Wait for Home Assistant Core to start

If sending the WebSocket message to Home Assistant Core to inform it
about the start of a backup fails, we should reject the backup process
entirely. This prevents taking a backup without Home Assistant Core
being aware of it.

As a side effect, this likely prevents situation where Core does not
learn about the backup progress since we also won't be able to send
WebSocket messages about Job progress updates.

.github/workflows/builder.yml

@@ -320,6 +320,15 @@ jobs:
             exit 1
           fi
 
+      - name: Wait for Home Assistant Core to start
+        run: |
+          echo "Waiting for Home Assistant Core to start"
+          timeout 10m ha supervisor logs -f -n 10000 -b 0 | grep -q "Detect a running Home Assistant instance"
+          if [ "$?" != "0" ]; then
+            echo "Home Assistant Core did not start within 10 minutes"
+            exit 1
+          fi
+
       - name: Create full backup
         id: backup
         run: |

requirements_tests.txt

@@ -8,7 +8,7 @@ pytest-asyncio==0.25.2
 pytest-cov==7.0.0
 pytest-timeout==2.4.0
 pytest==8.4.2
-ruff==0.14.1
+ruff==0.14.2
 time-machine==2.19.0
 types-docker==7.1.0.20251009
 types-pyyaml==6.0.12.20250915

supervisor/api/ingress.py

@@ -253,16 +253,6 @@ class APIIngress(CoreSysAttributes):
             skip_auto_headers={hdrs.CONTENT_TYPE},
         ) as result:
             headers = _response_header(result)
-
-            # Empty body responses (304, 204, HEAD, etc.) should not be streamed,
-            # otherwise aiohttp < 3.9.0 may generate an invalid "0\r\n\r\n" chunk
-            # This also avoids setting content_type for empty responses.
-            if must_be_empty_body(request.method, result.status):
-                return web.Response(
-                    headers=headers,
-                    status=result.status,
-                )
-
             # Avoid parsing content_type in simple cases for better performance
             if maybe_content_type := result.headers.get(hdrs.CONTENT_TYPE):
                 content_type = (maybe_content_type.partition(";"))[0].strip()
@@ -270,7 +260,11 @@ class APIIngress(CoreSysAttributes):
                 content_type = result.content_type
             # Simple request
             if (
-                hdrs.CONTENT_LENGTH in result.headers
+                # empty body responses should not be streamed,
+                # otherwise aiohttp < 3.9.0 may generate
+                # an invalid "0\r\n\r\n" chunk instead of an empty response.
+                must_be_empty_body(request.method, result.status)
+                or hdrs.CONTENT_LENGTH in result.headers
                 and int(result.headers.get(hdrs.CONTENT_LENGTH, 0)) < 4_194_000
             ):
                 # Return Response

supervisor/docker/interface.py

@@ -306,6 +306,8 @@ class DockerInterface(JobGroup, ABC):
         # Our filters have all passed. Time to update the job
         # Only downloading and extracting have progress details. Use that to set extra
         # We'll leave it around on later stages as the total bytes may be useful after that stage
+        # Enforce range to prevent float drift error
+        progress = max(0, min(progress, 100))
         if (
             stage in {PullImageLayerStage.DOWNLOADING, PullImageLayerStage.EXTRACTING}
             and reference.progress_detail
@@ -371,7 +373,7 @@ class DockerInterface(JobGroup, ABC):
 
         # To reduce noise, limit updates to when result has changed by an entire percent or when stage changed
         if stage != install_job.stage or progress >= install_job.progress + 1:
-            install_job.update(stage=stage.status, progress=progress)
+            install_job.update(stage=stage.status, progress=max(0, min(progress, 100)))
 
     @Job(
         name="docker_interface_install",

supervisor/hardware/disk.py

@@ -9,7 +9,12 @@ from typing import Any
 from supervisor.resolution.const import UnhealthyReason
 
 from ..coresys import CoreSys, CoreSysAttributes
-from ..exceptions import DBusError, DBusObjectError, HardwareNotFound
+from ..exceptions import (
+    DBusError,
+    DBusNotConnectedError,
+    DBusObjectError,
+    HardwareNotFound,
+)
 from .const import UdevSubsystem
 from .data import Device
 
@@ -207,6 +212,8 @@ class HwDisk(CoreSysAttributes):
         try:
             block_device = self.sys_dbus.udisks2.get_block_device_by_path(device_path)
             drive = self.sys_dbus.udisks2.get_drive(block_device.drive)
+        except DBusNotConnectedError:
+            return None
         except DBusObjectError:
             _LOGGER.warning(
                 "Unable to find UDisks2 drive for device at %s", device_path.as_posix()

supervisor/homeassistant/module.py

@@ -371,6 +371,12 @@ class HomeAssistant(FileConfiguration, CoreSysAttributes):
                 _LOGGER.error,
             ) from err
 
+        if not resp:
+            raise HomeAssistantBackupError(
+                "Preparing backup of Home Assistant Core failed. No response from HA Core.",
+                _LOGGER.error,
+            )
+
         if resp and not resp.get(ATTR_SUCCESS):
             raise HomeAssistantBackupError(
                 f"Preparing backup of Home Assistant Core failed due to: {resp.get(ATTR_ERROR, {}).get(ATTR_MESSAGE, '')}. Check HA Core logs.",

supervisor/homeassistant/websocket.py

@@ -225,6 +225,10 @@ class HomeAssistantWebSocket(CoreSysAttributes):
         # since it makes a new socket connection and we already have one.
         if not connected and not await self.sys_homeassistant.api.check_api_state():
             # No core access, don't try.
+            _LOGGER.debug(
+                "Home Assistant API is not accessible. Not sending WS message: %s",
+                message,
+            )
             return False
 
         if not self._client:

tests/hardware/test_disk.py

@@ -376,3 +376,14 @@ async def test_try_get_nvme_life_time_missing_percent_used(
         coresys.config.path_supervisor
     )
     assert lifetime is None
+
+
+async def test_try_get_nvme_life_time_dbus_not_connected(coresys: CoreSys):
+    """Test getting lifetime info from an NVMe when DBUS is not connected."""
+    # Set the dbus for udisks2 bus to be None, to make it forcibly disconnected.
+    coresys.dbus.udisks2.dbus = None
+
+    lifetime = await coresys.hardware.disk.get_disk_life_time(
+        coresys.config.path_supervisor
+    )
+    assert lifetime is None