feat(homeassistant): it shows system problems and installs updates on a weekly slot, and leaves device readings alone (#6748, #6749, #6747)

The connection's job is keeping Home Assistant running. Each poll now reads open
repairs, Supervisor health, disk, recovery mode and add-ons that should be running
over its websocket, for the page, the overview and the problems_open, disk_free_pct
and update_failed metrics. That needs a websocket client, so websockets joins the
dependencies.

A weekly slot, set on the page or through MCP and off until then, installs add-ons,
then Core, then the OS, each after its own confirmed backup, ending at the first
failure. The slot arithmetic moves from the host agent's cadence to steward.core.weekly
so both share it.

The device scrape goes: the page's Metrics section and a target's token_from from
4c9ed02 are removed, and a migration deletes targets using it, as they'd otherwise ask
with no token. A refused token still stops a scrape.

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

tests/plugins/homeassistant/test_service.py

@@ -4,7 +4,7 @@ from datetime import datetime, timedelta, timezone
 from types import SimpleNamespace as NS
 
 import plugins.homeassistant as homeassistant_plugin
-from plugins.homeassistant import mcp_tools, scheduler, service as s
+from plugins.homeassistant import mcp_tools, schedule, scheduler, service as s
 from steward.models.users import UserRole
 from tests._templates import render_plugin_template
 from tests.test_style_tokens import hex_colours
@@ -86,7 +86,7 @@ def _connection(**over):
                  "verify_tls": True, "signed_in_by": "someone", "signed_in_at": NOW,
                  "refused": False, "last_error": None, "last_polled_at": NOW, "last_ok_at": NOW,
                  "backup": {"known": True, "last_ok": NOW.isoformat(), "last_result": "completed"},
-                 "updates": [], **over})
+                 "updates": [], "health": {}, **over})
 
 
 def test_the_overview_names_what_needs_a_look_and_no_more():
@@ -101,6 +101,50 @@ def test_the_overview_names_what_needs_a_look_and_no_more():
     failed = NS(state="failed", title="Core", message="No new backup.", finished_at=NOW)
     [f] = mcp_tools.overview_findings(_connection(), failed, NOW)
     assert f["kind"] == "home_assistant_update_failed" and "No new backup." in f["summary"]
+    two = {"problems": [{"title": "Disk trouble"}, {"title": "Another"}]}
+    [f] = mcp_tools.overview_findings(_connection(health=two), None, NOW)
+    assert f["kind"] == "home_assistant_problem" and f["summary"] == "Disk trouble (and 1 more)"
+    plan = NS(stopped="Core: it failed.", last_slot_at=None)
+    [f] = mcp_tools.overview_findings(_connection(), None, NOW, plan)
+    assert f["kind"] == "home_assistant_schedule_stopped" and "Core: it failed." in f["summary"]
+
+
+def test_system_problems_are_read_from_what_home_assistant_answers():
+    healthy = s.read_health({"config": {"recovery_mode": False}, "issues": [], "resolution": {"unhealthy": [], "unsupported": []},
+                             "host": {"disk_free": 99.8, "disk_total": 116.5}, "addons": [{"slug": "a", "state": "started"}]})
+    assert healthy["problems"] == [] and healthy["repairs_read"] and healthy["supervised"]
+    assert healthy["disk_free_pct"] == 85.7
+    bad = s.read_health({
+        "config": {"recovery_mode": True},
+        "issues": [{"domain": "hassio", "translation_key": "issue_system_reboot_required", "severity": "warning"},
+                   {"domain": "demo", "translation_key": "named", "translation_placeholders": {"who": "X"}, "severity": "critical"},
+                   {"domain": "demo", "translation_key": "gone", "ignored": True},
+                   {"domain": "demo", "translation_key": "old", "dismissed_version": "1.0"}],
+        "titles": {"component.demo.issues.named.title": "{who} broke"},
+        "resolution": {"unhealthy": ["docker_gateway"], "unsupported": []},
+        "host": {"disk_free": 5.0, "disk_total": 100.0},
+        "addons": [{"slug": "should", "name": "Should run", "state": "stopped"}, {"slug": "once", "state": "stopped"}],
+        "boots": {"should": "auto", "once": "manual"}})
+    assert [p["kind"] for p in bad["problems"]] == ["recovery", "repair", "repair", "supervisor", "disk", "addon"]
+    titles = [p["title"] for p in bad["problems"]]
+    assert titles[1] == "System reboot required" and titles[2] == "X broke" and "docker gateway" in titles[3]
+    assert "5 GB of 100 GB" in titles[4] and "Should run" in titles[5]
+
+
+def test_what_home_assistant_wouldnt_answer_is_not_read_as_fine():
+    found = s.read_health({"config": {}, "issues": None, "resolution": None, "host": None, "addons": None})
+    assert not found["repairs_read"] and not found["supervised"] and found["disk_free_pct"] is None
+
+
+def test_a_schedule_installs_add_ons_then_core_then_the_os():
+    def update(entity_id, kind, in_progress=False):
+        return {"entity_id": entity_id, "kind": kind, "in_progress": in_progress}
+    waiting = [update("os", "os"), update("core", "core"), update("addon", "addon")]
+    assert schedule.next_to_install(waiting, set())["entity_id"] == "addon"
+    assert schedule.next_to_install(waiting, {"addon"})["entity_id"] == "core"
+    assert schedule.next_to_install(waiting, {"addon", "core"})["entity_id"] == "os"
+    assert schedule.next_to_install(waiting, {"addon", "core", "os"}) is None
+    assert schedule.next_to_install([update("addon", "addon", in_progress=True)], set()) is None
 
 
 def test_reading_is_open_to_a_viewer_and_installing_is_for_an_admin():
@@ -109,7 +153,8 @@ def test_reading_is_open_to_a_viewer_and_installing_is_for_an_admin():
     def register(fn, *, role, read_only):     # both required, as on the server
         seen[fn.__name__] = (role, read_only)
     homeassistant_plugin.register_mcp_tools(register)
-    assert seen == {"get_home_assistant": (UserRole.viewer, True), "update_home_assistant": (UserRole.admin, False)}
+    assert seen == {"get_home_assistant": (UserRole.viewer, True), "update_home_assistant": (UserRole.admin, False),
+                    "set_home_assistant_schedule": (UserRole.admin, False)}
 
 
 def _page_values(**over):
@@ -119,7 +164,9 @@ def _page_values(**over):
     return {"connection": _connection(), "reach": "up", "says": "Answering.", "installable": [update], "others": [],
             "current": [], "runs": [], "under_way": None, "backup": {}, "backup_tone": "ok",
             "backup_says": "Backups are up to date.", "backup_at": NOW, "backup_next": None, "backup_age": 1.0,
-            "stale_hours": 26, **over}
+            "stale_hours": 26, "health": {"repairs_read": True, "supervised": True}, "problems": [], "plan": None,
+            "plan_when": None, "plan_next": None, "weekdays": ("Monday", "Tuesday", "Wednesday", "Thursday", "Friday",
+                                                               "Saturday", "Sunday"), "default_zone": "UTC", **over}
 
 
 async def _render(app, role, **values):
@@ -151,6 +198,30 @@ async def test_install_is_held_back_while_it_couldnt_work_and_the_page_says_why(
     assert "isn't an administrator" in status_only
 
 
+async def test_the_page_shows_system_problems_and_what_it_couldnt_read(app):
+    clear = await _render(app, "viewer", **_page_values())
+    assert "No repairs open" in clear
+    problem = {"kind": "repair", "severity": "warning", "title": "System reboot required"}
+    shown = await _render(app, "viewer", **_page_values(problems=[problem]))
+    assert "System reboot required" in shown and "Settings → System → Repairs" in shown
+    hidden = await _render(app, "viewer", **_page_values(health={"repairs_read": False, "supervised": False}))
+    assert "wouldn't share its repairs" in hidden
+    bare = await _render(app, "viewer", **_page_values(health={"repairs_read": True, "supervised": False}))
+    assert "No Supervisor answered" in bare
+
+
+async def test_an_admin_sets_the_schedule_warned_of_what_it_does_and_a_viewer_only_reads_it(app):
+    off = await _render(app, "admin", **_page_values())
+    assert 'action="/plugins/homeassistant/schedule"' in off and "Off. Updates are only installed" in off
+    assert "with nobody there to watch" in off
+    assert '<option value="5" selected>Saturday</option>' in off and 'value="05:00"' in off
+    plan = NS(enabled=True, weekday=5, at_time="05:00", time_zone="Europe/London", stopped="Core: it failed.")
+    values = _page_values(plan=plan, plan_when="Saturdays at 05:00 (Europe/London), up to 4 h", plan_next=NOW)
+    on = await _render(app, "viewer", **values)
+    assert "Saturdays at 05:00" in on and "Core: it failed." in on
+    assert 'action="/plugins/homeassistant/schedule"' not in on
+
+
 async def test_an_unconnected_page_offers_to_connect(app):
     page = await _render(app, "admin", connection=None)
     assert 'href="/plugins/homeassistant/connection"' in page

tests/plugins/prometheus/test_service.py

@@ -141,10 +141,6 @@ def test_the_form_checks_what_it_stores_and_keeps_a_blank_token():
     assert again["token"] == "enc:v1:kept" and "enabled" not in again
     cleared = parse_form(_Form(name="Archives", url="http://a/metrics", clear_token="on"), existing)
     assert cleared["token"] == ""
-    lent = PrometheusTarget(name="Archives", url="http://a/metrics", token="", token_from="a_plugin")
-    assert parse_form(_Form(name="Archives", url="http://a/metrics"), lent)["token_from"] == "a_plugin"
-    typed = parse_form(_Form(name="Archives", url="http://a/metrics", token="s3cret"), lent)
-    assert typed["token_from"] is None and typed["token"]
     for bad in (dict(name="", url="http://a"), dict(name="a:b", url="http://a"),
                 dict(name="a", url="ftp://a"), dict(name="a", url="http://a", interval_seconds="5"),
                 dict(name="a", url="http://a", series_cap="0")):