fix(norms): templates grade by the cutoffs when handed no grade; the parts test reads inside transactions (#5742)

CI 9128: the host vitals and monitor templates required the new grades, so
callers (and tests) that render them without one failed; they now fall back
to status_class. The new integration test read outside begin(), which
autobegan the session.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

plugins/host_agent/templates/_host_vitals.html

@@ -5,6 +5,7 @@
    reports (the Agent panel below then shows provisioning). #}
 {% from "_macros.html" import status_badge, stat_tile %}
 {% macro pct(v) %}{{ '%.0f%%'|format(v) if v is not none else '—' }}{% endmacro %}
+{% set grades = grades or {} %}
 {% if reporting %}
 <section class="card host-health" aria-label="Host health">
   <div class="host-health-head">
@@ -22,10 +23,10 @@
     </span>
   </div>
   <div class="stat-strip host-health-tiles" style="margin-bottom:0;">
-    {{ stat_tile("CPU", pct(cpu), state=grades.cpu[0], hint=sparks.cpu|safe) }}
-    {{ stat_tile("Memory", pct(mem), state=grades.mem[0], hint=sparks.mem|safe) }}
-    {{ stat_tile("Disk /", pct(disk_root), state=grades.disk[0], hint=sparks.disk|safe) }}
-    {{ stat_tile("Load per core", pct(load_per_core), state=grades.load[0], hint=sparks.load|safe) }}
+    {{ stat_tile("CPU", pct(cpu), state=(grades.cpu or [status_class(cpu, 'cpu')])[0], hint=sparks.cpu|safe) }}
+    {{ stat_tile("Memory", pct(mem), state=(grades.mem or [status_class(mem, 'mem')])[0], hint=sparks.mem|safe) }}
+    {{ stat_tile("Disk /", pct(disk_root), state=(grades.disk or [status_class(disk_root, 'disk')])[0], hint=sparks.disk|safe) }}
+    {{ stat_tile("Load per core", pct(load_per_core), state=(grades.load or [status_class(load_per_core, 'load')])[0], hint=sparks.load|safe) }}
   </div>
   {% if usual and not stale %}
   <p class="field-hint" style="margin-top:0.6rem;">{{ usual | join("; ") | capitalize }}.</p>

steward/templates/monitors/rows.html

@@ -51,7 +51,7 @@
     {% if d.state in ('down', 'paused', 'pending') %}{{ status_badge(d.state) }}
     {% elif latest.response_ms is not none %}
       {% if d.spark %}<span class="mon-spark" title="Response time over the last 30 checks">{{ d.spark|safe }}</span>{% endif %}
-      <span class="tone-{{ d.latency }}"{% if d.latency_usual %} title="{{ d.latency_usual|capitalize }}"{% endif %}>{{ "%.0f"|format(latest.response_ms) }} ms</span>
+      <span class="tone-{{ d.latency or status_class(latest.response_ms, 'latency') }}"{% if d.latency_usual %} title="{{ d.latency_usual|capitalize }}"{% endif %}>{{ "%.0f"|format(latest.response_ms) }} ms</span>
     {% elif m.type == 'dns' and latest.resolved_ip %}<span class="tone-ok">{{ latest.resolved_ip }}</span>
     {% else %}{{ status_badge('up') }}{% endif %}
   </div>

steward/templates/monitors/widget.html

@@ -26,10 +26,11 @@
       {{- status_dot(d.state) -}}
       {% if session.user_id %}<a href="/monitors/{{ m.id }}">{{ m.name }}</a>{% else %}{{ m.name }}{% endif %}
       {%- if d.state == 'down' %}<span class="w-sub tone-crit">went down {{ d.down_since|ago if d.down_since else '' }}{% if d.down_reason %} · {{ d.down_reason }}{% endif %}</span>
-      {%- elif d.state == 'degraded' %}<span class="w-sub tone-warn">{% if d.latency == 'crit' %}slow replies{% else %}checks failing now and then{% endif %}</span>
+      {%- elif d.state == 'degraded' %}<span class="w-sub tone-warn">{% if (d.latency or status_class(latest.response_ms, 'latency') if latest else none) == 'crit' %}slow replies{% else %}checks failing now and then{% endif %}</span>
       {%- elif d.state == 'pending' %}<span class="w-sub">not checked yet</span>{% endif %}
     </td>
-    <td class="{{ 'tone-' ~ d.latency if d.latency in ('warn', 'crit') }}"{% if d.latency_usual %} title="{{ d.latency_usual|capitalize }}"{% endif %}>
+    {% set lat = (d.latency or status_class(latest.response_ms, 'latency')) if latest and latest.is_up and latest.response_ms is not none else none %}
+    <td class="{{ 'tone-' ~ lat if lat in ('warn', 'crit') }}"{% if d.latency_usual %} title="{{ d.latency_usual|capitalize }}"{% endif %}>
       {%- if latest and latest.is_up and latest.response_ms is not none %}{{ "%.0f"|format(latest.response_ms) }} ms{% else %}—{% endif -%}
     </td>
     {% for k in ("24h", "7d", "30d") %}{% set u = d.uptime.get(k) %}

tests/integration/test_anomalies.py

@@ -360,15 +360,19 @@ def test_a_metric_learned_without_its_parts_ignores_them(app, src):
                 s.add_all(rows)
             async with s.begin():
                 await rebuild_norms(s, NOW)
-            learned = await norms_at(s, [(src, host, METRIC), (src, part, METRIC)], NOW.replace(hour=14))
+            async with s.begin():
+                learned = await norms_at(s, [(src, host, METRIC), (src, part, METRIC)],
+                                         NOW.replace(hour=14))
             async with s.begin():
                 await _report(s, src, host, at, 90.0)
                 await _report(s, src, part, at, 90.0)
-            levels = await levels_now(s, at)
+            async with s.begin():
+                levels = await levels_now(s, at)
             async with s.begin():
                 await detect(s, at)
-            opened = list((await s.execute(select(MetricAnomaly.resource_name).where(
-                MetricAnomaly.source_module == src))).scalars())
+            async with s.begin():
+                opened = list((await s.execute(select(MetricAnomaly.resource_name).where(
+                    MetricAnomaly.source_module == src))).scalars())
             return learned, levels, opened
 
     learned, levels, opened = asyncio.run(_go())