driver / standard (pull_request) Successful in 14s
driver / lint (pull_request) Successful in 13s
driver / types (pull_request) Successful in 1m3s
driver / test (3.12) (pull_request) Successful in 1m10s
driver / test (3.13) (pull_request) Successful in 1m3s
driver / lowest (pull_request) Successful in 33s
driver / build (pull_request) Successful in 33s
driver / security (pull_request) Successful in 1m27s
CI / driver (pull_request) Successful in 3m4s
driver / publish (pull_request) Skipped
get_clusters() reads /cluster/ha/* (PVE 8 and 9), /cluster/backup with pool members, the newest backup per guest on each backup storage (PBS included) and /cluster/config/nodes votes; each read is best-effort. migrate_vm() moves a guest of the connected node and follows the task (HA-managed guests through the HA manager), raising with the task log's ERROR lines; get_vm_migration_preflight() maps PVE's own pre-flight; set_node_maintenance() runs ha-manager over SSH. Part of NetOrk/netork#1820 (stage 7: #759, #760, #761).
106 lines
3.7 KiB
Python
106 lines
3.7 KiB
Python
"""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"])
|