignore[union-attr]
if "Administrator" in str(roles):
level = 15
except Exception as exc:
logger.debug("Failed to fetch roles for user %s: %s", uid, exc)
result[uid] = {
"level": level,
# NAPALM's get_users shape; the password is never read.
"password": "", # nosec B105
"sshkeys": [],
}
groups = set(_words(user.get("groups"), sep=","))
is_admin = uid == _PVE_SUPERUSER or uid in admins or bool(groups & admin_groups)
result[uid] = _user_entry(15 if is_admin else 1)
# Merge local OS users from /etc/passwd
raw = self._exec_ssh_command("getent passwd 2>/dev/null || cat /etc/passwd")
for line in raw.splitlines():
parts = line.split(":")
if len(parts) < 7:
continue
uname, _, uid_str, *_ = parts
try:
uid_int = int(uid_str)
except ValueError:
continue
if uname not in result and uid_int < 1000 or uid_int == 0:
result[uname] = {
"level": 15 if uid_int == 0 else 0,
# NAPALM's get_users shape; the password is never read.
"password": "", # nosec B105
"sshkeys": [],
}
for uname, uid_int in _login_accounts(raw):
result.setdefault(uname, _user_entry(15 if uid_int == 0 else 0))
return result
def _pve_users(self) -> list[_JsonDict]:
try:
return self._api.access.users.get() or [] # type: ignore[union-attr]
except Exception as exc:
logger.debug("Failed to fetch Proxmox users: %s", exc)
return []
def _pve_administrators(self) -> tuple[set[str], set[str]]:
"""Users and groups holding the Administrator role on ``/`` (``/access/acl``)."""
try:
acl = self._api.access.acl.get() or [] # type: ignore[union-attr]
except Exception as exc:
logger.debug("Failed to fetch the Proxmox ACL: %s", exc)
acl = []
users: set[str] = set()
groups: set[str] = set()
for entry in acl:
if entry.get("path") != "/" or entry.get("roleid") != _PVE_ADMIN_ROLE:
continue
if entry.get("type") == "user":
users.add(entry.get("ugid", ""))
elif entry.get("type") == "group":
groups.add(entry.get("ugid", ""))
return users, groups
# ------------------------------------------------------------------ #
# Kernel facts (KernelFactsMixin supplies get_kernel_facts)
# ------------------------------------------------------------------ #
@@ -448,11 +485,17 @@ class ProxmoxSystemMixin:
if not re.match(r"^[a-zA-Z0-9_\-\+\.]+$", pkg):
raise ValueError(f"Invalid package name: {pkg!r}")
pkg_args = " ".join(packages)
output = self._exec_ssh_command(
"DEBIAN_FRONTEND=noninteractive apt-get install --only-upgrade -y "
f"{pkg_args} 2>&1 || true"
output, status = self._rc(
f"DEBIAN_FRONTEND=noninteractive apt-get install --only-upgrade -y {pkg_args}"
)
return {"success": True, "output": output}
if status == 0:
return {"success": True, "output": output}
error = (
f"apt-get exited with status {status}"
if status >= 0
else "apt-get's exit status could not be read"
)
return {"success": False, "output": output, "error": error}
# ------------------------------------------------------------------ #
# SNMP / Health
@@ -527,8 +570,12 @@ class ProxmoxSystemMixin:
return self._step(command, done)
def _write_apt_file(self, path: str, content: str) -> dict:
return self._write_root_file(path, content)
def _write_root_file(self, path: str, content: str) -> dict:
"""Write *content* to *path* as root, mode 644, through a temporary file."""
encoded = base64.b64encode(content.encode()).decode()
# No ``exit`` here: ``_rc`` reads the status after the command's braces.
# No ``exit`` here: it would end the shell before ``_rc`` reads the status.
return self._step(
f'umask 077 && t=$(mktemp) && echo {encoded} | base64 -d > "$t" && '
f'install -m 644 -o root -g root "$t" {shlex.quote(path)}; '
@@ -559,13 +606,7 @@ class ProxmoxSystemMixin:
)
def _write_snmp_agent_config(self, settings: dict) -> dict:
encoded = base64.b64encode(render_snmpd_conf(settings).encode()).decode()
return self._step(
f'umask 077 && t=$(mktemp) && echo {encoded} | base64 -d > "$t" && '
f'install -m 644 -o root -g root "$t" /etc/snmp/snmpd.conf; '
's=$?; rm -f "$t"; exit $s',
"wrote /etc/snmp/snmpd.conf",
)
return self._write_root_file(_SNMPD_CONF, render_snmpd_conf(settings))
def _allow_snmp_sources(self, settings: dict) -> dict:
sources = [_net(s) for s in settings.get("allowed_sources", [])]
@@ -582,6 +623,18 @@ class ProxmoxSystemMixin:
"snmpd enabled and running",
)
def _refuse_clustered_node(self) -> None:
_, status = self._rc(_CLUSTER_MEMBER_CHECK)
if status == 0:
raise CommandErrorException(
"set_hostname refused: the node is part of a cluster; rename a cluster"
" member by Proxmox's own procedure, not in place"
)
if status != 1:
raise CommandErrorException(
"set_hostname refused: could not tell whether the node is part of a cluster"
)
def set_hostname(self, new_hostname: str) -> None:
"""Set the system hostname on the Proxmox node.
@@ -599,16 +652,16 @@ class ProxmoxSystemMixin:
7. Proxmox node TLS certificate — regenerated via
``pvenode cert create --overwrite``.
**Important**: the Proxmox *node name* (shown in the web UI and stored
in ``/etc/pve/nodes/<name>/``) only changes after a full reboot.
Until then the web UI still shows the old node name, but the OS-level
hostname and the SSL certificate already reflect the new value.
The Proxmox *node name* (web UI, ``/etc/pve/nodes/<name>/``) only
changes after a reboot. Accepts a bare hostname or an FQDN.
Accepts a bare hostname (``pve-see``) or an FQDN
(``pve-see.see.local``).
Raises CommandErrorException, before changing anything, on a cluster
member (a corosync.conf exists): its rename is the cluster's procedure.
"""
import base64 as _b64
self._refuse_clustered_node()
if "." in new_hostname:
short_hostname, domain = new_hostname.split(".", 1)
fqdn = new_hostname
@@ -633,17 +686,7 @@ class ProxmoxSystemMixin:
# write back via base64 to avoid shell-escaping issues.
if old_short and old_short != short_hostname:
hosts_raw = self._exec_ssh_command("cat /etc/hosts 2>/dev/null")
# Replace whole-word occurrences of the old short hostname.
# Both "hostname" and "hostname.domain" forms are covered because
# the new value is the full fqdn when a domain was given.
import re as _re
new_replace = fqdn if domain else short_hostname
hosts_new = _re.sub(
r"\b" + _re.escape(old_short) + r"\b",
new_replace,
hosts_raw,
)
hosts_new = _rename_in_hosts(hosts_raw, old_short, short_hostname, domain)
hosts_b64 = _b64.b64encode(hosts_new.encode()).decode()
self._exec_ssh_command(
f"echo {hosts_b64} | base64 -d > /tmp/_netork_hosts"
+255
View File
@@ -0,0 +1,255 @@
"""HypervisorDriver migration methods for Proxmox VE: preflight, migrate, HA maintenance.
A guest is addressed like every other VM method of this driver: by vmid or name,
on the node the driver is connected to (``_resolve_vm``).
- ``get_vm_migration_preflight``: a VM asks ``GET .../qemu/{vmid}/migrate``; a
container has no such endpoint, so the online members are its candidates.
A member the cluster calls offline is refused (``"node offline"``) for both,
also where the VM endpoint would list it as allowed.
- ``migrate_vm``: ``POST .../{qemu|lxc}/{vmid}/migrate`` and the task is followed
to its end; a failure raises with the task log's ``ERROR:`` lines. A guest HA
manages migrates through a ``hamigrate`` task that only hands the request to
the HA manager; the driver then follows ``/cluster/ha/status/current`` until
the guest is on the target, and on a failure reads the migration task's log.
- ``set_node_maintenance``: PVE has no REST endpoint for it; it runs
``ha-manager crm-command node-maintenance`` over the driver's SSH session.
The HA manager moves the node's HA guests away after the command returned.
"""
from __future__ import annotations
import logging
import shlex
from time import monotonic, sleep
from typing import TYPE_CHECKING, Any
from napalm_proxmox.migration import lxc_local, qemu_local, refusal, task_errors, upid_field
from napalm_proxmox.pve_values import to_flag
if TYPE_CHECKING:
# Type-only: newer than the napalm_device_types floor in pyproject.toml.
from napalm_device_types import MigrationPreflightDict
logger = logging.getLogger(__name__)
_JsonDict = dict[str, Any]
#: Enough to reach the ERROR lines at the end of a long storage migration's log.
_TASK_LOG_LINES = 50000
_HA_POLL_SECONDS = 5
#: CRM states while the HA manager is moving a guest.
_HA_MOVING = frozenset({"migrate", "relocate"})
class ProxmoxVMMigrateMixin:
"""Relies on the driver, ``_resolve_vm``/``_guest_api``, ``_members`` and
``_wait_for_task``/``_run_node_command``/``_node_ssh``."""
_api: Any
_node_name: str
hostname: str
timeout: int
_node_api: Any
_resolve_vm: Any
_guest_api: Any
_members: Any
_wait_for_task: Any
_node_ssh: Any
_run_node_command: Any
# -- cluster membership -------------------------------------------------------
def _cluster_members(self) -> list[Any]:
status = self._api.cluster.status.get() or []
clustered = any(s.get("type") == "cluster" for s in status)
return list(self._members(status if clustered else []))
def _other_member(self, node: str) -> None:
names = [m.name for m in self._cluster_members()]
if node == self._node_name or node not in names:
raise ValueError(
f"{node!r} is not another member of the cluster of {self._node_name!r} "
f"(members: {', '.join(names)})"
)
def _storage_sharing(self) -> dict[str, bool]:
rows = self._api.cluster.resources.get(type="storage") or []
sharing: dict[str, bool] = {}
for row in rows:
if row.get("type") == "storage" and row.get("storage"):
name = str(row["storage"])
sharing[name] = sharing.get(name, False) or to_flag(row.get("shared"))
return sharing
# -- preflight ----------------------------------------------------------------
def get_vm_migration_preflight(self, name: str) -> MigrationPreflightDict:
"""Where the guest could move to within its cluster, and what blocks it."""
vmid, vm_type = self._resolve_vm(name)
others = [m for m in self._cluster_members() if m.name != self._node_name]
offline = [m.name for m in others if not m.online]
if vm_type == "vm":
return self._qemu_preflight(vmid, offline)
return self._lxc_preflight(vmid, [m.name for m in others if m.online], offline)
def _qemu_preflight(self, vmid: int, offline: list[str]) -> MigrationPreflightDict:
answer = self._guest_api(vmid, "vm").migrate.get() or {}
refused = {
str(node): refusal(blockers)
for node, blockers in (answer.get("not_allowed_nodes") or {}).items()
}
for node in offline:
refused[node] = "; ".join(filter(None, ["node offline", refused.get(node)]))
disks, resources = qemu_local(answer)
return {
"running": to_flag(answer.get("running")),
"allowed_nodes": [
str(n) for n in answer.get("allowed_nodes") or [] if n not in offline
],
"refused_nodes": refused,
"local_disks": disks,
"local_resources": resources,
}
def _lxc_running(self, vmid: int) -> bool:
status = self._guest_api(vmid, "container").status.current.get() or {}
return bool(status.get("status") == "running")
def _lxc_preflight(
self, vmid: int, online: list[str], offline: list[str]
) -> MigrationPreflightDict:
config = self._guest_api(vmid, "container").config.get() or {}
disks, resources = lxc_local(config, self._storage_sharing())
return {
"running": self._lxc_running(vmid),
"allowed_nodes": online,
"refused_nodes": {n: "node offline" for n in offline},
"local_disks": disks,
"local_resources": resources,
}
# -- migration ----------------------------------------------------------------
def migrate_vm(
self,
name: str,
target_node: str,
online: bool = True,
target_storage: str | None = None,
timeout: int = 3600,
) -> None:
"""Move the guest to *target_node*; block until Proxmox VE finished."""
vmid, vm_type = self._resolve_vm(name)
self._other_member(target_node)
deadline = monotonic() + timeout
params = self._qemu_params(vmid, online) if vm_type == "vm" else self._lxc_params(vmid)
if target_storage:
params["targetstorage" if vm_type == "vm" else "target-storage"] = target_storage
try:
upid = str(self._guest_api(vmid, vm_type).migrate.post(target=target_node, **params))
except Exception as exc:
raise RuntimeError(
f"Migration of {vm_type} {vmid} to {target_node!r} refused: {exc}"
) from exc
self._follow_task(upid, vmid, timeout)
if upid_field(upid, 5) == "hamigrate":
self._follow_ha(vmid, vm_type, target_node, deadline, upid)
def _qemu_params(self, vmid: int, online: bool) -> dict[str, Any]:
answer = self._guest_api(vmid, "vm").migrate.get() or {}
params: dict[str, Any] = {}
if online and to_flag(answer.get("running")):
params["online"] = 1
if qemu_local(answer)[0]:
params["with-local-disks"] = 1
return params
def _lxc_params(self, vmid: int) -> dict[str, Any]:
# A container cannot move while running; restart mode stops and restarts it.
return {"restart": 1} if self._lxc_running(vmid) else {}
def _task_errors(self, upid: str) -> list[str]:
try:
log = self._node_api().tasks(upid).log.get(limit=_TASK_LOG_LINES) or []
except Exception as exc:
logger.debug("Cannot read the log of task %s: %s", upid, exc)
return []
return task_errors(log)
def _follow_task(self, upid: str, vmid: int, timeout: int) -> None:
try:
self._wait_for_task(upid, timeout=timeout)
except RuntimeError as exc:
errors = self._task_errors(upid)
if not errors:
raise
raise RuntimeError(f"Migration of guest {vmid} failed: {'; '.join(errors)}") from exc
# -- a guest HA manages -------------------------------------------------------
def _ha_state(self, sid: str) -> tuple[str | None, str | None]:
rows = self._api.cluster.ha.status.current.get() or []
service: _JsonDict = next((r for r in rows if r.get("sid") == sid), {})
return service.get("node"), service.get("crm_state") or service.get("state")
def _follow_ha(
self, vmid: int, vm_type: str, target: str, deadline: float, ha_upid: str
) -> None:
sid = f"{'vm' if vm_type == 'vm' else 'ct'}:{vmid}"
moving_seen = False
while True:
node, state = self._ha_state(sid)
if node == target and state not in _HA_MOVING:
return
if state == "error":
self._ha_failed(vmid, vm_type, ha_upid, f"HA manager put {sid} into error state")
if state in _HA_MOVING:
moving_seen = True
elif moving_seen:
self._ha_failed(
vmid, vm_type, ha_upid, f"HA manager left {sid} on {node} ({state})"
)
if monotonic() > deadline:
raise RuntimeError(
f"Migration of {sid} to {target!r} timed out on {node} ({state})"
)
sleep(_HA_POLL_SECONDS)
def _ha_failed(self, vmid: int, vm_type: str, ha_upid: str, what: str) -> None:
"""Raise *what* with the ERROR lines of the migration the HA manager ran."""
errors: list[str] = []
kind = "qmigrate" if vm_type == "vm" else "vzmigrate"
try:
tasks = self._node_api().tasks.get(vmid=vmid, typefilter=kind, limit=1) or []
since = int(upid_field(ha_upid, 4) or "0", 16)
if tasks and int(tasks[0].get("starttime") or 0) >= since:
errors = self._task_errors(str(tasks[0].get("upid")))
except Exception as exc:
logger.debug("Cannot find the migration task of guest %s: %s", vmid, exc)
raise RuntimeError(f"{what}: {'; '.join(errors)}" if errors else what)
# -- HA maintenance -----------------------------------------------------------
def set_node_maintenance(self, node: str, enabled: bool) -> None:
"""Switch HA maintenance mode of a cluster node, over SSH (no REST endpoint)."""
names = [m.name for m in self._cluster_members()]
if node not in names:
raise ValueError(
f"{node!r} is not a member of the cluster (members: {', '.join(names)})"
)
action = "enable" if enabled else "disable"
try:
self._node_ssh()
except Exception as exc:
raise RuntimeError(
f"Cannot {action} HA maintenance of node {node!r}: SSH to {self.hostname} "
f"is not available ({exc})"
) from exc
command = f"ha-manager crm-command node-maintenance {action} {shlex.quote(node)}"
try:
self._run_node_command(command, timeout=self.timeout)
except Exception as exc:
raise RuntimeError(
f"ha-manager refused to {action} maintenance of {node!r}: {exc}"
) from exc
+34 -19
View File
@@ -210,12 +210,7 @@ class ProxmoxVMProvisionMixin:
timeout=30,
)
if "EXISTS" not in exists:
_logger.info(f"Downloading cloud image {image_url} -> {local_path}")
self._run_node_command(
f"wget -q -O {local_path}.tmp '{image_url}' "
f"&& mv {local_path}.tmp {local_path}",
timeout=timeout,
)
self._fetch_image(image_url, local_path, timeout)
if not image_checksum:
return local_path
@@ -240,6 +235,20 @@ class ProxmoxVMProvisionMixin:
raise AssertionError("unreachable") # loop always returns or raises above
def _fetch_image(self, image_url: str, local_path: str, timeout: int) -> None:
"""Download *image_url* to *local_path* on the node, through a ``.tmp``.
wget runs with ``-nv`` so its reason ("No space left on device")
reaches the error, and a failed download leaves no partial ``.tmp``.
"""
_logger.info(f"Downloading cloud image {image_url} -> {local_path}")
tmp = f"{local_path}.tmp"
self._run_node_command(
f"wget -nv -O {tmp} '{image_url}' && mv {tmp} {local_path} "
f"|| {{ s=$?; rm -f {tmp}; exit $s; }}",
timeout=timeout,
)
def _cloud_image_on_node(self, image_url: str, image_checksum: str | None, timeout: int) -> str:
"""The node path of the image, downloaded, verified and unpacked.
@@ -745,9 +754,9 @@ class ProxmoxVMProvisionMixin:
cloud_init_args[f"ipconfig{i}"] = "ip=dhcp"
if ssh_public_keys:
# URL-encode SSH keys for Proxmox API
sshkeys = ";".join(ssh_public_keys)
cloud_init_args["sshkeys"] = quote(sshkeys)
# One key per line, URL-encoded without exception: Proxmox's
# "urlencoded" format refuses an unescaped "/" or "+".
cloud_init_args["sshkeys"] = quote("\n".join(ssh_public_keys), safe="")
self._node_api().qemu(vmid).config.post(**cloud_init_args)
@@ -831,16 +840,7 @@ class ProxmoxVMProvisionMixin:
snippets = self._cicustom_snippets(vmid_int)
# Step 2: Delete VM
# Proxmox's API parameter is hyphenated (destroy-unreferenced-disks),
# not a valid Python identifier — proxmoxer forwards kwargs to the
# request verbatim with no underscore-to-hyphen translation, so this
# must be built as a dict and unpacked rather than passed as a kwarg.
_logger.debug(f"Deleting VM {vmid} configuration and disks")
delete_params = {
"purge": 1,
"destroy-unreferenced-disks": 1 if remove_disk else 0,
}
self._node_api().qemu(vmid_int).delete(**delete_params)
self._delete_vm(vmid_int, remove_disk, timeout)
# Step 3: Clean up Cloud-Init snippets
# (This is best-effort; snippet files may be unreachable if storage is unavailable)
@@ -857,6 +857,21 @@ class ProxmoxVMProvisionMixin:
_logger.exception(f"Failed to destroy VM {vmid}: {e}")
raise
def _delete_vm(self, vmid: int, remove_disk: bool, timeout: int) -> None:
"""Delete the VM and wait for it to be gone: the delete is a task."""
_logger.debug(f"Deleting VM {vmid} configuration and disks")
# Proxmox's API parameter is hyphenated (destroy-unreferenced-disks),
# not a valid Python identifier — proxmoxer forwards kwargs to the
# request verbatim with no underscore-to-hyphen translation, so this
# must be built as a dict and unpacked rather than passed as a kwarg.
delete_params = {
"purge": 1,
"destroy-unreferenced-disks": 1 if remove_disk else 0,
}
upid = self._node_api().qemu(vmid).delete(**delete_params)
if upid:
self._wait_for_task(upid, timeout=timeout)
def _cicustom_snippets(self, vmid: int) -> list[tuple[str, str]]:
"""``(storage, path)`` of every snippet the VM's cicustom names.
+3 -5
View File
@@ -16,7 +16,6 @@
"napalm_proxmox/sdn_mixin.py::ProxmoxSDNMixin.set_vlan": 13,
"napalm_proxmox/system_mixin.py::ProxmoxSystemMixin.get_disk_smart": 22,
"napalm_proxmox/system_mixin.py::ProxmoxSystemMixin.get_packages": 16,
"napalm_proxmox/system_mixin.py::ProxmoxSystemMixin.get_users": 15,
"napalm_proxmox/utils.py::addresses_from_node_network": 15,
"napalm_proxmox/vm_contract_mixin.py::parse_vm_config": 16,
"napalm_proxmox/vm_mixin.py::ProxmoxVMMixin._get_vm_disk_and_boot": 16,
@@ -35,13 +34,12 @@
"napalm_proxmox/sdn_mixin.py::ProxmoxSDNMixin.set_vlan": 70,
"napalm_proxmox/system_mixin.py::ProxmoxSystemMixin.get_disk_smart": 74,
"napalm_proxmox/system_mixin.py::ProxmoxSystemMixin.get_packages": 66,
"napalm_proxmox/system_mixin.py::ProxmoxSystemMixin.get_users": 51,
"napalm_proxmox/system_mixin.py::ProxmoxSystemMixin.set_hostname": 124,
"napalm_proxmox/system_mixin.py::ProxmoxSystemMixin.set_hostname": 114,
"napalm_proxmox/vm_mixin.py::ProxmoxVMMixin.get_vm_interfaces": 179,
"napalm_proxmox/vm_mixin.py::ProxmoxVMMixin.get_vms": 114,
"napalm_proxmox/vm_provision_mixin.py::ProxmoxVMProvisionMixin._download_cloud_image": 61,
"napalm_proxmox/vm_provision_mixin.py::ProxmoxVMProvisionMixin._download_cloud_image": 56,
"napalm_proxmox/vm_provision_mixin.py::ProxmoxVMProvisionMixin.create_vm_from_cloud_init": 220,
"napalm_proxmox/vm_provision_mixin.py::ProxmoxVMProvisionMixin.destroy_vm": 65,
"napalm_proxmox/vm_provision_mixin.py::ProxmoxVMProvisionMixin.destroy_vm": 56,
"napalm_proxmox/vm_provision_mixin.py::ProxmoxVMProvisionMixin.get_vm_status": 92
},
"mypy_relaxed": [
+555
View File
@@ -0,0 +1,555 @@
"""get_clusters(): quorum votes, the HA manager's guests, backup jobs and backups.
Each of these is read on top of the cluster report and is best-effort: a read
that fails leaves its key None and the rest of the report stands.
Payloads: ``/cluster/config/nodes`` is corosync's nodelist; ``/cluster/ha/*``
follow pve-ha-manager's API (``src/PVE/API2/HA/{Resources,Status,Rules}.pm``,
master = PVE 9 and stable-4 = PVE 8): a service entry of ``status/current``
carries ``crm_state`` and ``request_state``, a rule carries ``rule``, ``type``
and ``resources`` as a sorted, comma-separated list of sids, and PVE 8 has no
``/cluster/ha/rules``. ``/cluster/backup`` follows pve-manager's
``PVE/API2/Backup.pm``; storage content rows follow
``/nodes/{node}/storage/{storage}/content``.
"""
from __future__ import annotations
from typing import Any
from unittest.mock import MagicMock
from tests.test_clusters import (
CLUSTER_RESOURCES,
GIB,
_api,
_by_name,
_driver,
_storage_row,
)
def _report(api: MagicMock) -> dict[str, Any]:
[cluster] = _driver(api).get_clusters()
return dict(cluster)
class TestVotes:
def test_each_node_has_its_quorum_votes(self):
nodes = _by_name(_report(_api())["nodes"])
assert [nodes[n]["votes"] for n in ("pve1", "pve2", "pve3")] == [1, 1, 2]
def test_unreadable_nodelist_leaves_votes_unknown(self):
api = _api()
api.cluster.config.nodes.get.side_effect = RuntimeError("403 Forbidden")
nodes = _report(api)["nodes"]
assert [n["votes"] for n in nodes] == [None, None, None]
# The rest of the node still stands.
assert nodes[0]["cpu_threads"] == 32
def test_standalone_node_has_no_votes(self):
# A node without corosync.conf answers with an empty nodelist.
api = _api()
api.cluster.config.nodes.get.return_value = []
assert [n["votes"] for n in _report(api)["nodes"]] == [None, None, None]
def test_node_without_quorum_votes_counts_the_default(self):
"""corosync counts a node without quorum_votes as one vote."""
api = _api()
api.cluster.config.nodes.get.return_value = [{"node": "pve1", "name": "pve1"}]
nodes = _by_name(_report(api)["nodes"])
assert nodes["pve1"]["votes"] == 1
assert nodes["pve2"]["votes"] is None
# --------------------------------------------------------------------------- HA
HA_RESOURCES_PVE8 = [
{
"sid": "vm:100",
"type": "vm",
"state": "started",
"group": "prefer-pve1",
"max_restart": 1,
"max_relocate": 1,
"digest": "3f1c",
},
# No state in the config: HA's default, "started".
{"sid": "ct:101", "type": "ct", "digest": "3f1c"},
{"sid": "vm:102", "type": "vm", "state": "disabled", "digest": "3f1c"},
]
HA_STATUS = [
{"id": "quorum", "type": "quorum", "node": "pve1", "quorate": 1, "status": "OK"},
{
"id": "master",
"type": "master",
"node": "pve1",
"status": "pve1 (active, Sun Nov 1 10:00:00 2026)",
"timestamp": 1793527200,
},
{
"id": "lrm:pve1",
"type": "lrm",
"node": "pve1",
"status": "pve1 (active, Sun Nov 1 10:00:00 2026)",
"timestamp": 1793527200,
},
{
"id": "service:vm:100",
"type": "service",
"sid": "vm:100",
"node": "pve2",
"crm_state": "migrate",
"state": "migrate",
"status": "vm:100 (pve2, migrate)",
"request_state": "started",
"group": "prefer-pve1",
"max_restart": 1,
"max_relocate": 1,
},
{
"id": "service:ct:101",
"type": "service",
"sid": "ct:101",
"node": "pve1",
"crm_state": "started",
"state": "started",
"status": "ct:101 (pve1, started)",
"request_state": "started",
},
# Not yet processed by the CRM: no crm_state, a verbose state only.
{
"id": "service:vm:102",
"type": "service",
"sid": "vm:102",
"state": "queued",
"status": "vm:102 (---, queued)",
"request_state": "disabled",
},
]
# PVE 9: groups were migrated into node-affinity rules, so resources lose "group".
HA_RESOURCES_PVE9 = [{k: v for k, v in r.items() if k != "group"} for r in HA_RESOURCES_PVE8]
HA_RULES_PVE9 = [
{
"rule": "ha-group-prefer-pve1",
"type": "node-affinity",
"resources": "vm:100",
"nodes": "pve1:2,pve2:1",
"strict": 0,
"digest": "9ac2",
"order": 1,
},
{
"rule": "keep-apart",
"type": "resource-affinity",
"affinity": "negative",
"resources": "ct:101,vm:100",
"digest": "9ac2",
"order": 2,
},
]
def _ha_api(
resources: Any = HA_RESOURCES_PVE8,
status: Any = HA_STATUS,
rules: Any = None,
) -> MagicMock:
api = _api()
ha = api.cluster.ha
for endpoint, value in ((ha.resources, resources), (ha.status.current, status)):
if isinstance(value, Exception):
endpoint.get.side_effect = value
else:
endpoint.get.return_value = value
if rules is None:
# PVE 8 has no rules endpoint.
ha.rules.get.side_effect = RuntimeError(
"501 Not Implemented: Method 'GET /cluster/ha/rules' not implemented"
)
else:
ha.rules.get.return_value = rules
return api
def _ha(api: MagicMock) -> dict[int, dict[str, Any]] | None:
resources = _report(api)["ha_resources"]
return None if resources is None else {r["vmid"]: r for r in resources}
class TestHAResources:
def test_pve8_resource_with_group(self):
ha = _ha(_ha_api())
assert ha is not None
assert ha[100] == {
"vmid": 100,
"kind": "qemu",
"requested": "started",
"status": "migrate",
"node": "pve2",
"group": "prefer-pve1",
"rules": [],
}
def test_container_and_default_requested_state(self):
ha = _ha(_ha_api())
assert ha is not None
assert ha[101]["kind"] == "lxc"
assert ha[101]["requested"] == "started"
assert ha[101]["status"] == "started"
assert ha[101]["node"] == "pve1"
assert ha[101]["group"] is None
def test_not_yet_processed_reports_the_verbose_state(self):
ha = _ha(_ha_api())
assert ha is not None
assert ha[102]["requested"] == "disabled"
assert ha[102]["status"] == "queued"
assert ha[102]["node"] is None
def test_pve9_rules_name_their_resources(self):
ha = _ha(_ha_api(resources=HA_RESOURCES_PVE9, rules=HA_RULES_PVE9))
assert ha is not None
assert ha[100]["rules"] == ["ha-group-prefer-pve1", "keep-apart"]
assert ha[101]["rules"] == ["keep-apart"]
assert ha[102]["rules"] == []
assert ha[100]["group"] is None
def test_no_ha_configured_is_an_empty_list(self):
assert _report(_ha_api(resources=[], status=[]))["ha_resources"] == []
def test_unreadable_resources_leave_the_list_unknown(self):
api = _ha_api(resources=RuntimeError("403 Forbidden: Permission check failed"))
report = _report(api)
assert report["ha_resources"] is None
assert report["name"] == "lab"
def test_unreadable_status_leaves_status_and_node_unknown(self):
ha = _ha(_ha_api(status=RuntimeError("500 cfs lock")))
assert ha is not None
assert (ha[100]["status"], ha[100]["node"]) == (None, None)
assert ha[100]["requested"] == "started"
# ---------------------------------------------------------------------- backups
BACKUP_JOBS = [
{
"id": "backup-5a1b2c3d-e4f5",
"type": "vzdump",
"enabled": 1,
"schedule": "sat 02:00",
"storage": "pbs",
"mode": "snapshot",
"all": 1,
"exclude": "105,106",
"compress": "zstd",
"mailnotification": "always",
"next-run": 1793584800,
"notes-template": "{{guestname}}",
},
{
"id": "backup-7e8f9a0b-1c2d",
"type": "vzdump",
"enabled": 0,
"schedule": "mon..fri 21:00",
"storage": "local",
"mode": "stop",
"vmid": "100,101",
"node": "pve1",
},
# No "enabled": a job is enabled unless it says otherwise.
{
"id": "backup-0b1c2d3e-4f5a",
"type": "vzdump",
"schedule": "daily",
"storage": "pbs",
"mode": "suspend",
"pool": "prod",
},
]
# PVE >= 8.1: GET /pools?poolid=prod
POOL_INDEX = [
{
"poolid": "prod",
"comment": "",
"members": [
{"id": "qemu/100", "type": "qemu", "vmid": 100, "node": "pve1", "name": "web"},
{"id": "lxc/101", "type": "lxc", "vmid": 101, "node": "pve1", "name": "dns"},
{"id": "storage/pve1/local", "type": "storage", "storage": "local", "node": "pve1"},
],
}
]
def _backup_api(jobs: Any = BACKUP_JOBS) -> MagicMock:
api = _api()
if isinstance(jobs, Exception):
api.cluster.backup.get.side_effect = jobs
else:
api.cluster.backup.get.return_value = jobs
api.pools.get.return_value = POOL_INDEX
return api
def _jobs(api: MagicMock) -> dict[str, dict[str, Any]]:
jobs = _report(api)["backup_jobs"]
assert jobs is not None
return {j["id"]: j for j in jobs}
class TestBackupJobs:
def test_all_guests_job(self):
assert _jobs(_backup_api())["backup-5a1b2c3d-e4f5"] == {
"id": "backup-5a1b2c3d-e4f5",
"enabled": True,
"schedule": "sat 02:00",
"storage": "pbs",
"mode": "snapshot",
"all": True,
"vmids": [],
"exclude": [105, 106],
"pool": None,
"node": None,
}
def test_disabled_job_for_named_guests_on_one_node(self):
job = _jobs(_backup_api())["backup-7e8f9a0b-1c2d"]
assert job["enabled"] is False
assert job["all"] is False
assert job["vmids"] == [100, 101]
assert job["exclude"] == []
assert job["node"] == "pve1"
def test_pool_job_lists_the_pools_guests(self):
api = _backup_api()
job = _jobs(api)["backup-0b1c2d3e-4f5a"]
assert job["enabled"] is True
assert job["pool"] == "prod"
assert job["vmids"] == [100, 101]
api.pools.get.assert_called_with(poolid="prod")
def test_pool_read_falls_back_to_the_old_endpoint(self):
"""Before PVE 8.1, /pools takes no poolid; /pools/{poolid} answers instead."""
api = _backup_api()
api.pools.get.side_effect = RuntimeError("400 Bad Request: Parameter verification failed")
pool = MagicMock()
pool.get.return_value = {"comment": "", "members": POOL_INDEX[0]["members"]}
api.pools.side_effect = lambda name: pool if name == "prod" else MagicMock()
assert _jobs(api)["backup-0b1c2d3e-4f5a"]["vmids"] == [100, 101]
def test_old_style_start_time_and_days(self):
"""Jobs from vzdump.cron carry starttime/dow; PVE turns them into a schedule."""
jobs = [
{"id": "a", "starttime": "02:30", "dow": "sat,sun", "storage": "local", "vmid": "102"},
{
"id": "b",
"starttime": "03:00",
"dow": "mon,tue,wed,thu,fri,sat,sun",
"storage": "local",
"all": 1,
},
{"id": "c", "starttime": "04:00", "storage": "local", "all": 1},
]
found = _jobs(_backup_api(jobs))
assert [found[i]["schedule"] for i in "abc"] == ["sat,sun 02:30", "03:00", "04:00"]
def test_job_with_a_dump_directory_has_no_storage(self):
jobs = [{"id": "d", "schedule": "daily", "dumpdir": "/mnt/backup", "all": 1}]
job = _jobs(_backup_api(jobs))["d"]
assert job["storage"] is None
assert job["mode"] is None
def test_no_jobs_is_an_empty_list(self):
assert _report(_backup_api([]))["backup_jobs"] == []
def test_unreadable_jobs_leave_the_list_unknown(self):
report = _report(_backup_api(RuntimeError("403 Forbidden")))
assert report["backup_jobs"] is None
assert report["nodes"]
def test_unreadable_pool_leaves_the_list_unknown(self):
"""A pool job whose guests cannot be read would look like a job for none."""
api = _backup_api()
api.pools.get.side_effect = RuntimeError("403 Forbidden")
api.pools.return_value.get.side_effect = RuntimeError("403 Forbidden")
api.pools.side_effect = None
assert _report(api)["backup_jobs"] is None
BACKUP_RESOURCES = [
*CLUSTER_RESOURCES,
_storage_row("pve1", "pbs", "pbs", 1, "backup", 8000 * GIB, 2000 * GIB),
_storage_row("pve2", "pbs", "pbs", 1, "backup", 8000 * GIB, 2000 * GIB),
_storage_row("pve1", "local", "dir", 0, "iso,vztmpl,backup", 100 * GIB, 20 * GIB),
_storage_row("pve2", "local", "dir", 0, "iso,vztmpl,backup", 100 * GIB, 30 * GIB),
_storage_row("pve3", "local", "dir", 0, "iso,vztmpl,backup", 0, 0, status="unknown"),
]
PBS_CONTENT = [
{
"volid": "pbs:backup/vm/100/2026-10-31T01:00:00Z",
"content": "backup",
"format": "pbs-vm",
"ctime": 1793408400,
"size": 32 * GIB,
"vmid": 100,
"subtype": "qemu",
"notes": "web",
"protected": 0,
"verification": {"state": "ok", "upid": "UPID:pbs:0000:0000:0000:verify:x:root@pam:"},
},
{
"volid": "pbs:backup/vm/100/2026-10-24T01:00:00Z",
"content": "backup",
"format": "pbs-vm",
"ctime": 1792803600,
"size": 31 * GIB,
"vmid": 100,
"subtype": "qemu",
},
{
"volid": "pbs:backup/ct/101/2026-10-31T01:05:00Z",
"content": "backup",
"format": "pbs-ct",
"ctime": 1793408700,
"size": 2 * GIB,
"vmid": 101,
"subtype": "lxc",
},
]
LOCAL_CONTENT = {
"pve1": [
{
"volid": "local:backup/vzdump-qemu-100-2026_10_30-21_00_02.vma.zst",
"content": "backup",
"format": "vma.zst",
"ctime": 1793394002,
"size": 1234567,
"vmid": 100,
"subtype": "qemu",
},
# A file that does not follow vzdump's naming has no vmid.
{"volid": "local:backup/manual.tar", "content": "backup", "format": "tar", "size": 10},
],
"pve2": [
{
"volid": "local:backup/vzdump-qemu-100-2026_10_29-21_00_01.vma.zst",
"content": "backup",
"format": "vma.zst",
"ctime": 1793307601,
"size": 1200000,
"vmid": 100,
"subtype": "qemu",
},
],
}
def _storage_api(
content: dict[tuple[str, str], Any] | None = None,
) -> tuple[MagicMock, dict[tuple[str, str], MagicMock]]:
"""The cluster with backup storages; *content* per (node, storage)."""
rows = content or {
("pve1", "pbs"): PBS_CONTENT,
("pve2", "pbs"): PBS_CONTENT,
("pve1", "local"): LOCAL_CONTENT["pve1"],
("pve2", "local"): LOCAL_CONTENT["pve2"],
}
api = _api(resources=BACKUP_RESOURCES)
calls: dict[tuple[str, str], MagicMock] = {}
for node in ("pve1", "pve2"):
def storage(name: str, node: str = node) -> MagicMock:
res = calls.setdefault((node, name), MagicMock())
value = rows.get((node, name), [])
if isinstance(value, Exception):
res.content.get.side_effect = value
else:
res.content.get.return_value = value
return res
api.nodes(node).storage.side_effect = storage
return api, calls
def _backups(api: MagicMock) -> dict[tuple[int, str], dict[str, Any]] | None:
backups = _report(api)["backups"]
return None if backups is None else {(b["vmid"], b["storage"]): b for b in backups}
class TestBackups:
def test_newest_backup_per_guest_and_storage(self):
api, _ = _storage_api()
backups = _backups(api)
assert backups is not None
assert backups[(100, "pbs")] == {
"vmid": 100,
"storage": "pbs",
"time": 1793408400,
"size": 32 * GIB,
"count": 2,
}
assert backups[(101, "pbs")]["count"] == 1
def test_shared_storage_read_once_on_the_own_node(self):
api, calls = _storage_api()
_report(api)
calls[("pve1", "pbs")].content.get.assert_called_once_with(content="backup")
assert ("pve2", "pbs") not in calls
def test_local_storage_read_on_each_node_that_has_it(self):
api, calls = _storage_api()
backups = _backups(api)
assert backups is not None
# The same guest backed up to "local" on two nodes: counted together.
assert backups[(100, "local")] == {
"vmid": 100,
"storage": "local",
"time": 1793394002,
"size": 1234567,
"count": 2,
}
assert ("pve1", "local") in calls and ("pve2", "local") in calls
def test_storage_without_backup_content_is_not_read(self):
api, calls = _storage_api()
_report(api)
assert not [key for key in calls if key[1] in ("ceph-vm", "local-lvm", "nfs-iso")]
def test_entry_without_vmid_is_skipped(self):
api, _ = _storage_api()
backups = _backups(api)
assert backups is not None
assert {vmid for vmid, _ in backups} == {100, 101}
def test_shared_storage_tried_on_the_next_node(self):
api, _ = _storage_api(
{
("pve1", "pbs"): RuntimeError("pbs: datastore not reachable from pve1"),
("pve2", "pbs"): PBS_CONTENT,
}
)
backups = _backups(api)
assert backups is not None
assert backups[(100, "pbs")]["count"] == 2
def test_unreadable_storage_leaves_backups_unknown(self):
"""Half a list would say a guest has no backup when it may have one."""
api, _ = _storage_api(
{
("pve1", "pbs"): PBS_CONTENT,
("pve1", "local"): LOCAL_CONTENT["pve1"],
("pve2", "local"): RuntimeError("596 Connection timed out"),
}
)
report = _report(api)
assert report["backups"] is None
assert report["storages"]
def test_no_backup_storage_is_an_empty_list(self):
assert _report(_api())["backups"] == []
+14
View File
@@ -167,6 +167,13 @@ SDN_VNETS = [
{"vnet": "overlay", "zone": "fabric", "tag": 100000, "type": "vnet", "digest": "x"},
]
# /cluster/config/nodes: corosync's nodelist; corosync.conf keeps every value a string.
CONFIG_NODES = [
{"node": "pve1", "name": "pve1", "nodeid": "1", "quorum_votes": "1", "ring0_addr": "10.0.0.1"},
{"node": "pve2", "name": "pve2", "nodeid": "2", "quorum_votes": "1", "ring0_addr": "10.0.0.2"},
{"node": "pve3", "name": "pve3", "nodeid": "3", "quorum_votes": "2", "ring0_addr": "10.0.0.3"},
]
HA_MANAGER_STATUS = {
"manager_status": {
"master_node": "pve1",
@@ -213,6 +220,12 @@ def _api(
manager.side_effect = ha
else:
manager.return_value = HA_MANAGER_STATUS if ha is None else ha
# Read for get_clusters' votes, HA resources and backups; see test_cluster_ha_backups.
api.cluster.config.nodes.get.return_value = CONFIG_NODES
api.cluster.ha.resources.get.return_value = []
api.cluster.ha.status.current.get.return_value = []
api.cluster.ha.rules.get.return_value = []
api.cluster.backup.get.return_value = []
return api
@@ -268,6 +281,7 @@ class TestNodes:
"cpu_usage": 0.25,
"memory_total": 128 * GIB,
"memory_used": 64 * GIB,
"votes": 1,
}
def test_other_member_is_not_local(self):
+34
View File
@@ -10,6 +10,7 @@ contract's ``install_package``. Every step reads its exit status.
from __future__ import annotations
import re
import subprocess
import pytest
@@ -36,6 +37,18 @@ class _Node:
return f"{out}\n__PVE_RC={rc}" if "__PVE_RC" in command else out
def _shell_driver(prelude: str) -> ProxmoxDriver:
"""A driver whose node shell is a real local ``sh``, *prelude* run first."""
def shell(command: str) -> str:
run = subprocess.run(["sh", "-c", prelude + command], capture_output=True, text=True)
return run.stdout
d = ProxmoxDriver("pve", "root@pam", "pw")
d._exec_ssh_command = shell
return d
@pytest.fixture
def node():
return _Node()
@@ -87,6 +100,27 @@ class TestSnmpAgent:
assert result["success"] is False
assert not any("/etc/snmp/snmpd.conf" in c for c in node.ran)
def test_the_written_config_reports_success_through_a_real_shell(self, tmp_path):
"""The configure step's own exit status must reach ``_rc`` (#30).
``_Node`` answers ``__PVE_RC=0`` on its own; a real ``sh`` shows whether the
wrapped command still prints it. ``install`` is a function writing to tmp_path.
"""
target = tmp_path / "snmpd.conf"
d = _shell_driver(f'install() {{ cat "$7" > {target}; }}; ')
step = d._write_snmp_agent_config(SETTINGS)
assert step == {"ok": True, "output": "wrote /etc/snmp/snmpd.conf"}
assert "rocommunity c0mm 10.7.224.20/32" in target.read_text()
def test_a_refused_write_reports_failure_through_a_real_shell(self):
d = _shell_driver('install() { echo "install: read-only file system"; return 1; }; ')
step = d._write_snmp_agent_config(SETTINGS)
assert step == {"ok": False, "output": "install: read-only file system"}
def test_the_management_peer(self, driver):
assert driver.get_management_peer() == "10.7.224.20"
+65
View File
@@ -0,0 +1,65 @@
"""LLDP neighbours read from lldpd's ``lldpcli show neighbors``."""
from __future__ import annotations
from napalm_proxmox import ProxmoxDriver
# lldpd 1.0 output: one "Interface:" block per neighbour, so a port with two
# neighbours (a switch and a phone behind it) appears twice. The phone sends
# no SysName.
TWO_NEIGHBOURS_ON_ONE_PORT = """\
-------------------------------------------------------------------------------
LLDP neighbors:
-------------------------------------------------------------------------------
Interface: eno1, via: LLDP, RID: 1, Time: 0 day, 00:10:12
Chassis:
ChassisID: mac 00:11:22:33:44:55
SysName: sw1
Port:
PortID: ifname gi1/0/1
PortDescr: GigabitEthernet1/0/1
-------------------------------------------------------------------------------
Interface: eno1, via: LLDP, RID: 2, Time: 0 day, 00:01:02
Chassis:
ChassisID: mac aa:bb:cc:dd:ee:ff
Port:
PortID: mac aa:bb:cc:dd:ee:ff
-------------------------------------------------------------------------------
Interface: eno2, via: LLDP, RID: 3, Time: 0 day, 00:10:12
Chassis:
ChassisID: mac 00:11:22:33:44:66
SysName: sw2
Port:
PortID: ifname gi1/0/2
-------------------------------------------------------------------------------
"""
def _driver(output: str) -> ProxmoxDriver:
d = ProxmoxDriver("pve", "root@pam", "pw")
d._exec_ssh_command = lambda command: output
return d
class TestGetLldpNeighbors:
def test_every_neighbour_on_a_port_is_its_own_entry(self):
result = _driver(TWO_NEIGHBOURS_ON_ONE_PORT).get_lldp_neighbors()
assert result == {
"eno1": [
{"hostname": "sw1", "port": "gi1/0/1"},
{"hostname": "", "port": "aa:bb:cc:dd:ee:ff"},
],
"eno2": [{"hostname": "sw2", "port": "gi1/0/2"}],
}
def test_no_lldpd_output_is_no_neighbour(self):
assert _driver("").get_lldp_neighbors() == {}
class TestGetLldpNeighborsDetail:
def test_every_neighbour_on_a_port_is_its_own_entry(self):
result = _driver(TWO_NEIGHBOURS_ON_ONE_PORT).get_lldp_neighbors_detail()
assert [e["remote_system_name"] for e in result["eno1"]] == ["sw1", ""]
assert result["eno1"][1]["remote_chassis_id"] == "aa:bb:cc:dd:ee:ff"
+25
View File
@@ -175,6 +175,31 @@ class TestPing:
assert result["success"]["packet_loss"] == 0
assert result["success"]["rtt_min"] == pytest.approx(10.5, abs=0.01)
def test_every_reply_is_a_result(self, driver):
"""iputils puts "from <ip>" before icmp_seq (napalm-proxmox#12)."""
driver._node_api().execute.post.return_value = {"data": self.PING_SUCCESS}
result = driver.ping("8.8.8.8")
assert result["success"]["results"] == [
{"ip_address": "8.8.8.8", "rtt": 10.5},
{"ip_address": "8.8.8.8", "rtt": 11.2},
]
def test_a_reply_by_name_and_over_ipv6(self, driver):
output = (
"PING dns.google (8.8.4.4) 56(84) bytes of data.\n"
"64 bytes from dns.google (8.8.4.4): icmp_seq=1 ttl=118 time=123 ms\n"
"64 bytes from 2001:db8::1: icmp_seq=2 ttl=64 time=0.045 ms\n"
"--- dns.google ping statistics ---\n"
"2 packets transmitted, 2 received, 0% packet loss, time 1001ms\n"
"rtt min/avg/max/mdev = 0.045/61.5/123.0/61.4 ms\n"
)
driver._node_api().execute.post.return_value = {"data": output}
result = driver.ping("dns.google")
assert result["success"]["results"] == [
{"ip_address": "8.8.4.4", "rtt": 123.0},
{"ip_address": "2001:db8::1", "rtt": 0.045},
]
def test_failure(self, driver):
driver._node_api().execute.post.return_value = {"data": self.PING_FAIL}
result = driver.ping("badhost")
+86
View File
@@ -0,0 +1,86 @@
"""get_ntp_stats on chrony (Proxmox VE's default) and on ntpd (#18)."""
from __future__ import annotations
from unittest.mock import patch
import pytest
# chrony 4.x, ``chronyc -n -c sources`` then ``chronyc -n -c sourcestats``,
# behind the markers the driver's command prints.
CHRONY = """\
CHRONY_SOURCES
^,*,192.168.1.1,2,6,377,8,-0.000012345,-0.000013000,0.012345678
^,+,10.0.0.2,3,10,17,301,0.000250000,0.000251000,0.020000000
^,?,10.0.0.3,0,6,0,-,0.000000000,0.000000000,0.000000000
CHRONY_SOURCESTATS
192.168.1.1,10,6,320,-0.012,0.150,-0.000001234,0.000045678
10.0.0.2,6,4,1200,0.003,0.400,0.000200000,0.000500000
"""
# ntpd's ``ntpq -pn``: a candidate without tally mark keeps its leading space.
NTPQ = """\
remote refid st t when poll reach delay offset jitter
==============================================================================
*192.168.1.1 10.0.0.1 4 u 12 64 377 1.234 0.567 0.123
192.168.1.5 10.0.0.1 3 u 40 64 377 2.000 0.100 0.050
"""
def _stats(driver, output):
with patch.object(driver, "_exec_ssh_command", return_value=output):
return driver.get_ntp_stats()
class TestChrony:
def test_every_source_is_reported(self, driver):
assert [s["remote"] for s in _stats(driver, CHRONY)] == [
"192.168.1.1",
"10.0.0.2",
"10.0.0.3",
]
def test_the_selected_source_is_synchronized(self, driver):
stats = _stats(driver, CHRONY)
assert [s["synchronized"] for s in stats] == [True, False, False]
def test_fields_in_ntpq_units(self, driver):
first = _stats(driver, CHRONY)[0]
assert first == {
"remote": "192.168.1.1",
"referenceid": "",
"synchronized": True,
"stratum": 2,
"type": "u",
"when": "8",
"hostpoll": 64,
"reachability": 377,
"delay": 0.0,
"offset": pytest.approx(-0.013),
"jitter": pytest.approx(0.045678),
}
def test_a_source_without_statistics_has_no_jitter(self, driver):
last = _stats(driver, CHRONY)[2]
assert last["jitter"] == 0.0
assert last["when"] == "-"
assert last["reachability"] == 0
def test_the_command_asks_chrony_first(self, driver):
with patch.object(driver, "_exec_ssh_command", return_value="") as exec_:
driver.get_ntp_stats()
assert "chronyc -n -c sources" in exec_.call_args[0][0]
assert "ntpq -pn" in exec_.call_args[0][0]
class TestNtpd:
def test_a_peer_without_tally_mark_is_kept(self, driver):
stats = _stats(driver, NTPQ)
assert [s["remote"] for s in stats] == ["192.168.1.1", "192.168.1.5"]
assert [s["synchronized"] for s in stats] == [True, False]
assert stats[1]["stratum"] == 3
+107
View File
@@ -0,0 +1,107 @@
"""set_hostname: what it writes to /etc/hosts, and that it leaves a clustered node alone
(napalm-proxmox#21)."""
from __future__ import annotations
import base64
import re
import pytest
from napalm.base.exceptions import CommandErrorException
from napalm_proxmox import ProxmoxDriver
HOSTS = """\
127.0.0.1 localhost.localdomain localhost
# pve is the node; pve-backup its backup target
10.0.0.1 pve.old.lan pve
10.0.0.2 pve-backup.old.lan pve-backup
"""
class _Node:
"""The node's shell: /etc/hostname reads "pve", /etc/hosts reads HOSTS.
*cluster_check* is what the test for a corosync.conf prints: exit status 1
(no corosync.conf, a single node) unless a test says otherwise.
"""
def __init__(self) -> None:
self.ran: list[str] = []
self.cluster_check = "__PVE_RC=1"
def __call__(self, command: str) -> str:
self.ran.append(command)
if "corosync.conf" in command:
return self.cluster_check
if command.startswith("cat /etc/hostname"):
return "pve\n"
if command.startswith("cat /etc/hosts"):
return HOSTS
return ""
def written_hosts(self) -> str:
write = next(c for c in self.ran if c.rstrip().endswith("chmod 644 /etc/hosts"))
return base64.b64decode(re.match(r"echo (\S+) \| base64 -d", write).group(1)).decode()
@pytest.fixture
def node() -> _Node:
return _Node()
@pytest.fixture
def driver(node: _Node) -> ProxmoxDriver:
d = ProxmoxDriver("pve", "root@pam", "pw")
d._exec_ssh_command = node
return d
def test_an_fqdn_renames_the_nodes_names_and_nothing_else(driver, node):
driver.set_hostname("new.dom.lan")
assert node.written_hosts() == (
"127.0.0.1 localhost.localdomain localhost\n"
"# pve is the node; pve-backup its backup target\n"
"10.0.0.1 new.dom.lan new\n"
"10.0.0.2 pve-backup.old.lan pve-backup\n"
)
def test_a_short_name_keeps_the_old_domain(driver, node):
driver.set_hostname("new")
assert "10.0.0.1 new.old.lan new\n" in node.written_hosts()
assert "10.0.0.2 pve-backup.old.lan pve-backup\n" in node.written_hosts()
def test_the_same_name_leaves_hosts_alone(driver, node):
driver.set_hostname("pve")
assert not any("/etc/hosts" in c and "base64 -d" in c for c in node.ran)
def test_a_clustered_node_is_refused_before_anything_changes(driver, node):
node.cluster_check = "__PVE_RC=0"
with pytest.raises(CommandErrorException, match="part of a cluster"):
driver.set_hostname("new")
assert [c for c in node.ran if "corosync.conf" not in c] == []
def test_the_cluster_check_looks_at_both_corosync_confs(driver, node):
driver.set_hostname("new")
check = next(c for c in node.ran if "corosync.conf" in c)
assert "/etc/pve/corosync.conf" in check
assert "/etc/corosync/corosync.conf" in check
def test_an_unreadable_cluster_check_refuses_too(driver, node):
node.cluster_check = ""
with pytest.raises(CommandErrorException, match="could not tell"):
driver.set_hostname("new")
assert [c for c in node.ran if "corosync.conf" not in c] == []
+79
View File
@@ -0,0 +1,79 @@
"""The node's SSH login takes any key type, and a key wins over the API password (#17)."""
from __future__ import annotations
import io
import paramiko
import pytest
from cryptography.hazmat.primitives import serialization
from cryptography.hazmat.primitives.asymmetric import ed25519
from napalm_proxmox.driver import ProxmoxDriver
from tests.test_ssh_host_key import client # noqa: F401 - the fake paramiko client fixture
def _ed25519_pem() -> str:
key = ed25519.Ed25519PrivateKey.generate()
return key.private_bytes(
serialization.Encoding.PEM,
serialization.PrivateFormat.OpenSSH,
serialization.NoEncryption(),
).decode()
def _paramiko_pem(key: paramiko.PKey) -> str:
out = io.StringIO()
key.write_private_key(out)
return out.getvalue()
def _driver(key: str, ssh_password: str | None = None) -> ProxmoxDriver:
args = {"ssh_private_key_str": key}
if ssh_password is not None:
args["ssh_password"] = ssh_password
return ProxmoxDriver("10.0.0.9", "root", "api-s3cret", optional_args=args)
@pytest.mark.parametrize(
("pem", "kind"),
[
(_ed25519_pem(), paramiko.Ed25519Key),
(_paramiko_pem(paramiko.ECDSAKey.generate()), paramiko.ECDSAKey),
(_paramiko_pem(paramiko.RSAKey.generate(1024)), paramiko.RSAKey),
],
)
def test_any_key_type_logs_in_although_an_api_password_is_set(client, pem, kind): # noqa: F811
_driver(pem)._node_ssh()
assert isinstance(client.connected_with["pkey"], kind)
def test_the_password_stays_as_the_fallback(client): # noqa: F811
_driver(_ed25519_pem(), ssh_password="ssh-pw")._node_ssh()
assert client.connected_with["password"] == "ssh-pw"
def test_without_key_the_password_logs_in(client): # noqa: F811
ProxmoxDriver("10.0.0.9", "root", "api-s3cret")._node_ssh()
assert client.connected_with["password"] == "api-s3cret"
assert "pkey" not in client.connected_with
def test_a_key_that_cannot_be_read_leaves_the_password(client, caplog): # noqa: F811
_driver("not a key")._node_ssh()
assert "pkey" not in client.connected_with
assert client.connected_with["password"] == "api-s3cret"
assert "not a key" not in caplog.text
assert "could not be read" in caplog.text
def test_a_token_only_login_sends_no_empty_password(client): # noqa: F811
ProxmoxDriver(
"10.0.0.9", "root", "", optional_args={"ssh_private_key_str": _ed25519_pem()}
)._node_ssh()
assert "password" not in client.connected_with
+58
View File
@@ -0,0 +1,58 @@
"""The physical port set_vlan configures: _find_switch_uplink / _is_physical_uplink."""
from __future__ import annotations
from unittest.mock import MagicMock
from napalm_proxmox.sdn_mixin import ProxmoxSDNMixin
NETWORK = [
{"iface": "eno1", "type": "eth"},
{"iface": "vmbr0", "type": "bridge", "bridge_ports": "eno1"},
]
def _mixin(detail: dict, basic: dict) -> ProxmoxSDNMixin:
m = ProxmoxSDNMixin()
m._get_node_network = MagicMock(return_value=NETWORK)
m.get_lldp_neighbors_detail = MagicMock(return_value=detail)
m.get_lldp_neighbors = MagicMock(return_value=basic)
return m
class TestFindSwitchUplink:
def test_physical_port_with_a_bridge_neighbour_is_the_uplink(self):
m = _mixin(
detail={
"tap100i0": [{"remote_system_capab": ["bridge"]}],
"eno1": [{"remote_system_capab": ["bridge", "router"]}],
},
basic={},
)
assert m._find_switch_uplink() == "eno1"
def test_basic_lldp_fallback_skips_a_virtual_port(self):
m = _mixin(
detail={},
basic={"fwpr100p0": [{"hostname": "sw"}], "eno1": [{"hostname": "sw"}]},
)
assert m._find_switch_uplink() == "eno1"
def test_no_physical_neighbour_is_none(self):
m = _mixin(detail={"tap100i0": [{"remote_system_capab": ["bridge"]}]}, basic={})
assert m._find_switch_uplink() is None
class TestIsPhysicalUplink:
def test_eth_in_the_network_config_is_physical(self):
assert ProxmoxSDNMixin()._is_physical_uplink("eno1", {"eno1": {"type": "eth"}})
def test_virtual_prefix_is_never_physical(self):
network = {"tap100i0": {"type": ""}}
assert not ProxmoxSDNMixin()._is_physical_uplink("tap100i0", network)
def test_port_missing_from_the_network_config_is_not_physical(self):
assert not ProxmoxSDNMixin()._is_physical_uplink("eno9", {})
def test_bridge_is_not_physical(self):
assert not ProxmoxSDNMixin()._is_physical_uplink("vmbr0", {"vmbr0": {"type": "bridge"}})
+29
View File
@@ -104,6 +104,35 @@ class TestRefresh:
assert result == {"success": False, "output": "403 Permission check failed"}
class TestApplyUpdates:
def test_a_clean_upgrade_succeeds(self, driver):
with patch.object(driver, "_exec_ssh_command", return_value="Done.\n__PVE_RC=0") as exec_:
result = driver.apply_updates(["curl"])
assert "apt-get install --only-upgrade -y curl" in exec_.call_args[0][0]
assert result == {"success": True, "output": "Done."}
def test_a_failed_upgrade_is_a_failure(self, driver):
out = "E: Could not get lock /var/lib/dpkg/lock-frontend\n__PVE_RC=100"
with patch.object(driver, "_exec_ssh_command", return_value=out):
result = driver.apply_updates(["curl"])
assert result["success"] is False
assert result["output"] == "E: Could not get lock /var/lib/dpkg/lock-frontend"
assert result["error"] == "apt-get exited with status 100"
def test_an_unread_exit_status_is_no_success(self, driver):
with patch.object(driver, "_exec_ssh_command", return_value=""):
result = driver.apply_updates(["curl"])
assert result["success"] is False
assert result["error"] == "apt-get's exit status could not be read"
def test_a_name_that_is_no_package_name_is_refused(self, driver):
with pytest.raises(ValueError):
driver.apply_updates(["curl;reboot"])
class TestHostStatus:
def test_the_node_is_read_over_the_exec_path(self, driver):
report = (
+66
View File
@@ -0,0 +1,66 @@
"""get_users: Proxmox users with their level, and the node's login accounts (#19)."""
from __future__ import annotations
from unittest.mock import patch
PASSWD = """\
root:x:0:0:root:/root:/bin/bash
daemon:x:1:1:daemon:/usr/sbin:/usr/sbin/nologin
bin:x:2:2:bin:/bin:/usr/sbin/nologin
www-data:x:33:33:www-data:/var/www:/usr/sbin/nologin
alice:x:1000:1000:Alice:/home/alice:/bin/bash
nobody:x:65534:65534:nobody:/nonexistent:/usr/sbin/nologin
"""
# GET /access/acl: who holds which role where.
ACL = [
{"path": "/", "roleid": "Administrator", "type": "user", "ugid": "admin@pve", "propagate": 1},
{"path": "/", "roleid": "Administrator", "type": "group", "ugid": "ops", "propagate": 1},
{"path": "/vms/100", "roleid": "Administrator", "type": "user", "ugid": "vm@pve"},
{"path": "/", "roleid": "PVEAuditor", "type": "user", "ugid": "audit@pve", "propagate": 1},
]
USERS = [
{"userid": "root@pam", "groups": ""},
{"userid": "admin@pve", "groups": ""},
{"userid": "bob@pve", "groups": "ops,dev"},
{"userid": "vm@pve", "groups": ""},
{"userid": "audit@pve", "groups": ""},
]
def _users(driver, passwd=PASSWD):
driver._api.access.users.get.return_value = USERS
driver._api.access.acl.get.return_value = ACL
with patch.object(driver, "_exec_ssh_command", return_value=passwd):
return driver.get_users()
class TestLocalAccounts:
def test_root_and_real_users_not_system_accounts(self, driver):
result = _users(driver)
assert result["root"]["level"] == 15
assert "alice" in result
for system in ("daemon", "bin", "www-data", "nobody"):
assert system not in result
class TestProxmoxUsers:
def test_administrator_on_the_whole_datacenter_is_level_15(self, driver):
result = _users(driver)
assert result["admin@pve"]["level"] == 15
def test_administrator_through_a_group_is_level_15(self, driver):
assert _users(driver)["bob@pve"]["level"] == 15
def test_root_at_pam_is_always_level_15(self, driver):
assert _users(driver)["root@pam"]["level"] == 15
def test_another_role_or_a_narrower_path_is_level_1(self, driver):
result = _users(driver)
assert result["audit@pve"]["level"] == 1
assert result["vm@pve"]["level"] == 1
+382
View File
@@ -0,0 +1,382 @@
"""Moving a guest to another node: preflight, migration, HA maintenance mode.
Payloads: ``GET /nodes/{node}/qemu/{vmid}/migrate`` follows qemu-server's
``migrate_vm_precondition`` (PVE 9: ``not_allowed_nodes`` may name
``unavailable_storages``, ``unavailable-resources`` and
``blocking-ha-resources``); the task log follows ``/nodes/{node}/tasks/{upid}/log``
(``n``/``t`` rows). A guest that HA manages migrates through a ``hamigrate``
task that only hands the request to the HA manager.
"""
from __future__ import annotations
from typing import Any
from unittest.mock import MagicMock, patch
import pytest
from tests.test_clusters import CLUSTER_RESOURCES, CLUSTER_STATUS, GIB, _api, _driver, _storage_row
QEMU_LIST = [
{"vmid": 100, "name": "web", "status": "running", "maxmem": 4 * GIB},
{"vmid": 102, "name": "idle", "status": "stopped", "maxmem": 2 * GIB},
]
LXC_LIST = [{"vmid": "101", "name": "dns", "status": "running", "type": "lxc"}]
PREFLIGHT_RUNNING = {
"running": 1,
"allowed_nodes": ["pve2", "pve3"],
"not_allowed_nodes": {},
"local_disks": [
{
"volid": "local-lvm:vm-100-disk-0",
"size": 32 * GIB,
"drivename": "scsi0",
"is_attached": 1,
"referenced_in_config": 1,
},
{"volid": "local:iso/debian-13.iso", "cdrom": 1, "size": 700 * 1024**2},
# Regenerated on the target: neither a disk to copy nor a blocker.
{"volid": "local-lvm:vm-100-cloudinit", "cdrom": 1, "size": 4 * 1024**2},
],
"local_resources": [],
"mapped-resources": [],
"mapped-resource-info": {},
"has-dbus-vmstate": 1,
}
PREFLIGHT_BLOCKED = {
"running": 0,
"allowed_nodes": [],
"not_allowed_nodes": {
"pve2": {"unavailable_storages": ["local-zfs"]},
"pve3": {
"blocking-ha-resources": [{"sid": "vm:101", "cause": "resource-affinity"}],
"unavailable-resources": ["gpu-a"],
},
},
"local_disks": [],
"local_resources": ["hostpci0", "usb0"],
"mapped-resources": ["gpu-a"],
"mapped-resource-info": {"gpu-a": {"live-migration": 0}},
"has-dbus-vmstate": 1,
}
LXC_CONFIG = {
"hostname": "dns",
"rootfs": "local-lvm:vm-101-disk-0,size=8G",
"mp0": "ceph-vm:vm-101-disk-1,mp=/data,size=32G",
"mp1": "/srv/share,mp=/share",
"mp2": "/mnt/nfs,mp=/nfs,shared=1",
"unused0": "local-lvm:vm-101-disk-2",
"net0": "name=eth0,bridge=vmbr0,ip=dhcp",
"digest": "abc",
}
UPID_QEMU = "UPID:pve1:000ABCDE:0123ABCD:672AB000:qmigrate:100:root@pam:"
UPID_LXC = "UPID:pve1:000ABCDF:0123ABCE:672AB001:vzmigrate:101:root@pam:"
UPID_HA = "UPID:pve1:000ABCE0:0123ABCF:672AB002:hamigrate:100:root@pam:"
FAILED_LOG = [
{"n": 1, "t": "2026-11-01 10:00:00 starting migration of VM 100 to node 'pve2' (10.0.0.2)"},
{"n": 2, "t": "2026-11-01 10:00:01 starting VM 100 on remote node 'pve2'"},
{
"n": 3,
"t": "2026-11-01 10:00:02 ERROR: online migrate failure - remote command failed "
"with exit code 255",
},
{"n": 4, "t": "2026-11-01 10:00:02 aborting phase 2 - cleanup resources"},
{
"n": 5,
"t": "2026-11-01 10:00:03 ERROR: migration finished with problems (duration 00:00:03)",
},
{"n": 6, "t": "TASK ERROR: migration problems"},
]
def _migration_api(
preflight: dict[str, Any] | None = None,
task: dict[str, Any] | None = None,
log: list[dict[str, Any]] | None = None,
status: list[dict[str, Any]] | None = None,
) -> tuple[MagicMock, MagicMock]:
resources = [
*CLUSTER_RESOURCES,
_storage_row("pve1", "local", "dir", 0, "iso,vztmpl,backup", 100 * GIB, 20 * GIB),
]
api = _api(cluster_status=status, resources=resources)
node = api.nodes("pve1")
node.qemu.get.return_value = QEMU_LIST
node.lxc.get.return_value = LXC_LIST
node.qemu.return_value.migrate.get.return_value = preflight or PREFLIGHT_RUNNING
node.qemu.return_value.migrate.post.return_value = UPID_QEMU
node.lxc.return_value.migrate.post.return_value = UPID_LXC
node.lxc.return_value.status.current.get.return_value = {"status": "running", "vmid": 101}
node.lxc.return_value.config.get.return_value = LXC_CONFIG
node.tasks.return_value.status.get.return_value = task or {
"status": "stopped",
"exitstatus": "OK",
}
node.tasks.return_value.log.get.return_value = log or []
return api, node
class TestPreflightQemu:
def test_allowed_nodes_and_local_disks(self):
api, node = _migration_api()
result = _driver(api).get_vm_migration_preflight("web")
assert result == {
"running": True,
# pve3 is offline: the platform would allow it, it cannot take the VM now.
"allowed_nodes": ["pve2"],
"refused_nodes": {"pve3": "node offline"},
"local_disks": ["local-lvm:vm-100-disk-0"],
# A local ISO in a CD drive stops the migration.
"local_resources": ["cdrom=local:iso/debian-13.iso"],
}
node.qemu.assert_any_call(100)
def test_refusal_reasons_in_plain_words(self):
all_online = [dict(s, online=1) if s["type"] == "node" else s for s in CLUSTER_STATUS]
api, _ = _migration_api(preflight=PREFLIGHT_BLOCKED, status=all_online)
result = _driver(api).get_vm_migration_preflight("100")
assert result["running"] is False
assert result["allowed_nodes"] == []
assert result["refused_nodes"] == {
"pve2": "storage local-zfs is not available there",
"pve3": "HA resource vm:101 blocks it (resource-affinity rule); "
"mapped resource gpu-a is not available there",
}
assert result["local_resources"] == ["hostpci0", "usb0"]
def test_unknown_reason_is_passed_on(self):
preflight = {
"running": 0,
"allowed_nodes": [],
"not_allowed_nodes": {"pve2": {"some-future-check": ["x", "y"]}},
"local_disks": [],
"local_resources": [],
}
api, _ = _migration_api(preflight=preflight)
refused = _driver(api).get_vm_migration_preflight("web")["refused_nodes"]
assert refused["pve2"] == "some-future-check: x, y"
def test_unknown_guest(self):
api, _ = _migration_api()
with pytest.raises(ValueError, match="nope"):
_driver(api).get_vm_migration_preflight("nope")
class TestPreflightLxc:
def test_members_and_local_volumes(self):
api, node = _migration_api()
result = _driver(api).get_vm_migration_preflight("dns")
assert result == {
"running": True,
"allowed_nodes": ["pve2"],
"refused_nodes": {"pve3": "node offline"},
# ceph-vm is shared; local-lvm is not.
"local_disks": ["local-lvm:vm-101-disk-0", "local-lvm:vm-101-disk-2"],
# A bind mount that is not marked shared stops the migration.
"local_resources": ["mp1"],
}
node.qemu.return_value.migrate.get.assert_not_called()
def test_stopped_container(self):
api, node = _migration_api()
node.lxc.return_value.status.current.get.return_value = {"status": "stopped"}
assert _driver(api).get_vm_migration_preflight("101")["running"] is False
class TestMigrateQemu:
def test_running_vm_moves_online_with_its_local_disks(self):
api, node = _migration_api()
_driver(api).migrate_vm("web", "pve2")
node.qemu.return_value.migrate.post.assert_called_once_with(
target="pve2", online=1, **{"with-local-disks": 1}
)
node.tasks.assert_called_with(UPID_QEMU)
def test_target_storage_and_offline(self):
preflight = dict(PREFLIGHT_RUNNING, local_disks=[])
api, node = _migration_api(preflight=preflight)
_driver(api).migrate_vm("web", "pve2", online=False, target_storage="ceph-vm")
node.qemu.return_value.migrate.post.assert_called_once_with(
target="pve2", targetstorage="ceph-vm"
)
def test_stopped_vm_is_not_moved_online(self):
preflight = dict(PREFLIGHT_RUNNING, running=0, local_disks=[])
api, node = _migration_api(preflight=preflight)
_driver(api).migrate_vm("idle", "pve2")
node.qemu.return_value.migrate.post.assert_called_once_with(target="pve2")
def test_failure_raises_with_the_task_logs_errors(self):
api, node = _migration_api(
task={"status": "stopped", "exitstatus": "migration problems"}, log=FAILED_LOG
)
with pytest.raises(RuntimeError) as exc:
_driver(api).migrate_vm("web", "pve2")
message = str(exc.value)
assert "online migrate failure - remote command failed with exit code 255" in message
assert "migration finished with problems (duration 00:00:03)" in message
assert "starting VM 100" not in message
node.tasks.return_value.log.get.assert_called_once()
def test_failure_without_error_lines_keeps_the_task_status(self):
api, _ = _migration_api(
task={"status": "stopped", "exitstatus": "migration aborted"},
log=[{"n": 1, "t": "starting migration"}],
)
with pytest.raises(RuntimeError, match="migration aborted"):
_driver(api).migrate_vm("web", "pve2")
def test_refused_request_raises(self):
api, node = _migration_api()
node.qemu.return_value.migrate.post.side_effect = RuntimeError(
"500 Internal Server Error: VM is locked (backup)"
)
with pytest.raises(RuntimeError, match="VM is locked"):
_driver(api).migrate_vm("web", "pve2")
class TestMigrateLxc:
def test_running_container_moves_in_restart_mode(self):
api, node = _migration_api()
_driver(api).migrate_vm("dns", "pve2", target_storage="local-lvm")
node.lxc.return_value.migrate.post.assert_called_once_with(
target="pve2", restart=1, **{"target-storage": "local-lvm"}
)
node.tasks.assert_called_with(UPID_LXC)
def test_stopped_container(self):
api, node = _migration_api()
node.lxc.return_value.status.current.get.return_value = {"status": "stopped"}
_driver(api).migrate_vm("dns", "pve2")
node.lxc.return_value.migrate.post.assert_called_once_with(target="pve2")
class TestTargetRefusedBeforeTheApi:
@pytest.mark.parametrize("target", ["pve1", "pve9", "pve2; reboot"])
def test_not_another_member(self, target):
api, node = _migration_api()
with pytest.raises(ValueError, match="not another member"):
_driver(api).migrate_vm("web", target)
node.qemu.return_value.migrate.post.assert_not_called()
def _ha_status(node: str, crm_state: str) -> list[dict[str, Any]]:
return [
{"id": "quorum", "type": "quorum", "node": "pve1", "quorate": 1, "status": "OK"},
{
"id": "service:vm:100",
"type": "service",
"sid": "vm:100",
"node": node,
"crm_state": crm_state,
"state": crm_state,
"status": f"vm:100 ({node}, {crm_state})",
"request_state": "started",
},
]
class TestMigrateHaManaged:
"""PVE hands an HA guest's migration to the HA manager: the hamigrate task only
queues the request, so the driver follows the HA manager until the guest is there."""
def _api(self, states: list[list[dict[str, Any]]]) -> tuple[MagicMock, MagicMock]:
api, node = _migration_api()
node.qemu.return_value.migrate.post.return_value = UPID_HA
api.cluster.ha.status.current.get.side_effect = states
return api, node
def test_waits_until_the_guest_runs_on_the_target(self):
api, _ = self._api(
[
_ha_status("pve1", "started"),
_ha_status("pve1", "migrate"),
_ha_status("pve2", "started"),
]
)
with patch("napalm_proxmox.vm_migrate_mixin.sleep") as sleep:
_driver(api).migrate_vm("web", "pve2")
assert api.cluster.ha.status.current.get.call_count == 3
assert sleep.call_count == 2
def test_back_on_the_source_node_is_a_failure_with_the_migration_log(self):
api, node = self._api([_ha_status("pve1", "migrate"), _ha_status("pve1", "started")])
node.tasks.get.return_value = [
{
"upid": UPID_QEMU,
"type": "qmigrate",
"id": "100",
"starttime": 0x672AB002 + 12,
"status": "migration problems",
}
]
node.tasks.return_value.log.get.return_value = FAILED_LOG
with patch("napalm_proxmox.vm_migrate_mixin.sleep"):
with pytest.raises(RuntimeError, match="online migrate failure"):
_driver(api).migrate_vm("web", "pve2")
node.tasks.get.assert_called_once_with(vmid=100, typefilter="qmigrate", limit=1)
def test_error_state_raises(self):
api, node = self._api([_ha_status("pve1", "error")])
node.tasks.get.return_value = []
with patch("napalm_proxmox.vm_migrate_mixin.sleep"):
with pytest.raises(RuntimeError, match="error"):
_driver(api).migrate_vm("web", "pve2")
def test_times_out(self):
api, _ = self._api([_ha_status("pve1", "migrate")] * 50)
clock = iter(range(0, 10000, 30))
with (
patch("napalm_proxmox.vm_migrate_mixin.sleep"),
patch("napalm_proxmox.vm_migrate_mixin.monotonic", lambda: next(clock)),
):
with pytest.raises(RuntimeError, match="timed out"):
_driver(api).migrate_vm("web", "pve2", timeout=120)
class TestNodeMaintenance:
def _driver(self) -> tuple[Any, MagicMock]:
api, _ = _migration_api()
drv = _driver(api)
drv._node_ssh = MagicMock() # type: ignore[method-assign]
drv._run_node_command = MagicMock(return_value="") # type: ignore[method-assign]
return drv, drv._run_node_command
def test_enable(self):
drv, run = self._driver()
drv.set_node_maintenance("pve2", True)
run.assert_called_once_with(
"ha-manager crm-command node-maintenance enable pve2", timeout=drv.timeout
)
def test_disable(self):
drv, run = self._driver()
drv.set_node_maintenance("pve3", False)
run.assert_called_once_with(
"ha-manager crm-command node-maintenance disable pve3", timeout=drv.timeout
)
@pytest.mark.parametrize("node", ["pve9", "pve2; reboot", "$(reboot)"])
def test_not_a_member_runs_nothing(self, node):
drv, run = self._driver()
with pytest.raises(ValueError, match="not a member"):
drv.set_node_maintenance(node, True)
run.assert_not_called()
def test_no_ssh(self):
drv, run = self._driver()
drv._node_ssh.side_effect = OSError("Authentication failed")
with pytest.raises(RuntimeError, match="SSH"):
drv.set_node_maintenance("pve2", True)
run.assert_not_called()
def test_command_refused(self):
drv, run = self._driver()
run.side_effect = RuntimeError("Command failed (exit 255): ...\nno such node 'pve2'")
with pytest.raises(RuntimeError, match="no such node"):
drv.set_node_maintenance("pve2", True)
+133
View File
@@ -488,6 +488,42 @@ def test_destroy_vm_already_stopped():
assert mock_vm.delete.called
def _destroy_mixin(delete_task: dict) -> tuple[ProxmoxVMProvisionMixin, MagicMock]:
mixin = ProxmoxVMProvisionMixin()
node = MagicMock()
mixin._node_api = MagicMock(return_value=node)
vm = node.qemu.return_value
vm.status.stop.post.return_value = "UPID:pve1:125:stop"
vm.config.get.return_value = {}
vm.delete.return_value = "UPID:pve1:126:qmdestroy"
tasks = {
"UPID:pve1:125:stop": {"status": "stopped", "exitstatus": "OK"},
"UPID:pve1:126:qmdestroy": delete_task,
}
node.tasks.side_effect = lambda upid: MagicMock(**{"status.get.return_value": tasks[upid]})
return mixin, node
def test_destroy_vm_waits_for_the_delete_task():
"""destroy_vm returns once the VM is gone, not when its delete was queued (#15)."""
mixin, node = _destroy_mixin({"status": "stopped", "exitstatus": "OK"})
with patch("time.sleep"):
mixin.destroy_vm("101")
assert [c.args[0] for c in node.tasks.call_args_list] == [
"UPID:pve1:125:stop",
"UPID:pve1:126:qmdestroy",
]
def test_destroy_vm_raises_when_the_delete_task_fails():
mixin, _ = _destroy_mixin({"status": "stopped", "exitstatus": "VM is locked (backup)"})
with patch("time.sleep"), pytest.raises(RuntimeError, match="VM is locked"):
mixin.destroy_vm("101")
# ---------------------------------------------------------------------------
# get_network_targets
# ---------------------------------------------------------------------------
@@ -1284,3 +1320,100 @@ def test_create_vm_cloudinit_drive_uses_image_storage_not_snippet_storage():
cloud_init_call = next(c for c in mock_vm.config.post.call_args_list if "ide2" in c[1])
assert cloud_init_call[1]["ide2"] == "local-zfs:cloudinit"
assert cloud_init_call[1]["cicustom"].startswith("user=local:snippets/")
def test_create_vm_passes_every_ssh_key_one_per_line():
"""Proxmox's sshkeys is one key per line, URL-encoded (napalm-proxmox#23).
Joined with ";", only the first key worked. Base64 keys carry "/" and "+",
which Proxmox's urlencoded format refuses unescaped.
"""
import re
from urllib.parse import unquote
mixin = ProxmoxVMProvisionMixin()
mixin._node_name = "pve1"
mock_api = MagicMock()
mock_api.cluster.nextid.get.return_value = 110
mock_api.storage.return_value.get.return_value = {"path": "/var/lib/vz"}
mock_node = MagicMock()
mock_node.storage.get.return_value = [
{"storage": "local", "type": "dir", "content": "images,snippets", "enabled": 1},
]
mixin._api = mock_api
mixin._node_api = MagicMock(return_value=mock_node)
mixin._download_cloud_image = MagicMock(return_value="/var/lib/vz/x.qcow2")
mixin._run_node_command = MagicMock(return_value="")
mock_vm = mock_node.qemu.return_value
mock_vm.config.get.return_value = {"unused0": "local:vm-110-disk-0"}
mock_vm.status.start.post.return_value = "UPID:pve1:136:start"
mock_node.tasks.return_value.status.get.return_value = {"status": "stopped", "exitstatus": "OK"}
keys = [
"ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAI/+k1 alice@example",
"ssh-rsa AAAAB3NzaC1yc2EAAAADAQAB/+k2 bob@example",
]
with patch("time.sleep"):
mixin.create_vm_from_cloud_init(
name="keys-vm",
image_url="https://cloud.debian.org/images/cloud/bookworm/latest/debian-12-genericcloud-amd64.qcow2",
cpu=1,
memory=512,
nics=[{"bridge": "vmbr0"}],
cloud_init_config={},
ssh_public_keys=keys,
)
call = next(c for c in mock_vm.config.post.call_args_list if "sshkeys" in c[1])
sshkeys = call[1]["sshkeys"]
assert unquote(sshkeys) == "\n".join(keys)
assert re.fullmatch(r"[-%a-zA-Z0-9_.!~*'()]*", sshkeys)
class _LocalShell:
"""The node's SSH session as a local ``sh``, *prelude* run before each command."""
def __init__(self, prelude: str) -> None:
self.prelude = prelude
def exec_command(self, command: str, timeout: int | None = None):
import subprocess
run = subprocess.run(
["sh", "-c", self.prelude + command], capture_output=True, timeout=timeout
)
return _mock_ssh_exec(run.returncode, run.stdout.decode(), run.stderr.decode())
# wget as it fails on a full disk: part of the file written, the reason on stderr.
_WGET_DISK_FULL = (
"wget() { "
'while [ $# -gt 0 ]; do [ "$1" = -O ] && o=$2; shift; done; '
'echo partial > "$o"; '
'echo "$o: No space left on device" >&2; return 3; }; '
)
def test_download_cloud_image_failure_says_why_and_leaves_no_tmp(tmp_path, monkeypatch):
"""A failed download names wget's reason and removes its .tmp (napalm-proxmox#3)."""
import napalm_proxmox.vm_provision_mixin as module
monkeypatch.setattr(module, "_IMAGE_CACHE_DIR", str(tmp_path))
mixin = ProxmoxVMProvisionMixin()
mixin._node_ssh = MagicMock(return_value=_LocalShell(_WGET_DISK_FULL))
with pytest.raises(RuntimeError, match="No space left on device") as err:
mixin._download_cloud_image("https://example.org/img/debian.qcow2", None, timeout=10)
assert "exit 3" in str(err.value)
assert list(tmp_path.iterdir()) == []
def test_download_cloud_image_does_not_silence_wget():
mixin = ProxmoxVMProvisionMixin()
mixin._run_node_command = MagicMock(side_effect=["MISSING", ""])
mixin._download_cloud_image("https://example.org/img/debian.qcow2", None, timeout=10)
download = mixin._run_node_command.call_args_list[1][0][0]
assert download.startswith("wget -nv ")