Instrument _check_sigs_and_hash_and_fetch to trace time spent in child concurrent calls

changelog.d/13588.misc

@@ -0,0 +1 @@
+Instrument `_check_sigs_and_hash_and_fetch` to trace time spent in child concurrent calls for understandable traces in Jaeger.

synapse/crypto/event_signing.py

@@ -28,13 +28,15 @@
 from synapse.api.room_versions import RoomVersion
 from synapse.events import EventBase
 from synapse.events.utils import prune_event, prune_event_dict
+from synapse.logging.tracing import trace
 from synapse.types import JsonDict
 
 logger = logging.getLogger(__name__)
 
 Hasher = Callable[[bytes], "hashlib._Hash"]
 
 
+@trace
 def check_event_content_hash(
     event: EventBase, hash_algorithm: Hasher = hashlib.sha256
 ) -> bool:

synapse/events/spamcheck.py

@@ -32,6 +32,7 @@
 
 import synapse
 from synapse.api.errors import Codes
+from synapse.logging.tracing import trace
 from synapse.rest.media.v1._base import FileInfo
 from synapse.rest.media.v1.media_storage import ReadableFileWrapper
 from synapse.spam_checker_api import RegistrationBehaviour
@@ -378,6 +379,7 @@ def register_callbacks(
         if check_media_file_for_spam is not None:
             self._check_media_file_for_spam_callbacks.append(check_media_file_for_spam)
 
+    @trace
     async def check_event_for_spam(
         self, event: "synapse.events.EventBase"
     ) -> Union[Tuple[Codes, JsonDict], str]:

synapse/federation/federation_base.py

@@ -23,6 +23,7 @@
 from synapse.events import EventBase, make_event_from_dict
 from synapse.events.utils import prune_event, validate_canonicaljson
 from synapse.http.servlet import assert_params_in_dict
+from synapse.logging.tracing import log_kv, trace
 from synapse.types import JsonDict, get_domain_from_id
 
 if TYPE_CHECKING:
@@ -55,6 +56,7 @@ def __init__(self, hs: "HomeServer"):
         self._clock = hs.get_clock()
         self._storage_controllers = hs.get_storage_controllers()
 
+    @trace
     async def _check_sigs_and_hash(
         self, room_version: RoomVersion, pdu: EventBase
     ) -> EventBase:
@@ -97,17 +99,36 @@ async def _check_sigs_and_hash(
                     "Event %s seems to have been redacted; using our redacted copy",
                     pdu.event_id,
                 )
+                log_kv(
+                    {
+                        "message": "Event seems to have been redacted; using our redacted copy",
+                        "event_id": pdu.event_id,
+                    }
+                )
             else:
                 logger.warning(
                     "Event %s content has been tampered, redacting",
                     pdu.event_id,
                 )
+                log_kv(
+                    {
+                        "message": "Event content has been tampered, redacting",
+                        "event_id": pdu.event_id,
+                    }
+                )
             return redacted_event
 
         spam_check = await self.spam_checker.check_event_for_spam(pdu)
 
         if spam_check != self.spam_checker.NOT_SPAM:
             logger.warning("Event contains spam, soft-failing %s", pdu.event_id)
+            log_kv(
+                {
+                    "message": "Event contains spam, redacting (to save disk space) "
+                    "as well as soft-failing (to stop using the event in prev_events)",
+                    "event_id": pdu.event_id,
+                }
+            )
             # we redact (to save disk space) as well as soft-failing (to stop
             # using the event in prev_events).
             redacted_event = prune_event(pdu)
@@ -117,6 +138,7 @@ async def _check_sigs_and_hash(
         return pdu
 
 
+@trace
 async def _check_sigs_on_pdu(
     keyring: Keyring, room_version: RoomVersion, pdu: EventBase
 ) -> None:

synapse/federation/federation_client.py

@@ -61,7 +61,7 @@
 )
 from synapse.federation.transport.client import SendJoinResponse
 from synapse.http.types import QueryParams
-from synapse.logging.opentracing import SynapseTags, set_tag, tag_args, trace
+from synapse.logging.opentracing import SynapseTags, log_kv, set_tag, tag_args, trace
 from synapse.types import JsonDict, UserID, get_domain_from_id
 from synapse.util.async_helpers import concurrently_execute
 from synapse.util.caches.expiringcache import ExpiringCache
@@ -587,6 +587,10 @@ async def _check_sigs_and_hash_and_fetch(
         Returns:
             A list of PDUs that have valid signatures and hashes.
         """
+        set_tag(
+            SynapseTags.RESULT_PREFIX + "pdus.length",
+            str(len(pdus)),
+        )
 
         # We limit how many PDUs we check at once, as if we try to do hundreds
         # of thousands of PDUs at once we see large memory spikes.
@@ -607,6 +611,8 @@ async def _execute(pdu: EventBase) -> None:
 
         return valid_pdus
 
+    @trace
+    @tag_args
     async def _check_sigs_and_hash_and_fetch_one(
         self,
         pdu: EventBase,
@@ -643,6 +649,14 @@ async def _check_sigs_and_hash_and_fetch_one(
                 pdu.event_id,
                 e,
             )
+            log_kv(
+                {
+                    "message": "Signature on retrieved event was invalid. "
+                    "Checking local store/orgin server",
+                    "event_id": pdu.event_id,
+                    "InvalidEventSignatureError": e,
+                }
+            )
 
         # Check local db.
         res = await self.store.get_event(

synapse/util/ratelimitutils.py

@@ -168,6 +168,11 @@ def should_reject(self) -> bool:
         (either sleeping or in the ready queue).
         """
         queue_size = len(self.ready_request_queue) + len(self.sleeping_requests)
+        logger.info(
+            "should_reject queue_size=%s self.reject_limit=%s",
+            queue_size,
+            self.reject_limit,
+        )
         return queue_size > self.reject_limit
 
     def should_sleep(self) -> bool: