bvandeusenandClaude Opus 5.5
4fbe93bdf7
The Alerts page's active list now says each alert in one line around the resource it watches, e.g. "cpu_pct on web-1 is 94% (> 90%), firing for 12m", with web-1 linking to its host, monitor or container page (or the plugin's page for Traefik, UniFi and SNMP). Values take their unit from the metric name through formatting.metric_value. A silent series reads "No cpu_pct from web-1 for 20m"; an up/down metric says the word. Actions follow the routes' roles: acknowledge, edit, toggle and New rule for operators and admins, Delete for admins only, nothing for viewers. The sidebar shows how many alerts need attention beside Alerts, from an async context processor that skips htmx fragments, logged-out pages and the database-down page, and renders without the count if the query fails. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
135 lines
6.4 KiB
Python
135 lines
6.4 KiB
Python
"""The Alerts page says each alert as a sentence around a link to what it
|
|
watches, offers only the actions the viewer's role can take, and the sidebar
|
|
counts what needs attention (#5124)."""
|
|
from datetime import datetime, timedelta, timezone
|
|
from types import SimpleNamespace
|
|
|
|
import pytest
|
|
from quart import render_template, session
|
|
|
|
from steward.alerts import service as alert_svc
|
|
from steward.alerts.service import alert_sentence
|
|
from steward.models.alerts import AlertOperator, AlertStateEnum
|
|
from tests.test_style_tokens import hex_colours
|
|
|
|
|
|
def _rule(metric="cpu_pct", op=AlertOperator.gt, threshold=90.0, resource="web-1", rid="r1"):
|
|
return SimpleNamespace(id=rid, name=f"{resource} {metric}", source_module="host_agent",
|
|
resource_name=resource, metric_name=metric, operator=op,
|
|
threshold=threshold, enabled=True,
|
|
consecutive_failures_required=1, consecutive_ok_required=1)
|
|
|
|
|
|
def _said(says):
|
|
return says["lead"] + says["resource"] + says["tail"]
|
|
|
|
|
|
@pytest.mark.parametrize("op, threshold, value, reads", [
|
|
(AlertOperator.gt, 90.0, 94.0, "cpu_pct on web-1 is 94% (> 90%), firing for 12m"),
|
|
(AlertOperator.gte, 90.0, 90.0, "cpu_pct on web-1 is 90% (>= 90%), firing for 12m"),
|
|
(AlertOperator.lt, 5.0, 2.5, "cpu_pct on web-1 is 2.5% (< 5.0%), firing for 12m"),
|
|
(AlertOperator.lte, 5.0, 5.0, "cpu_pct on web-1 is 5.0% (<= 5.0%), firing for 12m"),
|
|
(AlertOperator.eq, 100.0, 100.0, "cpu_pct on web-1 is 100% (== 100%), firing for 12m"),
|
|
(AlertOperator.ne, 0.0, 3.0, "cpu_pct on web-1 is 3.0% (!= 0.0%), firing for 12m"),
|
|
])
|
|
def test_a_firing_alert_reads_as_one_sentence(op, threshold, value, reads):
|
|
now = datetime.now(timezone.utc)
|
|
says = alert_sentence(_rule(op=op, threshold=threshold), "firing", value,
|
|
now - timedelta(minutes=12), now)
|
|
assert says["resource"] == "web-1"
|
|
assert _said(says) == reads
|
|
|
|
|
|
def test_units_follow_the_metric_name():
|
|
now = datetime.now(timezone.utc)
|
|
says = alert_sentence(_rule("response_ms", threshold=200.0, resource="api"), "pending", 431.0,
|
|
now - timedelta(seconds=45), now)
|
|
assert _said(says) == "response_ms on api is 431 ms (> 200 ms), pending for 45s"
|
|
|
|
|
|
def test_an_up_or_down_metric_says_the_word():
|
|
now = datetime.now(timezone.utc)
|
|
says = alert_sentence(_rule("is_up", AlertOperator.lt, 1.0, "nas"), "acknowledged", 0.0,
|
|
now - timedelta(hours=2, minutes=5), now)
|
|
assert _said(says) == "nas is down, acknowledged for 2h 5m"
|
|
|
|
|
|
def test_no_reading_yet_is_said_not_shown_as_a_number():
|
|
now = datetime.now(timezone.utc)
|
|
says = alert_sentence(_rule(), "firing", None, now - timedelta(minutes=3), now)
|
|
assert _said(says) == "cpu_pct on web-1 has no reading yet (> 90%), firing for 3m"
|
|
|
|
|
|
def test_a_silent_series_says_how_long_it_has_been_quiet():
|
|
now = datetime.now(timezone.utc)
|
|
says = alert_sentence(_rule(), "no_data", 50.0, now - timedelta(minutes=20), now)
|
|
assert _said(says) == "No cpu_pct from web-1 for 20m"
|
|
|
|
|
|
def _row(rule, label, url, state=AlertStateEnum.firing):
|
|
now = datetime.now(timezone.utc)
|
|
st = SimpleNamespace(state=state, no_data_since=None, last_sample_at=None,
|
|
last_transition_at=now - timedelta(minutes=12))
|
|
return {"state": st, "rule": rule, "label": label, "url": url,
|
|
"says": alert_sentence(rule, label, 94.0, st.last_transition_at, now)}
|
|
|
|
|
|
async def test_an_operator_sees_the_sentence_linked_and_can_acknowledge(app):
|
|
rule = _rule()
|
|
async with app.test_request_context("/alerts/"):
|
|
session["user_role"] = "operator"
|
|
page = await render_template("alerts/list.html", active=[_row(rule, "firing", "/hosts/h1")],
|
|
rules=[rule], operators=[])
|
|
|
|
assert 'cpu_pct on <a href="/hosts/h1">web-1</a> is 94% (> 90%), firing for 12m' in page
|
|
assert '<span class="badge badge-crit">Firing</span>' in page
|
|
assert 'action="/alerts/r1/acknowledge"' in page
|
|
assert 'href="/alerts/rules/new"' in page and 'href="/alerts/rules/r1/edit"' in page
|
|
assert 'action="/alerts/rules/r1/delete"' not in page # deleting a rule is the admin's
|
|
main = page[page.index("<main"):] # the sidebar brand mark carries its own colours
|
|
assert "onclick" not in main and not hex_colours(main), hex_colours(main)
|
|
|
|
|
|
async def test_a_viewer_reads_but_gets_no_actions(app):
|
|
rule = _rule()
|
|
async with app.test_request_context("/alerts/"):
|
|
session["user_role"] = "viewer"
|
|
page = await render_template("alerts/list.html", active=[_row(rule, "firing", None)],
|
|
rules=[rule], operators=[])
|
|
|
|
assert "cpu_pct on <strong>web-1</strong>" in page # nothing to link to: no bare href
|
|
for action in ("/acknowledge", "/alerts/rules/new", "/edit", "/toggle", "/delete"):
|
|
assert action not in page, action
|
|
assert 'href="/alerts/history?rule=r1"' in page
|
|
|
|
|
|
async def test_an_admin_can_delete_a_rule(app):
|
|
rule = _rule()
|
|
async with app.test_request_context("/alerts/"):
|
|
session["user_role"] = "admin"
|
|
page = await render_template("alerts/list.html", active=[], rules=[rule], operators=[])
|
|
assert 'action="/alerts/rules/r1/delete"' in page
|
|
assert 'data-confirm="Delete rule web-1 cpu_pct?"' in page
|
|
|
|
|
|
async def _active_count(_db):
|
|
return 3
|
|
|
|
|
|
async def test_the_sidebar_counts_alerts_needing_attention(app, monkeypatch):
|
|
monkeypatch.setattr(alert_svc, "active_count", _active_count)
|
|
async with app.test_request_context("/alerts/"):
|
|
session["user_id"], session["user_role"] = "u1", "viewer"
|
|
page = await render_template("alerts/list.html", active=[], rules=[], operators=[])
|
|
assert '<span class="nav-count" title="3 alerts need attention">3</span>' in page
|
|
|
|
|
|
async def test_no_count_for_a_fragment_or_a_logged_out_page(app, monkeypatch):
|
|
monkeypatch.setattr(alert_svc, "active_count", _active_count)
|
|
async with app.test_request_context("/alerts/", headers={"HX-Request": "true"}):
|
|
session["user_id"], session["user_role"] = "u1", "viewer"
|
|
fragment = await render_template("alerts/list.html", active=[], rules=[], operators=[])
|
|
async with app.test_request_context("/alerts/"):
|
|
logged_out = await render_template("alerts/list.html", active=[], rules=[], operators=[])
|
|
assert "nav-count" not in fragment and "nav-count" not in logged_out
|