feat: classify offline and degraded connectivity
This commit is contained in:
@@ -0,0 +1,178 @@
|
||||
"""Execute and parse multi-packet fping probes without shell interpolation."""
|
||||
|
||||
import asyncio
|
||||
from dataclasses import dataclass
|
||||
import re
|
||||
from typing import Iterable
|
||||
|
||||
|
||||
FPING_RESULT_PATTERN = re.compile(r"^\s*(\S+)\s+:\s*(.*?)\s*$")
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class ProbeResult:
|
||||
"""Aggregate result for one IP in one monitoring round."""
|
||||
|
||||
ip: str
|
||||
sent_count: int
|
||||
received_count: int
|
||||
average_rtt_ms: float | None
|
||||
is_valid: bool
|
||||
failure_reason: str | None = None
|
||||
|
||||
@property
|
||||
def packet_loss_percent(self) -> float:
|
||||
"""Return packet loss as a percentage of the attempted probes."""
|
||||
if self.sent_count <= 0:
|
||||
return 0.0
|
||||
return (
|
||||
(self.sent_count - self.received_count)
|
||||
/ self.sent_count
|
||||
* 100.0
|
||||
)
|
||||
|
||||
|
||||
def _invalid_result(
|
||||
ip: str,
|
||||
packets_per_round: int,
|
||||
reason: str,
|
||||
) -> ProbeResult:
|
||||
"""Build an invalid aggregate that cannot resemble measured packet loss."""
|
||||
return ProbeResult(
|
||||
ip=ip,
|
||||
sent_count=packets_per_round,
|
||||
received_count=0,
|
||||
average_rtt_ms=None,
|
||||
is_valid=False,
|
||||
failure_reason=reason,
|
||||
)
|
||||
|
||||
|
||||
def parse_fping_count_output(
|
||||
output: str,
|
||||
expected_ips: set[str],
|
||||
packets_per_round: int,
|
||||
) -> dict[str, ProbeResult]:
|
||||
"""Parse ``fping -C`` reply tokens for every expected IP.
|
||||
|
||||
A numeric token is a measured RTT and ``-`` is packet loss. Missing,
|
||||
duplicated, or malformed lines are represented as invalid aggregates.
|
||||
"""
|
||||
if packets_per_round <= 0:
|
||||
raise ValueError("packets_per_round must be positive")
|
||||
|
||||
parsed_lines: dict[str, list[str]] = {}
|
||||
duplicate_ips: set[str] = set()
|
||||
for line in output.splitlines():
|
||||
match = FPING_RESULT_PATTERN.match(line)
|
||||
if match is None:
|
||||
continue
|
||||
ip, replies = match.groups()
|
||||
if ip not in expected_ips:
|
||||
continue
|
||||
if ip in parsed_lines:
|
||||
duplicate_ips.add(ip)
|
||||
continue
|
||||
parsed_lines[ip] = replies.split()
|
||||
|
||||
results: dict[str, ProbeResult] = {}
|
||||
for ip in expected_ips:
|
||||
tokens = parsed_lines.get(ip)
|
||||
if tokens is None:
|
||||
results[ip] = _invalid_result(
|
||||
ip,
|
||||
packets_per_round,
|
||||
"missing fping output",
|
||||
)
|
||||
continue
|
||||
if ip in duplicate_ips or len(tokens) != packets_per_round:
|
||||
results[ip] = _invalid_result(
|
||||
ip,
|
||||
packets_per_round,
|
||||
"malformed fping replies",
|
||||
)
|
||||
continue
|
||||
|
||||
reply_times: list[float] = []
|
||||
malformed = False
|
||||
for token in tokens:
|
||||
if token == "-":
|
||||
continue
|
||||
try:
|
||||
reply_time = float(token)
|
||||
except ValueError:
|
||||
malformed = True
|
||||
break
|
||||
if reply_time < 0:
|
||||
malformed = True
|
||||
break
|
||||
reply_times.append(reply_time)
|
||||
|
||||
if malformed:
|
||||
results[ip] = _invalid_result(
|
||||
ip,
|
||||
packets_per_round,
|
||||
"malformed fping replies",
|
||||
)
|
||||
continue
|
||||
|
||||
average_rtt_ms = (
|
||||
sum(reply_times) / len(reply_times)
|
||||
if reply_times
|
||||
else None
|
||||
)
|
||||
results[ip] = ProbeResult(
|
||||
ip=ip,
|
||||
sent_count=packets_per_round,
|
||||
received_count=len(reply_times),
|
||||
average_rtt_ms=average_rtt_ms,
|
||||
is_valid=True,
|
||||
)
|
||||
|
||||
return results
|
||||
|
||||
|
||||
async def run_fping_count(
|
||||
ips: Iterable[str],
|
||||
packets_per_round: int,
|
||||
timeout_ms: int,
|
||||
executable: str,
|
||||
) -> dict[str, ProbeResult]:
|
||||
"""Run one fping count probe and return an aggregate for every IP."""
|
||||
ip_list = list(ips)
|
||||
if not ip_list:
|
||||
return {}
|
||||
|
||||
command = [
|
||||
executable,
|
||||
"-C",
|
||||
str(packets_per_round),
|
||||
"-q",
|
||||
"-t",
|
||||
str(timeout_ms),
|
||||
*ip_list,
|
||||
]
|
||||
try:
|
||||
process = await asyncio.create_subprocess_exec(
|
||||
*command,
|
||||
stdout=asyncio.subprocess.PIPE,
|
||||
stderr=asyncio.subprocess.PIPE,
|
||||
)
|
||||
stdout, stderr = await process.communicate()
|
||||
except Exception as error:
|
||||
reason = f"fping execution failed: {type(error).__name__}"
|
||||
return {
|
||||
ip: _invalid_result(ip, packets_per_round, reason)
|
||||
for ip in ip_list
|
||||
}
|
||||
|
||||
output = "\n".join(
|
||||
stream.decode("utf-8", errors="replace")
|
||||
for stream in (stdout, stderr)
|
||||
if stream
|
||||
)
|
||||
return parse_fping_count_output(
|
||||
output,
|
||||
set(ip_list),
|
||||
packets_per_round,
|
||||
)
|
||||
+221
-198
@@ -1,242 +1,265 @@
|
||||
"""
|
||||
异步 Ping 引擎
|
||||
"""Persist batched probe summaries and apply device health transitions."""
|
||||
|
||||
核心逻辑：
|
||||
1. 每轮从数据库加载所有启用设备，批量 fping
|
||||
2. 记录每台设备本轮 ping 结果（存活/响应时间）
|
||||
3. 状态机管理设备状态，判断是否从 online→offline 或 offline→online
|
||||
4. 结果通过回调或队列通知 alerter
|
||||
"""
|
||||
|
||||
import asyncio
|
||||
import subprocess
|
||||
import time
|
||||
import logging
|
||||
from datetime import datetime
|
||||
from typing import Optional, Callable, Awaitable
|
||||
import time
|
||||
from collections import defaultdict
|
||||
from dataclasses import dataclass
|
||||
from datetime import datetime
|
||||
from typing import Awaitable, Callable
|
||||
|
||||
from sqlalchemy import select, update
|
||||
from sqlalchemy import select
|
||||
from sqlalchemy.ext.asyncio import AsyncSession
|
||||
|
||||
from app.config import settings
|
||||
from app.models.device import Device, DeviceTypeEnum
|
||||
from app.config import Settings, settings
|
||||
from app.models.device import Device, DeviceMonitoringPolicy
|
||||
from app.models.ping_record import PingRecord
|
||||
from app.services.fping_runner import ProbeResult, run_fping_count
|
||||
from app.services.state_machine import StateDecision, evaluate_health
|
||||
|
||||
|
||||
logger = logging.getLogger("pingwatch.pinger")
|
||||
|
||||
|
||||
class PingResult:
|
||||
"""单台设备一轮 ping 的结果"""
|
||||
def __init__(self, device_id: int, is_alive: bool, response_time_ms: Optional[float] = None):
|
||||
self.device_id = device_id
|
||||
self.is_alive = is_alive
|
||||
self.response_time_ms = response_time_ms
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class DeviceStateChange:
|
||||
"""设备状态变化事件"""
|
||||
def __init__(self, device: Device, old_status: str, new_status: str, consecutive_failures: int):
|
||||
self.device = device
|
||||
self.old_status = old_status
|
||||
self.new_status = new_status
|
||||
self.consecutive_failures = consecutive_failures
|
||||
"""A persisted device health transition produced by one probe round."""
|
||||
|
||||
device: Device
|
||||
old_status: str
|
||||
new_status: str
|
||||
consecutive_failures: int
|
||||
event_type: str | None = None
|
||||
reason: str = ""
|
||||
|
||||
|
||||
class Pinger:
|
||||
"""
|
||||
Ping 引擎，使用 fping 批量并发检测。
|
||||
对所有设备进 ping，返回存活状态和响应时间。
|
||||
"""
|
||||
"""Run multi-packet fping probes and update monitored device health."""
|
||||
|
||||
def __init__(self):
|
||||
def __init__(self, runtime_settings: Settings | None = None):
|
||||
self._settings = runtime_settings or settings
|
||||
self._round_num = 0
|
||||
self._on_state_change: Optional[Callable[[DeviceStateChange], Awaitable[None]]] = None
|
||||
self._on_state_change: (
|
||||
Callable[[DeviceStateChange], Awaitable[None]] | None
|
||||
) = None
|
||||
|
||||
def on_state_change(self, callback: Callable[[DeviceStateChange], Awaitable[None]]):
|
||||
"""注册状态变化回调"""
|
||||
def on_state_change(
|
||||
self,
|
||||
callback: Callable[[DeviceStateChange], Awaitable[None]],
|
||||
) -> None:
|
||||
"""Register the legacy transition callback during scheduler migration."""
|
||||
self._on_state_change = callback
|
||||
|
||||
async def run_one_round(self, db: AsyncSession) -> list[PingResult]:
|
||||
"""
|
||||
执行一轮 ping 检测：
|
||||
1. 加载所有启用设备
|
||||
2. 批量 fping
|
||||
3. 记录结果
|
||||
4. 更新设备状态
|
||||
"""
|
||||
async def run_one_round(
|
||||
self,
|
||||
db: AsyncSession,
|
||||
) -> list[DeviceStateChange]:
|
||||
"""Persist one summary per enabled device and commit state changes once."""
|
||||
self._round_num += 1
|
||||
round_num = self._round_num
|
||||
|
||||
# 1. 加载启用设备
|
||||
result = await db.execute(
|
||||
select(Device).where(Device.is_enabled == True)
|
||||
query_result = await db.execute(
|
||||
select(Device).where(Device.is_enabled.is_(True))
|
||||
)
|
||||
devices = list(result.scalars().all())
|
||||
|
||||
devices = list(query_result.scalars().all())
|
||||
if not devices:
|
||||
logger.info(f"[Round {round_num}] 没有启用的设备")
|
||||
logger.info("[Round %s] no enabled devices", round_num)
|
||||
return []
|
||||
|
||||
logger.info(f"[Round {round_num}] 开始检测 {len(devices)} 台设备")
|
||||
|
||||
# 2. 批量 ping
|
||||
start_time = time.time()
|
||||
ip_to_device = {d.ip: d for d in devices}
|
||||
ip_list = list(ip_to_device.keys())
|
||||
|
||||
ping_results_map = await self._batch_ping(ip_list)
|
||||
|
||||
# 3. 构造结果
|
||||
results: list[PingResult] = []
|
||||
for ip, dev in ip_to_device.items():
|
||||
is_alive, rtt = ping_results_map.get(ip, (False, None))
|
||||
results.append(PingResult(device_id=dev.id, is_alive=is_alive, response_time_ms=rtt))
|
||||
|
||||
elapsed = time.time() - start_time
|
||||
alive_count = sum(1 for r in results if r.is_alive)
|
||||
|
||||
# 4. 批量写入 ping_records
|
||||
now = datetime.now()
|
||||
records = [
|
||||
PingRecord(
|
||||
device_id=r.device_id,
|
||||
is_alive=r.is_alive,
|
||||
response_time_ms=r.response_time_ms,
|
||||
round_num=round_num,
|
||||
created_at=now,
|
||||
started_at = time.monotonic()
|
||||
policies = {
|
||||
device.id: DeviceMonitoringPolicy.from_device(
|
||||
device,
|
||||
self._settings,
|
||||
)
|
||||
for r in results
|
||||
]
|
||||
db.add_all(records)
|
||||
for device in devices
|
||||
}
|
||||
probe_results = await self._probe_by_packet_count(devices, policies)
|
||||
observed_at = datetime.now()
|
||||
|
||||
records_by_device: dict[int, PingRecord] = {}
|
||||
for device in devices:
|
||||
policy = policies[device.id]
|
||||
probe = probe_results.get(
|
||||
(device.id, policy.probe_packets_per_round)
|
||||
)
|
||||
if probe is None:
|
||||
probe = ProbeResult(
|
||||
ip=device.ip,
|
||||
sent_count=policy.probe_packets_per_round,
|
||||
received_count=0,
|
||||
average_rtt_ms=None,
|
||||
is_valid=False,
|
||||
failure_reason="missing runner result",
|
||||
)
|
||||
record = self._to_record(
|
||||
device=device,
|
||||
probe=probe,
|
||||
round_num=round_num,
|
||||
observed_at=observed_at,
|
||||
)
|
||||
records_by_device[device.id] = record
|
||||
db.add(record)
|
||||
|
||||
await db.flush()
|
||||
|
||||
# 5. 更新设备状态（状态机）
|
||||
device_map = {d.id: d for d in devices}
|
||||
for r in results:
|
||||
dev = device_map.get(r.device_id)
|
||||
if not dev:
|
||||
changes: list[DeviceStateChange] = []
|
||||
for device in devices:
|
||||
record = records_by_device[device.id]
|
||||
device.last_ping_time = observed_at
|
||||
if not record.is_valid:
|
||||
continue
|
||||
|
||||
old_status = dev.current_status
|
||||
if r.is_alive:
|
||||
dev.consecutive_failures = 0
|
||||
dev.last_ping_time = now
|
||||
dev.last_online_time = now
|
||||
dev.current_status = "online"
|
||||
else:
|
||||
dev.consecutive_failures = (dev.consecutive_failures or 0) + 1
|
||||
dev.last_ping_time = now
|
||||
if dev.consecutive_failures >= dev.alert_threshold:
|
||||
if dev.current_status != "offline":
|
||||
dev.current_status = "offline"
|
||||
dev.last_offline_time = now
|
||||
else:
|
||||
if dev.current_status == "online":
|
||||
dev.current_status = "checking"
|
||||
|
||||
# 状态变化回调
|
||||
if old_status != dev.current_status and self._on_state_change:
|
||||
change = DeviceStateChange(
|
||||
device=dev,
|
||||
old_status=old_status,
|
||||
new_status=dev.current_status,
|
||||
consecutive_failures=dev.consecutive_failures,
|
||||
)
|
||||
await self._on_state_change(change)
|
||||
policy = policies[device.id]
|
||||
recent = await self._load_recent_valid(
|
||||
db,
|
||||
device.id,
|
||||
policy,
|
||||
)
|
||||
decision = evaluate_health(
|
||||
device.current_status or "unknown",
|
||||
recent,
|
||||
policy,
|
||||
)
|
||||
self._update_observation_fields(device, recent, observed_at)
|
||||
change = self._apply_decision(
|
||||
device,
|
||||
decision,
|
||||
observed_at,
|
||||
)
|
||||
if change is not None:
|
||||
changes.append(change)
|
||||
|
||||
await db.commit()
|
||||
|
||||
if self._on_state_change is not None:
|
||||
for change in changes:
|
||||
await self._on_state_change(change)
|
||||
|
||||
elapsed = time.monotonic() - started_at
|
||||
logger.info(
|
||||
f"[Round {round_num}] 完成: {alive_count}/{len(devices)} 在线, "
|
||||
f"耗时 {elapsed:.2f}s"
|
||||
"[Round %s] persisted %s device summaries with %s transitions "
|
||||
"in %.2fs",
|
||||
round_num,
|
||||
len(devices),
|
||||
len(changes),
|
||||
elapsed,
|
||||
)
|
||||
return changes
|
||||
|
||||
async def _probe_by_packet_count(
|
||||
self,
|
||||
devices: list[Device],
|
||||
policies: dict[int, DeviceMonitoringPolicy],
|
||||
) -> dict[tuple[int, int], ProbeResult]:
|
||||
"""Run one argv-safe batch for each distinct packet-count policy."""
|
||||
groups: dict[int, list[Device]] = defaultdict(list)
|
||||
for device in devices:
|
||||
groups[policies[device.id].probe_packets_per_round].append(device)
|
||||
|
||||
results: dict[tuple[int, int], ProbeResult] = {}
|
||||
timeout_ms = int(self._settings.PING_TIMEOUT_SECONDS * 1000)
|
||||
for packet_count, grouped_devices in groups.items():
|
||||
unique_ips = list(dict.fromkeys(
|
||||
device.ip for device in grouped_devices
|
||||
))
|
||||
batch = await run_fping_count(
|
||||
unique_ips,
|
||||
packets_per_round=packet_count,
|
||||
timeout_ms=timeout_ms,
|
||||
executable=self._settings.FPING_PATH,
|
||||
)
|
||||
for device in grouped_devices:
|
||||
probe = batch.get(device.ip)
|
||||
if probe is not None:
|
||||
results[(device.id, packet_count)] = probe
|
||||
return results
|
||||
|
||||
async def _batch_ping(self, ip_list: list[str]) -> dict[str, tuple[bool, Optional[float]]]:
|
||||
"""
|
||||
使用 fping 批量 ping
|
||||
返回: { ip: (is_alive, response_time_ms) }
|
||||
"""
|
||||
if not ip_list:
|
||||
return {}
|
||||
@staticmethod
|
||||
def _to_record(
|
||||
device: Device,
|
||||
probe: ProbeResult,
|
||||
round_num: int,
|
||||
observed_at: datetime,
|
||||
) -> PingRecord:
|
||||
"""Convert a runner aggregate into the persisted compatibility model."""
|
||||
return PingRecord(
|
||||
device_id=device.id,
|
||||
is_alive=probe.is_valid and probe.received_count > 0,
|
||||
response_time_ms=probe.average_rtt_ms,
|
||||
round_num=round_num,
|
||||
created_at=observed_at,
|
||||
sent_count=probe.sent_count,
|
||||
received_count=probe.received_count,
|
||||
packet_loss_percent=(
|
||||
probe.packet_loss_percent if probe.is_valid else None
|
||||
),
|
||||
average_rtt_ms=probe.average_rtt_ms,
|
||||
is_valid=probe.is_valid,
|
||||
failure_reason=probe.failure_reason,
|
||||
)
|
||||
|
||||
try:
|
||||
# fping 一次性 ping 多个 IP
|
||||
# -c 1: 每个 IP 发 1 个包
|
||||
# -t: 超时毫秒
|
||||
timeout_ms = int(settings.PING_TIMEOUT_SECONDS * 1000)
|
||||
cmd = [
|
||||
settings.FPING_PATH,
|
||||
"-c", "1",
|
||||
"-t", str(timeout_ms),
|
||||
"-e", # 显示响应时间
|
||||
] + ip_list
|
||||
@staticmethod
|
||||
async def _load_recent_valid(
|
||||
db: AsyncSession,
|
||||
device_id: int,
|
||||
policy: DeviceMonitoringPolicy,
|
||||
) -> list[PingRecord]:
|
||||
"""Load enough valid summaries for every configured transition rule."""
|
||||
history_size = max(
|
||||
policy.offline_consecutive_rounds,
|
||||
policy.degraded_window_rounds,
|
||||
policy.recovery_consecutive_clean_rounds,
|
||||
)
|
||||
query_result = await db.execute(
|
||||
select(PingRecord)
|
||||
.where(PingRecord.device_id == device_id)
|
||||
.where(PingRecord.is_valid.is_(True))
|
||||
.order_by(PingRecord.created_at.desc(), PingRecord.id.desc())
|
||||
.limit(history_size)
|
||||
)
|
||||
newest_first = list(query_result.scalars().all())
|
||||
return list(reversed(newest_first))
|
||||
|
||||
proc = await asyncio.create_subprocess_exec(
|
||||
*cmd,
|
||||
stdout=subprocess.PIPE,
|
||||
stderr=subprocess.PIPE,
|
||||
)
|
||||
stdout, stderr = await proc.communicate()
|
||||
@staticmethod
|
||||
def _update_observation_fields(
|
||||
device: Device,
|
||||
recent: list[PingRecord],
|
||||
observed_at: datetime,
|
||||
) -> None:
|
||||
"""Maintain compatibility timestamps and the full-loss streak counter."""
|
||||
current = recent[-1]
|
||||
if current.received_count > 0:
|
||||
device.last_online_time = observed_at
|
||||
|
||||
result_map: dict[str, tuple[bool, Optional[float]]] = {}
|
||||
full_loss_count = 0
|
||||
for record in reversed(recent):
|
||||
if record.sent_count and record.received_count == 0:
|
||||
full_loss_count += 1
|
||||
else:
|
||||
break
|
||||
device.consecutive_failures = full_loss_count
|
||||
|
||||
# fping 标准输出逐行: "IP : xmt/rcv/%loss = 1/1/0%, rtt min/avg/max = 0.12/0.12/0.12"
|
||||
# 或 "IP : xmt/rcv/%loss = 1/0/100%"
|
||||
for line in stdout.decode("utf-8", errors="replace").splitlines():
|
||||
line = line.strip()
|
||||
if ":" not in line:
|
||||
continue
|
||||
ip = line.split(":")[0].strip()
|
||||
# 解析响应时间
|
||||
if "rtt" in line:
|
||||
try:
|
||||
# 提取 avg rtt
|
||||
rtt_part = line.split("rtt")[1]
|
||||
# 格式: min/avg/max = 0.12/0.12/0.12
|
||||
if "=" in rtt_part:
|
||||
avg_rtt_str = rtt_part.split("=")[1].strip().split("/")[1]
|
||||
rtt_ms = float(avg_rtt_str)
|
||||
else:
|
||||
rtt_ms = None
|
||||
except (IndexError, ValueError):
|
||||
rtt_ms = None
|
||||
result_map[ip] = (True, rtt_ms)
|
||||
else:
|
||||
result_map[ip] = (False, None)
|
||||
@staticmethod
|
||||
def _apply_decision(
|
||||
device: Device,
|
||||
decision: StateDecision,
|
||||
observed_at: datetime,
|
||||
) -> DeviceStateChange | None:
|
||||
"""Apply one pure decision and materialize its transition value."""
|
||||
old_status = device.current_status or "unknown"
|
||||
if decision.next_status == old_status:
|
||||
return None
|
||||
|
||||
return result_map
|
||||
device.current_status = decision.next_status
|
||||
if decision.next_status == "offline":
|
||||
device.last_offline_time = observed_at
|
||||
if decision.next_status == "online":
|
||||
device.last_online_time = observed_at
|
||||
|
||||
except FileNotFoundError:
|
||||
logger.warning("fping 未找到，回退到系统 ping (串行)")
|
||||
return await self._fallback_ping(ip_list)
|
||||
except Exception as e:
|
||||
logger.error(f"fping 异常: {e}")
|
||||
return await self._fallback_ping(ip_list)
|
||||
|
||||
async def _fallback_ping(self, ip_list: list[str]) -> dict[str, tuple[bool, Optional[float]]]:
|
||||
"""回退方案：使用系统 ping，并发执行"""
|
||||
async def ping_one(ip: str) -> tuple[str, bool, Optional[float]]:
|
||||
try:
|
||||
timeout = settings.PING_TIMEOUT_SECONDS
|
||||
cmd = ["ping", "-c", "1", "-W", str(int(timeout)), ip]
|
||||
start = time.time()
|
||||
proc = await asyncio.create_subprocess_exec(
|
||||
*cmd, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL,
|
||||
)
|
||||
await proc.wait()
|
||||
elapsed = (time.time() - start) * 1000
|
||||
return ip, proc.returncode == 0, round(elapsed, 2)
|
||||
except Exception:
|
||||
return ip, False, None
|
||||
|
||||
tasks = [ping_one(ip) for ip in ip_list]
|
||||
sem = asyncio.Semaphore(settings.PING_CONCURRENCY)
|
||||
|
||||
async def bounded_ping(ip: str):
|
||||
async with sem:
|
||||
return await ping_one(ip)
|
||||
|
||||
results = await asyncio.gather(*[bounded_ping(ip) for ip in ip_list])
|
||||
return {ip: (alive, rtt) for ip, alive, rtt in results}
|
||||
return DeviceStateChange(
|
||||
device=device,
|
||||
old_status=old_status,
|
||||
new_status=decision.next_status,
|
||||
consecutive_failures=device.consecutive_failures or 0,
|
||||
event_type=decision.event_type,
|
||||
reason=decision.reason,
|
||||
)
|
||||
|
||||
@@ -0,0 +1,126 @@
|
||||
"""Pure state transitions for device connectivity health."""
|
||||
|
||||
from dataclasses import dataclass
|
||||
from typing import Protocol, Sequence
|
||||
|
||||
from app.models.device import DeviceMonitoringPolicy
|
||||
|
||||
|
||||
class ProbeSample(Protocol):
|
||||
"""Packet summary attributes required for health evaluation."""
|
||||
|
||||
sent_count: int
|
||||
received_count: int
|
||||
is_valid: bool
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class StateDecision:
|
||||
"""Result of evaluating recent probes against a monitoring policy."""
|
||||
|
||||
next_status: str
|
||||
event_type: str | None
|
||||
should_notify: bool
|
||||
reason: str
|
||||
|
||||
|
||||
def _retain(previous_status: str, reason: str) -> StateDecision:
|
||||
"""Return a non-notifying decision that preserves current health."""
|
||||
return StateDecision(
|
||||
next_status=previous_status,
|
||||
event_type=None,
|
||||
should_notify=False,
|
||||
reason=reason,
|
||||
)
|
||||
|
||||
|
||||
def _transition(
|
||||
previous_status: str,
|
||||
next_status: str,
|
||||
event_type: str,
|
||||
reason: str,
|
||||
) -> StateDecision:
|
||||
"""Return a transition event only when the persisted status changes."""
|
||||
if previous_status == next_status:
|
||||
return _retain(previous_status, reason)
|
||||
return StateDecision(
|
||||
next_status=next_status,
|
||||
event_type=event_type,
|
||||
should_notify=True,
|
||||
reason=reason,
|
||||
)
|
||||
|
||||
|
||||
def _is_valid(sample: ProbeSample) -> bool:
|
||||
"""Reject impossible summaries before they influence device health."""
|
||||
return (
|
||||
sample.is_valid
|
||||
and sample.sent_count > 0
|
||||
and 0 <= sample.received_count <= sample.sent_count
|
||||
)
|
||||
|
||||
|
||||
def evaluate_health(
|
||||
previous_status: str,
|
||||
recent: Sequence[ProbeSample],
|
||||
policy: DeviceMonitoringPolicy,
|
||||
) -> StateDecision:
|
||||
"""Evaluate ordered oldest-to-newest probe summaries.
|
||||
|
||||
Invalid input retains state. Offline has priority over aggregate degraded
|
||||
loss, and an alerted state recovers only after the configured clean streak.
|
||||
"""
|
||||
if not recent:
|
||||
return _retain(previous_status, "no probe history")
|
||||
if any(not _is_valid(sample) for sample in recent):
|
||||
return _retain(previous_status, "invalid probe history")
|
||||
|
||||
offline_rounds = policy.offline_consecutive_rounds
|
||||
if len(recent) >= offline_rounds and all(
|
||||
sample.received_count == 0
|
||||
for sample in recent[-offline_rounds:]
|
||||
):
|
||||
return _transition(
|
||||
previous_status,
|
||||
"offline",
|
||||
"offline",
|
||||
f"{offline_rounds} consecutive full-loss rounds",
|
||||
)
|
||||
|
||||
recovery_rounds = policy.recovery_consecutive_clean_rounds
|
||||
if (
|
||||
previous_status in {"offline", "degraded"}
|
||||
and len(recent) >= recovery_rounds
|
||||
and all(
|
||||
sample.received_count == sample.sent_count
|
||||
for sample in recent[-recovery_rounds:]
|
||||
)
|
||||
):
|
||||
return _transition(
|
||||
previous_status,
|
||||
"online",
|
||||
"recovered",
|
||||
f"{recovery_rounds} consecutive clean rounds",
|
||||
)
|
||||
|
||||
degraded_rounds = policy.degraded_window_rounds
|
||||
if len(recent) >= degraded_rounds:
|
||||
window = recent[-degraded_rounds:]
|
||||
sent_count = sum(sample.sent_count for sample in window)
|
||||
received_count = sum(sample.received_count for sample in window)
|
||||
loss_percent = (sent_count - received_count) / sent_count * 100.0
|
||||
if (
|
||||
previous_status != "offline"
|
||||
and loss_percent >= policy.degraded_loss_percent
|
||||
):
|
||||
return _transition(
|
||||
previous_status,
|
||||
"degraded",
|
||||
"degraded",
|
||||
(
|
||||
f"window packet loss {loss_percent:.2f}% is at or above "
|
||||
f"{policy.degraded_loss_percent:.2f}%"
|
||||
),
|
||||
)
|
||||
|
||||
return _retain(previous_status, "no transition threshold met")
|
||||
@@ -0,0 +1,322 @@
|
||||
"""Behavior tests for multi-packet fping execution and persistence."""
|
||||
|
||||
import asyncio
|
||||
from datetime import datetime, timedelta
|
||||
|
||||
import pytest
|
||||
from sqlalchemy import select
|
||||
from sqlalchemy.ext.asyncio import AsyncSession, create_async_engine
|
||||
|
||||
from app.config import Settings
|
||||
from app.models.device import Base, Device, DeviceTypeEnum
|
||||
from app.models.ping_record import PingRecord
|
||||
from app.services.fping_runner import (
|
||||
ProbeResult,
|
||||
parse_fping_count_output,
|
||||
run_fping_count,
|
||||
)
|
||||
from app.services.pinger import Pinger
|
||||
|
||||
|
||||
@pytest.fixture
|
||||
async def db_session() -> AsyncSession:
|
||||
"""Provide an isolated database for pinger transaction tests."""
|
||||
engine = create_async_engine("sqlite+aiosqlite:///:memory:")
|
||||
try:
|
||||
async with engine.begin() as connection:
|
||||
await connection.run_sync(Base.metadata.create_all)
|
||||
async with AsyncSession(engine, expire_on_commit=False) as session:
|
||||
yield session
|
||||
finally:
|
||||
await engine.dispose()
|
||||
|
||||
|
||||
def test_parses_loss_and_average_rtt():
|
||||
"""Numeric replies count as received packets and dashes count as loss."""
|
||||
output = "10.0.0.8 : 12.4 - 12.6\n10.0.0.9 : - - -"
|
||||
|
||||
result = parse_fping_count_output(
|
||||
output,
|
||||
{"10.0.0.8", "10.0.0.9"},
|
||||
3,
|
||||
)
|
||||
|
||||
assert result["10.0.0.8"].received_count == 2
|
||||
assert result["10.0.0.8"].average_rtt_ms == 12.5
|
||||
assert result["10.0.0.8"].packet_loss_percent == pytest.approx(
|
||||
33.3333333333
|
||||
)
|
||||
assert result["10.0.0.9"].packet_loss_percent == 100.0
|
||||
|
||||
|
||||
def test_marks_missing_and_malformed_expected_output_invalid():
|
||||
"""Incomplete fping output cannot be interpreted as device packet loss."""
|
||||
output = "10.0.0.8 : 10.2 bad-token -"
|
||||
|
||||
result = parse_fping_count_output(
|
||||
output,
|
||||
{"10.0.0.8", "10.0.0.9"},
|
||||
3,
|
||||
)
|
||||
|
||||
assert result["10.0.0.8"].is_valid is False
|
||||
assert result["10.0.0.8"].failure_reason == "malformed fping replies"
|
||||
assert result["10.0.0.9"].is_valid is False
|
||||
assert result["10.0.0.9"].failure_reason == "missing fping output"
|
||||
|
||||
|
||||
def test_parses_ipv6_without_confusing_address_colons_for_delimiter():
|
||||
"""An IPv6 address remains the result key when parsing the output delimiter."""
|
||||
result = parse_fping_count_output(
|
||||
"2001:db8::8 : 1.25 - 1.75",
|
||||
{"2001:db8::8"},
|
||||
3,
|
||||
)
|
||||
|
||||
assert result["2001:db8::8"].is_valid is True
|
||||
assert result["2001:db8::8"].received_count == 2
|
||||
assert result["2001:db8::8"].average_rtt_ms == 1.5
|
||||
|
||||
|
||||
async def test_runner_uses_argument_vector_and_parses_quiet_stderr(monkeypatch):
|
||||
"""The runner passes IPs as argv entries and reads fping -C quiet output."""
|
||||
observed = {}
|
||||
|
||||
class CompletedProcess:
|
||||
returncode = 1
|
||||
|
||||
async def communicate(self):
|
||||
return b"", b"10.0.0.8 : 4.2 - 4.4\n"
|
||||
|
||||
async def fake_create_subprocess_exec(*args, **kwargs):
|
||||
observed["args"] = args
|
||||
observed["kwargs"] = kwargs
|
||||
return CompletedProcess()
|
||||
|
||||
monkeypatch.setattr(
|
||||
asyncio,
|
||||
"create_subprocess_exec",
|
||||
fake_create_subprocess_exec,
|
||||
)
|
||||
|
||||
result = await run_fping_count(
|
||||
["10.0.0.8"],
|
||||
packets_per_round=3,
|
||||
timeout_ms=800,
|
||||
executable="/test/fping",
|
||||
)
|
||||
|
||||
assert observed["args"] == (
|
||||
"/test/fping",
|
||||
"-C",
|
||||
"3",
|
||||
"-q",
|
||||
"-t",
|
||||
"800",
|
||||
"10.0.0.8",
|
||||
)
|
||||
assert observed["kwargs"] == {
|
||||
"stdout": asyncio.subprocess.PIPE,
|
||||
"stderr": asyncio.subprocess.PIPE,
|
||||
}
|
||||
assert result["10.0.0.8"].received_count == 2
|
||||
|
||||
|
||||
async def test_runner_failure_returns_invalid_result_instead_of_packet_loss(
|
||||
monkeypatch,
|
||||
):
|
||||
"""A process communication failure remains an invalid probe aggregate."""
|
||||
|
||||
class BrokenProcess:
|
||||
async def communicate(self):
|
||||
raise RuntimeError("simulated process pipe failure")
|
||||
|
||||
async def fake_create_subprocess_exec(*args, **kwargs):
|
||||
return BrokenProcess()
|
||||
|
||||
monkeypatch.setattr(
|
||||
asyncio,
|
||||
"create_subprocess_exec",
|
||||
fake_create_subprocess_exec,
|
||||
)
|
||||
|
||||
result = await run_fping_count(
|
||||
["10.0.0.8"],
|
||||
packets_per_round=3,
|
||||
timeout_ms=800,
|
||||
executable="/test/fping",
|
||||
)
|
||||
|
||||
assert result["10.0.0.8"].is_valid is False
|
||||
assert result["10.0.0.8"].failure_reason == (
|
||||
"fping execution failed: RuntimeError"
|
||||
)
|
||||
|
||||
|
||||
async def test_pinger_persists_one_summary_for_each_enabled_device(
|
||||
db_session,
|
||||
monkeypatch,
|
||||
):
|
||||
"""One monitoring round writes exactly one aggregate per enabled device."""
|
||||
first = Device(
|
||||
name="edge-a",
|
||||
ip="10.0.0.8",
|
||||
device_type=DeviceTypeEnum.switch,
|
||||
current_status="online",
|
||||
is_enabled=True,
|
||||
)
|
||||
second = Device(
|
||||
name="edge-b",
|
||||
ip="10.0.0.9",
|
||||
device_type=DeviceTypeEnum.switch,
|
||||
current_status="online",
|
||||
is_enabled=True,
|
||||
)
|
||||
disabled = Device(
|
||||
name="edge-disabled",
|
||||
ip="10.0.0.10",
|
||||
device_type=DeviceTypeEnum.switch,
|
||||
current_status="online",
|
||||
is_enabled=False,
|
||||
)
|
||||
db_session.add_all([first, second, disabled])
|
||||
await db_session.commit()
|
||||
|
||||
async def fake_run_fping_count(
|
||||
ips,
|
||||
packets_per_round,
|
||||
timeout_ms,
|
||||
executable,
|
||||
):
|
||||
assert set(ips) == {"10.0.0.8", "10.0.0.9"}
|
||||
return {
|
||||
"10.0.0.8": ProbeResult("10.0.0.8", 3, 2, 8.5, True),
|
||||
"10.0.0.9": ProbeResult("10.0.0.9", 3, 3, 9.5, True),
|
||||
}
|
||||
|
||||
monkeypatch.setattr(
|
||||
"app.services.pinger.run_fping_count",
|
||||
fake_run_fping_count,
|
||||
)
|
||||
|
||||
changes = await Pinger(Settings()).run_one_round(db_session)
|
||||
records = list(
|
||||
(
|
||||
await db_session.execute(
|
||||
select(PingRecord).order_by(PingRecord.device_id)
|
||||
)
|
||||
)
|
||||
.scalars()
|
||||
.all()
|
||||
)
|
||||
|
||||
assert changes == []
|
||||
assert len(records) == 2
|
||||
assert [record.device_id for record in records] == [first.id, second.id]
|
||||
assert records[0].sent_count == 3
|
||||
assert records[0].received_count == 2
|
||||
assert records[0].packet_loss_percent == pytest.approx(33.3333333333)
|
||||
assert records[0].average_rtt_ms == 8.5
|
||||
assert records[0].is_valid is True
|
||||
|
||||
|
||||
async def test_pinger_persists_invalid_probe_without_changing_device_state(
|
||||
db_session,
|
||||
monkeypatch,
|
||||
):
|
||||
"""A runner failure remains observable but cannot mark a device offline."""
|
||||
device = Device(
|
||||
name="edge-a",
|
||||
ip="10.0.0.8",
|
||||
device_type=DeviceTypeEnum.switch,
|
||||
current_status="online",
|
||||
is_enabled=True,
|
||||
)
|
||||
db_session.add(device)
|
||||
await db_session.commit()
|
||||
|
||||
async def fake_run_fping_count(*args, **kwargs):
|
||||
return {
|
||||
device.ip: ProbeResult(
|
||||
device.ip,
|
||||
3,
|
||||
0,
|
||||
None,
|
||||
False,
|
||||
"fping executable not found",
|
||||
)
|
||||
}
|
||||
|
||||
monkeypatch.setattr(
|
||||
"app.services.pinger.run_fping_count",
|
||||
fake_run_fping_count,
|
||||
)
|
||||
|
||||
changes = await Pinger(Settings()).run_one_round(db_session)
|
||||
record = (
|
||||
await db_session.execute(select(PingRecord).where(PingRecord.device_id == device.id))
|
||||
).scalar_one()
|
||||
|
||||
assert changes == []
|
||||
assert device.current_status == "online"
|
||||
assert record.is_valid is False
|
||||
assert record.packet_loss_percent is None
|
||||
assert record.failure_reason == "fping executable not found"
|
||||
|
||||
|
||||
async def test_pinger_evaluates_recent_valid_summaries_and_returns_change(
|
||||
db_session,
|
||||
monkeypatch,
|
||||
):
|
||||
"""A newly persisted full-loss summary completes the offline streak."""
|
||||
device = Device(
|
||||
name="edge-a",
|
||||
ip="10.0.0.8",
|
||||
device_type=DeviceTypeEnum.switch,
|
||||
current_status="online",
|
||||
is_enabled=True,
|
||||
offline_consecutive_rounds=2,
|
||||
)
|
||||
db_session.add(device)
|
||||
await db_session.flush()
|
||||
db_session.add(
|
||||
PingRecord(
|
||||
device_id=device.id,
|
||||
is_alive=False,
|
||||
response_time_ms=None,
|
||||
round_num=1,
|
||||
sent_count=3,
|
||||
received_count=0,
|
||||
packet_loss_percent=100.0,
|
||||
average_rtt_ms=None,
|
||||
is_valid=True,
|
||||
created_at=datetime.now() - timedelta(seconds=30),
|
||||
)
|
||||
)
|
||||
await db_session.commit()
|
||||
|
||||
async def fake_run_fping_count(*args, **kwargs):
|
||||
return {
|
||||
device.ip: ProbeResult(
|
||||
device.ip,
|
||||
3,
|
||||
0,
|
||||
None,
|
||||
True,
|
||||
)
|
||||
}
|
||||
|
||||
monkeypatch.setattr(
|
||||
"app.services.pinger.run_fping_count",
|
||||
fake_run_fping_count,
|
||||
)
|
||||
|
||||
changes = await Pinger(Settings()).run_one_round(db_session)
|
||||
|
||||
assert len(changes) == 1
|
||||
assert changes[0].device is device
|
||||
assert changes[0].old_status == "online"
|
||||
assert changes[0].new_status == "offline"
|
||||
assert changes[0].event_type == "offline"
|
||||
assert device.current_status == "offline"
|
||||
assert device.last_offline_time is not None
|
||||
@@ -0,0 +1,139 @@
|
||||
"""Unit tests for the connectivity health state machine."""
|
||||
|
||||
from dataclasses import dataclass
|
||||
|
||||
import pytest
|
||||
|
||||
from app.models.device import DeviceMonitoringPolicy
|
||||
from app.services.state_machine import evaluate_health
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class RecentProbe:
|
||||
"""Minimal real-data-shaped record consumed by the pure state machine."""
|
||||
|
||||
sent_count: int
|
||||
received_count: int
|
||||
is_valid: bool = True
|
||||
|
||||
|
||||
def loss(sent_count: int) -> RecentProbe:
|
||||
return RecentProbe(sent_count=sent_count, received_count=0)
|
||||
|
||||
|
||||
def partial(sent_count: int, received_count: int) -> RecentProbe:
|
||||
return RecentProbe(
|
||||
sent_count=sent_count,
|
||||
received_count=received_count,
|
||||
)
|
||||
|
||||
|
||||
def clean(sent_count: int) -> RecentProbe:
|
||||
return RecentProbe(sent_count=sent_count, received_count=sent_count)
|
||||
|
||||
|
||||
@pytest.fixture
|
||||
def policy() -> DeviceMonitoringPolicy:
|
||||
return DeviceMonitoringPolicy(
|
||||
probe_packets_per_round=3,
|
||||
offline_consecutive_rounds=2,
|
||||
degraded_window_rounds=5,
|
||||
degraded_loss_percent=20.0,
|
||||
recovery_consecutive_clean_rounds=3,
|
||||
)
|
||||
|
||||
|
||||
def test_two_full_loss_rounds_take_device_offline(policy):
|
||||
"""Configured consecutive full-loss rounds transition to offline once."""
|
||||
decision = evaluate_health("online", [loss(3), loss(3)], policy)
|
||||
|
||||
assert (decision.next_status, decision.event_type) == (
|
||||
"offline",
|
||||
"offline",
|
||||
)
|
||||
assert decision.should_notify is True
|
||||
|
||||
|
||||
def test_offline_priority_wins_over_aggregate_degraded_loss(policy):
|
||||
"""Full-loss streaks are classified offline even when the window also degrades."""
|
||||
recent = [partial(3, 2)] * 3 + [loss(3), loss(3)]
|
||||
|
||||
decision = evaluate_health("online", recent, policy)
|
||||
|
||||
assert decision.next_status == "offline"
|
||||
assert decision.event_type == "offline"
|
||||
|
||||
|
||||
def test_five_round_window_with_loss_is_degraded(policy):
|
||||
"""Aggregate loss at the configured window threshold is degraded."""
|
||||
decision = evaluate_health(
|
||||
"online",
|
||||
[partial(3, 2)] * 5,
|
||||
policy,
|
||||
)
|
||||
|
||||
assert decision.next_status == "degraded"
|
||||
assert decision.event_type == "degraded"
|
||||
|
||||
|
||||
def test_degraded_requires_a_complete_window(policy):
|
||||
"""A partial history cannot satisfy an aggregate-window decision."""
|
||||
decision = evaluate_health(
|
||||
"online",
|
||||
[partial(3, 2)] * 4,
|
||||
policy,
|
||||
)
|
||||
|
||||
assert decision.next_status == "online"
|
||||
assert decision.event_type is None
|
||||
|
||||
|
||||
def test_three_clean_rounds_recovers(policy):
|
||||
"""An alerted device recovers only after the configured clean streak."""
|
||||
decision = evaluate_health(
|
||||
"offline",
|
||||
[clean(3)] * 3,
|
||||
policy,
|
||||
)
|
||||
|
||||
assert (decision.next_status, decision.event_type) == (
|
||||
"online",
|
||||
"recovered",
|
||||
)
|
||||
assert decision.should_notify is True
|
||||
|
||||
|
||||
def test_clean_online_device_does_not_emit_duplicate_event(policy):
|
||||
"""Clean probes retain an already-online state without notifying."""
|
||||
decision = evaluate_health(
|
||||
"online",
|
||||
[clean(3)] * 5,
|
||||
policy,
|
||||
)
|
||||
|
||||
assert decision.next_status == "online"
|
||||
assert decision.event_type is None
|
||||
assert decision.should_notify is False
|
||||
|
||||
|
||||
def test_latest_invalid_probe_retains_state(policy):
|
||||
"""A malformed current result cannot reuse old loss to trigger a fault."""
|
||||
recent = [loss(3), RecentProbe(3, 0, is_valid=False)]
|
||||
|
||||
decision = evaluate_health("online", recent, policy)
|
||||
|
||||
assert decision.next_status == "online"
|
||||
assert decision.event_type is None
|
||||
assert decision.should_notify is False
|
||||
|
||||
|
||||
def test_offline_does_not_recover_before_clean_streak_is_complete(policy):
|
||||
"""Recovery debounce retains offline until enough clean rounds exist."""
|
||||
decision = evaluate_health(
|
||||
"offline",
|
||||
[clean(3), clean(3)],
|
||||
policy,
|
||||
)
|
||||
|
||||
assert decision.next_status == "offline"
|
||||
assert decision.event_type is None
|
||||
Reference in New Issue
Block a user