dashboards/tests/test_poller.py

"""The poll loop: isolation, last-good-value, and events that mark transitions only.

These three behaviours are why the dashboard can be trusted when something is broken, so
they are asserted directly rather than through HTTP:

* a source that throws must not take the loop, the hub, or its neighbours down;
* a failed poll keeps the previous payload and is marked stale — never zeroed;
* ok→fail and fail→ok each write exactly one event; 500 more failures write nothing;
* maintenance failing is logged, because a silent rollup failure means every long-window
  chart is quietly empty (that bug is what the logging was added for).
"""
from __future__ import annotations

import asyncio
import logging
import sqlite3
import time
from collections.abc import Iterator
from typing import Any

import pytest

from dashboards.poller import Hub, Job, Result
from dashboards.rates import RateEstimator
from dashboards.store import Event, Sample, SampleStore


@pytest.fixture
def store(tmp_path: Any) -> Iterator[SampleStore]:
    instance = SampleStore(tmp_path / "poller.sqlite3")
    try:
        yield instance
    finally:
        instance.close()


def ok_result(value: float = 1.0) -> Result:
    return Result(payload={"value": value},
                  samples=[Sample("host", "gw", "cpu.usage_pct", value, time.time())])


async def failing() -> Result:
    raise ConnectionError("agent is down")


async def test_a_successful_poll_stores_payload_samples_and_health(store) -> None:
    job = Job("host:gw", "system", "gw", 0.5, lambda: asyncio.sleep(0, ok_result(7.0)))
    outcome = await job.run_once(store)
    assert outcome is True, job.state.error
    assert job.state.ok is True and job.state.error is None
    assert job.state.payload == {"value": 7.0}
    assert job.state.successes == 1 and job.state.samples_written == 1
    assert job.state.latency_ms is not None
    assert store.recent_events(limit=5) == [], "a first successful poll is not a recovery"


async def test_a_failing_poll_keeps_the_last_good_payload(store) -> None:
    """The UI shows `stale 12s` against real numbers; it must not show zeros or n/a."""
    job = Job("host:gw", "system", "gw", 0.5, lambda: asyncio.sleep(0, ok_result(7.0)))
    await job.run_once(store)
    job.fn = failing
    assert await job.run_once(store) is False
    assert job.state.ok is False
    assert job.state.payload == {"value": 7.0}, "a dead host must not blank its last values"
    assert job.state.error == "ConnectionError: agent is down"
    assert job.state.as_json()["age_seconds"] is not None


async def test_transitions_write_one_event_each_and_spam_writes_none(store) -> None:
    job = Job("host:gw", "system", "gw", 0.5, lambda: asyncio.sleep(0, ok_result()),
              grace_failures=1)
    await job.run_once(store)
    job.fn = failing
    for _ in range(5):
        await job.run_once(store)
    events = store.recent_events(limit=50)
    down = [event for event in events if event.kind == "unreachable"]
    assert len(down) == 1, "five consecutive failures must not log five events"
    assert down[0].host == "gw", "events carry the display label, not the internal job key"
    assert down[0].severity == "error"

    job.fn = lambda: asyncio.sleep(0, ok_result())
    await job.run_once(store)
    events = store.recent_events(limit=50)
    up = [event for event in events if event.kind == "reachable"]
    assert len(up) == 1
    assert up[0].message == "gw is answering again after 5 failures"


async def test_one_failure_reads_as_a_failure_not_failures(store) -> None:
    job = Job("host:gw", "system", "gw", 0.5, lambda: asyncio.sleep(0, ok_result()))
    await job.run_once(store)
    job.fn = failing
    await job.run_once(store)
    job.fn = lambda: asyncio.sleep(0, ok_result())
    await job.run_once(store)
    up = [event for event in store.recent_events(limit=50) if event.kind == "reachable"]
    assert up[0].message.endswith("after 1 failure")


async def test_grace_allows_a_blip_without_an_event(store) -> None:
    """One failure inside the grace window must not page anybody."""
    job = Job("host:gw", "system", "gw", 0.5, failing, grace_failures=2)
    await job.run_once(store)
    assert store.recent_events(limit=10) == []
    await job.run_once(store)
    events = store.recent_events(limit=10)
    assert [event.kind for event in events] == ["unreachable"]
    assert events[0].severity == "warn", "never answered is a config problem, not an outage"
    assert "never answered" in events[0].message
    for _ in range(4):
        await job.run_once(store)
    assert len(store.recent_events(limit=10)) == 1, "one event per outage, not per poll"


async def test_a_failed_disks_poll_ages_its_cards_instead_of_zeroing_them(store) -> None:
    """Invariant 4 on the phase-09 job: SMART goes quiet, the numbers stay and age.

    The `/disks` page is the one a operator reads *because* something looks wrong, so a
    failed poll must not turn "last read 12 minutes ago" into a page of zeros — and the
    outage must be filed under the `disks` dashboard, which is where that page filters.
    """
    payload = {"name": "driveripper", "smartctl_available": True,
               "disks": [{"device": "nvme0n1", "health": "ok",
                          "smart": {"temp_c": 38.0, "power_on_hours": 8760.0}}],
               "summary": {"disks": 1, "with_smart": 1, "health_failed": 0,
                           "alarm_disks": 0}}

    async def good() -> Result:
        return Result(payload=payload,
                      samples=[Sample("disks", "driveripper", "disk.temp_c@nvme0n1",
                                      38.0, time.time())])

    job = Job("disks:driveripper", "disks", "driveripper", 300.0, good)
    assert job.interval == 300.0, "SMART is not a 10 s signal"
    assert await job.run_once(store) is True
    assert store.query("disks", "driveripper", "disk.temp_c@nvme0n1",
                       start=time.time() - 60), "the series is written under `disks`"

    job.fn = failing
    assert await job.run_once(store) is False
    assert job.state.payload == payload, "the card keeps the last temperature it read"
    assert job.state.ok is False
    assert job.state.as_json()["age_seconds"] is not None
    events = store.recent_events(limit=10)
    assert [event.source for event in events] == ["disks"], (
        "the outage must land where /disks filters its feed, not under the sample source")


async def test_a_slow_poll_never_overlaps_itself(store) -> None:
    calls = 0
    gate = asyncio.Event()

    async def slow() -> Result:
        nonlocal calls
        calls += 1
        await gate.wait()
        return ok_result()

    job = Job("host:gw", "system", "gw", 0.5, slow)
    first = asyncio.create_task(job.run_once(store))
    await asyncio.sleep(0.01)
    assert await job.run_once(store) is False, "an overlapping poll double-counts rates"
    gate.set()
    assert await first is True
    assert calls == 1


async def test_a_counter_that_resets_is_reported_not_silently_negative() -> None:
    """The reset path is what ships: an agent restart must not read as negative throughput."""
    rates = RateEstimator(span=10.0)
    first, _ = rates.rate("cpu.total", 1000.0, now=100.0)
    assert first is None, "one sample cannot produce a rate"
    second, window = rates.rate("cpu.total", 2000.0, now=110.0)
    assert second == pytest.approx(100.0) and window == pytest.approx(10.0)
    after, _ = rates.rate("cpu.total", 5.0, now=120.0)  # agent restarted
    assert after is None, "a reset counter must read n/a, never as negative throughput"


async def test_the_hub_runs_every_job_and_stops_cleanly(store) -> None:
    counts = {"a": 0, "b": 0}

    async def good(key: str) -> Result:
        counts[key] += 1
        return ok_result()

    hub = Hub(store)
    hub.add(Job("host:a", "system", "a", 0.5, lambda: good("a")))
    hub.add(Job("host:b", "system", "b", 0.5, failing))
    hub.add(Job("llm:off", "llm", "off", 0.5, failing, enabled=False))
    hub.start()
    try:
        await asyncio.sleep(1.2)
    finally:
        await hub.stop()
    assert counts["a"] >= 2, "a neighbour that throws must not slow the healthy one"
    states = {state["key"]: state for state in hub.states()}
    assert states["host:a"]["ok"] is True
    assert states["host:b"]["ok"] is False
    assert states["llm:off"]["configured"] is False, "not configured is not the same as broken"
    assert hub.jobs["host:b"].state.failures >= 1


async def test_states_can_be_filtered_to_one_dashboard(store) -> None:
    hub = Hub(store)
    hub.add(Job("host:a", "system", "a", 0.5, lambda: asyncio.sleep(0, ok_result())))
    hub.add(Job("nut", "ups", "NUT server", 0.5, lambda: asyncio.sleep(0, ok_result())))
    assert [s["key"] for s in hub.states("ups")] == ["nut"]


async def test_maintenance_failure_is_logged_not_swallowed(store, monkeypatch,
                                                           caplog: pytest.LogCaptureFixture) -> None:
    """A broken rollup means every chart past the raw window is empty. Retrying quietly
    turned a data-loss bug into nothing at all — this is the regression that added the log."""
    def boom() -> dict[str, int]:
        raise RuntimeError("no such table: sample_rollup")

    monkeypatch.setattr(store, "maintain", boom)
    hub = Hub(store)
    task = asyncio.create_task(hub._maintain_loop(interval=0.02))
    try:
        with caplog.at_level(logging.ERROR, logger="dashboards.poller"):
            await asyncio.sleep(0.15)
    finally:
        task.cancel()
        with pytest.raises(asyncio.CancelledError):
            await task
    assert "store maintenance failed" in caplog.text
    assert "no such table" in caplog.text, "the log line must carry the real reason"


async def test_maintenance_runs_from_the_hub_and_records_its_pass(store) -> None:
    hub = Hub(store)
    calls = []
    store.maintain = lambda: calls.append(1) or {"to_minute": 0, "to_hour": 0,  # type: ignore[method-assign]
                                                 "pruned": 0}
    task = asyncio.create_task(hub._maintain_loop(interval=0.02))
    try:
        await asyncio.sleep(0.1)
    finally:
        task.cancel()
        with pytest.raises(asyncio.CancelledError):
            await task
    assert calls, "the maintenance loop must actually call maintain"


async def test_events_written_by_a_job_land_in_the_store(store) -> None:
    event = Event(1_700_000_000.0, "ups", "cyberpower", "status", "warn", "OL → OB")
    job = Job("nut", "ups", "NUT server", 0.5,
              lambda: asyncio.sleep(0, Result(payload={}, events=[event])))
    await job.run_once(store)
    assert [e.message for e in store.recent_events(limit=5)] == ["OL → OB"]


class RefusingStore(SampleStore):
    """A store that turns good polls away for a while — what SQLite does when it is locked.

    This is not hypothetical: the dev server lost four sources at once this way, and nothing
    in the UI, the health record or the event log said so.
    """

    def __init__(self, path: Any, refusals: int) -> None:
        super().__init__(path)
        self.refusals = refusals

    def add_many(self, samples: Any) -> int:
        if self.refusals > 0:
            self.refusals -= 1
            raise sqlite3.OperationalError("database is locked")
        return super().add_many(samples)


async def test_a_store_that_refuses_a_good_poll_is_reported_not_survived_silently(
        tmp_path: Any, caplog: pytest.LogCaptureFixture) -> None:
    """A source whose history is not landing is broken, and must read that way.

    The upstream answered, so the freshest payload stays on the card; everything else says the
    opposite of healthy — `ok: false`, the real reason in `error`, and `last_ok` frozen so the
    page ages the numbers instead of calling a write failure a live source.
    """
    store = RefusingStore(tmp_path / "locked.sqlite3", refusals=0)
    try:
        job = Job("nut", "ups", "NUT server", 0.5, lambda: asyncio.sleep(0, ok_result(42.0)))
        assert await job.run_once(store) is True, "the first poll writes before the store turns nasty"
        store.refusals = 1
        with caplog.at_level(logging.ERROR, logger="dashboards.poller"):
            assert await job.run_once(store) is False
        assert job.state.ok is False
        assert job.state.error and "database is locked" in job.state.error
        assert job.state.payload == {"value": 42.0}, "a refused write must not blank the card"
        assert job.state.last_ok is not None and job.state.last_attempt is not None
        assert job.state.last_attempt >= job.state.last_ok
        assert "could not record a good poll" in caplog.text
    finally:
        store.close()


async def test_a_refusing_store_keeps_the_loop_polling_and_says_so_once(
        tmp_path: Any, caplog: pytest.LogCaptureFixture) -> None:
    store = RefusingStore(tmp_path / "locked.sqlite3", refusals=2)
    try:
        job = Job("nut", "ups", "NUT server", 0.5, lambda: asyncio.sleep(0, ok_result()))
        with caplog.at_level(logging.ERROR, logger="dashboards.poller"):
            for _ in range(3):
                await job.run_once(store)
        assert job.state.ok is True, "the third poll got through, so the source is ok again"
        assert job.state.last_ok is not None
        kinds = [e.kind for e in store.recent_events(limit=10)]
        assert kinds.count("store-error") == 1, "one line per outage, not one per refused poll"
        assert "recording" in kinds, "the recovery must be as visible as the failure"
        assert "could not record a good poll" in caplog.text
        assert "database is locked" in caplog.text, "the log must carry the real reason"
    finally:
        store.close()


async def test_last_good_payload_survives_a_refused_write(tmp_path: Any) -> None:
    """Invariant 4 with the store, not the upstream, at fault: keep the numbers, age them."""
    store = RefusingStore(tmp_path / "locked.sqlite3", refusals=99)
    try:
        job = Job("nut", "ups", "NUT server", 0.5, lambda: asyncio.sleep(0, ok_result(42.0)))
        await job.run_once(store)
        assert job.state.payload == {"value": 42.0}, "a refused write must not blank the card"
        assert job.state.last_ok is None, "and it must not be called current either"
        assert job.state.error and "the store refused it" in job.state.error
        assert job.state.attempts == 1
    finally:
        store.close()


async def test_an_unexpected_error_cannot_kill_the_poll_loop(
        store, caplog: pytest.LogCaptureFixture) -> None:
    """The task *is* a source's polling; an escaping bug must not retire it in silence.

    Four dev-server sources once stopped advancing `attempts` while `/api/health` went on
    reporting them `ok` — an exception that escaped the loop body, retrieved by nobody. This
    guard is the reason that class of failure can only ever be loud.
    """
    job = Job("host:gw", "system", "gw", 0.5, lambda: asyncio.sleep(0, ok_result()))
    real = job.run_once
    calls = 0

    async def exploding(incoming: SampleStore) -> bool:
        nonlocal calls
        calls += 1
        if calls == 1:
            raise RuntimeError("a bug in the bookkeeping")
        return await real(incoming)

    job.run_once = exploding  # type: ignore[method-assign]
    with caplog.at_level(logging.ERROR, logger="dashboards.poller"):
        job.start(store)
        try:
            await asyncio.sleep(1.4)    # the interval floor is 0.5s, so this is ~3 polls
        finally:
            await job.stop()
    assert calls >= 2, "the loop must still be calling the poll body after the escape"
    assert job.state.attempts >= 1 and job.state.ok is True, "and the next poll must have landed"
    assert "raised outside its own error handling" in caplog.text