Add more context to Sentry reports early during startup (#5682)

* Initialize machine information before Sentry

* Set user and machine for all reports

Now that we initialize machine earlier we can report user and machine
for all events, even before Supervisor is completely initialized.

Also use the new tag format which is a dictionary.

Note that it seems that with the current Sentry SDK version the
AioHttpIntegration no longer sets the URL as a tag. So sanitation is
no longer reuqired.

* Update pytests

supervisor/bootstrap.py

@@ -18,7 +18,6 @@ from .const import (
     ENV_SUPERVISOR_MACHINE,
     ENV_SUPERVISOR_NAME,
     ENV_SUPERVISOR_SHARE,
-    MACHINE_ID,
     SOCKET_DOCKER,
     LogLevel,
     UpdateChannel,
@@ -81,6 +80,9 @@ async def initialize_coresys() -> CoreSys:
     coresys.bus = Bus(coresys)
     coresys.mounts = await MountManager(coresys).load_config()
 
+    # Set Machine/Host ID
+    await coresys.init_machine()
+
     # diagnostics
     if coresys.config.diagnostics:
         init_sentry(coresys)
@@ -88,10 +90,6 @@ async def initialize_coresys() -> CoreSys:
     # bootstrap config
     initialize_system(coresys)
 
-    # Set Machine/Host ID
-    if MACHINE_ID.exists():
-        coresys.machine_id = MACHINE_ID.read_text(encoding="utf-8").strip()
-
     # Check if ENV is in development mode
     if coresys.dev:
         _LOGGER.warning("Environment variable 'SUPERVISOR_DEV' is set")
@@ -105,16 +103,6 @@ async def initialize_coresys() -> CoreSys:
     # Convert datetime
     logging.Formatter.converter = lambda *args: coresys.now().timetuple()
 
-    # Set machine type
-    if os.environ.get(ENV_SUPERVISOR_MACHINE):
-        coresys.machine = os.environ[ENV_SUPERVISOR_MACHINE]
-    elif os.environ.get(ENV_HOMEASSISTANT_REPOSITORY):
-        coresys.machine = os.environ[ENV_HOMEASSISTANT_REPOSITORY][14:-14]
-        _LOGGER.warning(
-            "Missing SUPERVISOR_MACHINE environment variable. Fallback to deprecated extraction!"
-        )
-    _LOGGER.info("Setting up coresys for machine: %s", coresys.machine)
-
     return coresys
 
 

supervisor/coresys.py

@@ -15,7 +15,13 @@ from typing import TYPE_CHECKING, Any, Self, TypeVar
 import aiohttp
 
 from .config import CoreConfig
-from .const import ENV_SUPERVISOR_DEV, SERVER_SOFTWARE
+from .const import (
+    ENV_HOMEASSISTANT_REPOSITORY,
+    ENV_SUPERVISOR_DEV,
+    ENV_SUPERVISOR_MACHINE,
+    MACHINE_ID,
+    SERVER_SOFTWARE,
+)
 from .utils.dt import UTC, get_time_zone
 
 if TYPE_CHECKING:
@@ -107,6 +113,26 @@ class CoreSys:
         await self.config.read_data()
         return self
 
+    async def init_machine(self):
+        """Initialize machine information."""
+
+        def _load_machine_id() -> str | None:
+            if MACHINE_ID.exists():
+                return MACHINE_ID.read_text(encoding="utf-8").strip()
+            return None
+
+        self.machine_id = await self.run_in_executor(_load_machine_id)
+
+        # Set machine type
+        if os.environ.get(ENV_SUPERVISOR_MACHINE):
+            self.machine = os.environ[ENV_SUPERVISOR_MACHINE]
+        elif os.environ.get(ENV_HOMEASSISTANT_REPOSITORY):
+            self.machine = os.environ[ENV_HOMEASSISTANT_REPOSITORY][14:-14]
+            _LOGGER.warning(
+                "Missing SUPERVISOR_MACHINE environment variable. Fallback to deprecated extraction!"
+            )
+        _LOGGER.info("Setting up coresys for machine: %s", self.machine)
+
     @property
     def dev(self) -> bool:
         """Return True if we run dev mode."""

supervisor/misc/filter.py

@@ -35,6 +35,12 @@ def filter_data(coresys: CoreSys, event: dict, hint: dict) -> dict:
         return None
 
     event.setdefault("extra", {}).update({"os.environ": dict(os.environ)})
+    event.setdefault("user", {}).update({"id": coresys.machine_id})
+    event.setdefault("tags", {}).update(
+        {
+            "machine": coresys.machine,
+        }
+    )
 
     # Not full startup - missing information
     if coresys.core.state in (CoreState.INITIALIZE, CoreState.SETUP):
@@ -47,7 +53,6 @@ def filter_data(coresys: CoreSys, event: dict, hint: dict) -> dict:
     ]
 
     # Update information
-    event.setdefault("user", {}).update({"id": coresys.machine_id})
     event.setdefault("contexts", {}).update(
         {
             "supervisor": {
@@ -92,19 +97,12 @@ def filter_data(coresys: CoreSys, event: dict, hint: dict) -> dict:
         {plugin.slug: plugin.version for plugin in coresys.plugins.all_plugins}
     )
 
-    event.setdefault("tags", []).extend(
-        [
-            ["installation_type", "os" if coresys.os.available else "supervised"],
-            ["machine", coresys.machine],
-        ],
+    event["tags"].update(
+        {
+            "installation_type": "os" if coresys.os.available else "supervised",
+        }
     )
 
-    # Sanitize event
-    for i, tag in enumerate(event.get("tags", [])):
-        key, value = tag
-        if key == "url":
-            event["tags"][i] = [key, sanitize_url(value)]
-
     if event.get("request"):
         if event["request"].get("url"):
             event["request"]["url"] = sanitize_url(event["request"]["url"])

tests/misc/test_filter_data.py

@@ -72,7 +72,7 @@ def test_defaults(coresys):
     with patch("shutil.disk_usage", return_value=(42, 42, 2 * (1024.0**3))):
         filtered = filter_data(coresys, SAMPLE_EVENT, {})
 
-    assert ["installation_type", "supervised"] in filtered["tags"]
+    assert filtered["tags"]["installation_type"] == "supervised"
     assert filtered["contexts"]["host"]["arch"] == "amd64"
     assert filtered["contexts"]["host"]["machine"] == "qemux86-64"
     assert filtered["contexts"]["versions"]["supervisor"] == AwesomeVersion(
@@ -84,7 +84,6 @@ def test_defaults(coresys):
 def test_sanitize(coresys):
     """Test event sanitation."""
     event = {
-        "tags": [["url", "https://mydomain.com"]],
         "request": {
             "url": "https://mydomain.com",
             "headers": [
@@ -101,8 +100,6 @@ def test_sanitize(coresys):
     with patch("shutil.disk_usage", return_value=(42, 42, 2 * (1024.0**3))):
         filtered = filter_data(coresys, event, {})
 
-    assert ["url", "https://example.com"] in filtered["tags"]
-
     assert filtered["request"]["url"] == "https://example.com"
 
     assert ["Host", "example.com"] in filtered["request"]["headers"]