diff --git a/synapse/handlers/account_data.py b/synapse/handlers/account_data.py
index 0478448b47..fc21d58001 100644
--- a/synapse/handlers/account_data.py
+++ b/synapse/handlers/account_data.py
@@ -225,7 +225,7 @@ class AccountDataEventSource(EventSource[int, JsonDict]):
self,
user: UserID,
from_key: int,
- limit: Optional[int],
+ limit: int,
room_ids: Collection[str],
is_guest: bool,
explicit_room_id: Optional[str] = None,
diff --git a/synapse/handlers/appservice.py b/synapse/handlers/appservice.py
index 203b62e015..66f5b8d108 100644
--- a/synapse/handlers/appservice.py
+++ b/synapse/handlers/appservice.py
@@ -109,10 +109,13 @@ class ApplicationServicesHandler:
last_token = await self.store.get_appservice_last_pos()
(
upper_bound,
- events,
event_to_received_ts,
- ) = await self.store.get_all_new_events_stream(
- last_token, self.current_max, limit=100, get_prev_content=True
+ ) = await self.store.get_all_new_event_ids_stream(
+ last_token, self.current_max, limit=100
+ )
+
+ events = await self.store.get_events_as_list(
+ event_to_received_ts.keys(), get_prev_content=True
)
events_by_room: Dict[str, List[EventBase]] = {}
diff --git a/synapse/handlers/directory.py b/synapse/handlers/directory.py
index 7127d5aefc..d52ebada6b 100644
--- a/synapse/handlers/directory.py
+++ b/synapse/handlers/directory.py
@@ -16,6 +16,8 @@ import logging
import string
from typing import TYPE_CHECKING, Iterable, List, Optional
+from typing_extensions import Literal
+
from synapse.api.constants import MAX_ALIAS_LENGTH, EventTypes
from synapse.api.errors import (
AuthError,
@@ -429,7 +431,10 @@ class DirectoryHandler:
return await self.auth.check_can_change_room_list(room_id, requester)
async def edit_published_room_list(
- self, requester: Requester, room_id: str, visibility: str
+ self,
+ requester: Requester,
+ room_id: str,
+ visibility: Literal["public", "private"],
) -> None:
"""Edit the entry of the room in the published room list.
@@ -451,9 +456,6 @@ class DirectoryHandler:
if requester.is_guest:
raise AuthError(403, "Guests cannot edit the published room list")
- if visibility not in ["public", "private"]:
- raise SynapseError(400, "Invalid visibility setting")
-
if visibility == "public" and not self.enable_room_list_search:
# The room list has been disabled.
raise AuthError(
@@ -505,7 +507,11 @@ class DirectoryHandler:
await self.store.set_room_is_public(room_id, making_public)
async def edit_published_appservice_room_list(
- self, appservice_id: str, network_id: str, room_id: str, visibility: str
+ self,
+ appservice_id: str,
+ network_id: str,
+ room_id: str,
+ visibility: Literal["public", "private"],
) -> None:
"""Add or remove a room from the appservice/network specific public
room list.
@@ -516,9 +522,6 @@ class DirectoryHandler:
room_id
visibility: either "public" or "private"
"""
- if visibility not in ["public", "private"]:
- raise SynapseError(400, "Invalid visibility setting")
-
await self.store.set_room_is_public_appservice(
room_id, appservice_id, network_id, visibility == "public"
)
diff --git a/synapse/handlers/federation.py b/synapse/handlers/federation.py
index 986ffed3d5..5f7e0a1f79 100644
--- a/synapse/handlers/federation.py
+++ b/synapse/handlers/federation.py
@@ -45,6 +45,7 @@ from synapse.api.errors import (
Codes,
FederationDeniedError,
FederationError,
+ FederationPullAttemptBackoffError,
HttpResponseException,
LimitExceededError,
NotFoundError,
@@ -781,15 +782,27 @@ class FederationHandler:
# Send the signed event back to the room, and potentially receive some
# further information about the room in the form of partial state events
- stripped_room_state = await self.federation_client.send_knock(
- target_hosts, event
- )
+ knock_response = await self.federation_client.send_knock(target_hosts, event)
# Store any stripped room state events in the "unsigned" key of the event.
# This is a bit of a hack and is cribbing off of invites. Basically we
# store the room state here and retrieve it again when this event appears
# in the invitee's sync stream. It is stripped out for all other local users.
- event.unsigned["knock_room_state"] = stripped_room_state["knock_state_events"]
+ stripped_room_state = (
+ knock_response.get("knock_room_state")
+ # Since v1.37, Synapse incorrectly used "knock_state_events" for this field.
+ # Thus, we also check for a 'knock_state_events' to support old instances.
+ # See https://github.com/matrix-org/synapse/issues/14088.
+ or knock_response.get("knock_state_events")
+ )
+
+ if stripped_room_state is None:
+ raise KeyError(
+ "Missing 'knock_room_state' (or legacy 'knock_state_events') field in "
+ "send_knock response"
+ )
+
+ event.unsigned["knock_room_state"] = stripped_room_state
context = EventContext.for_outlier(self._storage_controllers)
stream_id = await self._federation_event_handler.persist_events_and_notify(
@@ -1708,7 +1721,22 @@ class FederationHandler:
destination, event
)
break
+ except FederationPullAttemptBackoffError as exc:
+ # Log a warning about why we failed to process the event (the error message
+ # for `FederationPullAttemptBackoffError` is pretty good)
+ logger.warning("_sync_partial_state_room: %s", exc)
+ # We do not record a failed pull attempt when we backoff fetching a missing
+ # `prev_event` because not being able to fetch the `prev_events` just means
+ # we won't be able to de-outlier the pulled event. But we can still use an
+ # `outlier` in the state/auth chain for another event. So we shouldn't stop
+ # a downstream event from trying to pull it.
+ #
+ # This avoids a cascade of backoff for all events in the DAG downstream from
+ # one event backoff upstream.
except FederationError as e:
+ # TODO: We should `record_event_failed_pull_attempt` here,
+ # see https://github.com/matrix-org/synapse/issues/13700
+
if attempt == len(destinations) - 1:
# We have tried every remote server for this event. Give up.
# TODO(faster_joins) giving up isn't the right thing to do
diff --git a/synapse/handlers/federation_event.py b/synapse/handlers/federation_event.py
index b94641a87d..06e41b5cc0 100644
--- a/synapse/handlers/federation_event.py
+++ b/synapse/handlers/federation_event.py
@@ -44,6 +44,7 @@ from synapse.api.errors import (
AuthError,
Codes,
FederationError,
+ FederationPullAttemptBackoffError,
HttpResponseException,
RequestSendFailed,
SynapseError,
@@ -414,7 +415,9 @@ class FederationEventHandler:
# First, precalculate the joined hosts so that the federation sender doesn't
# need to.
- await self._event_creation_handler.cache_joined_hosts_for_event(event, context)
+ await self._event_creation_handler.cache_joined_hosts_for_events(
+ [(event, context)]
+ )
await self._check_for_soft_fail(event, context=context, origin=origin)
await self._run_push_actions_and_persist_event(event, context)
@@ -565,6 +568,9 @@ class FederationEventHandler:
event: partial-state event to be de-partial-stated
Raises:
+ FederationPullAttemptBackoffError if we are are deliberately not attempting
+ to pull the given event over federation because we've already done so
+ recently and are backing off.
FederationError if we fail to request state from the remote server.
"""
logger.info("Updating state for %s", event.event_id)
@@ -920,6 +926,18 @@ class FederationEventHandler:
context,
backfilled=backfilled,
)
+ except FederationPullAttemptBackoffError as exc:
+ # Log a warning about why we failed to process the event (the error message
+ # for `FederationPullAttemptBackoffError` is pretty good)
+ logger.warning("_process_pulled_event: %s", exc)
+ # We do not record a failed pull attempt when we backoff fetching a missing
+ # `prev_event` because not being able to fetch the `prev_events` just means
+ # we won't be able to de-outlier the pulled event. But we can still use an
+ # `outlier` in the state/auth chain for another event. So we shouldn't stop
+ # a downstream event from trying to pull it.
+ #
+ # This avoids a cascade of backoff for all events in the DAG downstream from
+ # one event backoff upstream.
except FederationError as e:
await self._store.record_event_failed_pull_attempt(
event.room_id, event_id, str(e)
@@ -966,6 +984,9 @@ class FederationEventHandler:
The event context.
Raises:
+ FederationPullAttemptBackoffError if we are are deliberately not attempting
+ to pull the given event over federation because we've already done so
+ recently and are backing off.
FederationError if we fail to get the state from the remote server after any
missing `prev_event`s.
"""
@@ -976,6 +997,18 @@ class FederationEventHandler:
seen = await self._store.have_events_in_timeline(prevs)
missing_prevs = prevs - seen
+ # If we've already recently attempted to pull this missing event, don't
+ # try it again so soon. Since we have to fetch all of the prev_events, we can
+ # bail early here if we find any to ignore.
+ prevs_to_ignore = await self._store.get_event_ids_to_not_pull_from_backoff(
+ room_id, missing_prevs
+ )
+ if len(prevs_to_ignore) > 0:
+ raise FederationPullAttemptBackoffError(
+ event_ids=prevs_to_ignore,
+ message=f"While computing context for event={event_id}, not attempting to pull missing prev_event={prevs_to_ignore[0]} because we already tried to pull recently (backing off).",
+ )
+
if not missing_prevs:
return await self._state_handler.compute_event_context(event)
diff --git a/synapse/handlers/initial_sync.py b/synapse/handlers/initial_sync.py
index 860c82c110..9c335e6863 100644
--- a/synapse/handlers/initial_sync.py
+++ b/synapse/handlers/initial_sync.py
@@ -57,13 +57,7 @@ class InitialSyncHandler:
self.validator = EventValidator()
self.snapshot_cache: ResponseCache[
Tuple[
- str,
- Optional[StreamToken],
- Optional[StreamToken],
- str,
- Optional[int],
- bool,
- bool,
+ str, Optional[StreamToken], Optional[StreamToken], str, int, bool, bool
]
] = ResponseCache(hs.get_clock(), "initial_sync_cache")
self._event_serializer = hs.get_event_client_serializer()
@@ -154,11 +148,6 @@ class InitialSyncHandler:
public_room_ids = await self.store.get_public_room_ids()
- if pagin_config.limit is not None:
- limit = pagin_config.limit
- else:
- limit = 10
-
serializer_options = SerializeEventConfig(as_client_event=as_client_event)
async def handle_room(event: RoomsForUser) -> None:
@@ -210,7 +199,7 @@ class InitialSyncHandler:
run_in_background(
self.store.get_recent_events_for_room,
event.room_id,
- limit=limit,
+ limit=pagin_config.limit,
end_token=room_end_token,
),
deferred_room_state,
@@ -360,15 +349,11 @@ class InitialSyncHandler:
member_event_id
)
- limit = pagin_config.limit if pagin_config else None
- if limit is None:
- limit = 10
-
leave_position = await self.store.get_position_for_event(member_event_id)
stream_token = leave_position.to_room_stream_token()
messages, token = await self.store.get_recent_events_for_room(
- room_id, limit=limit, end_token=stream_token
+ room_id, limit=pagin_config.limit, end_token=stream_token
)
messages = await filter_events_for_client(
@@ -420,10 +405,6 @@ class InitialSyncHandler:
now_token = self.hs.get_event_sources().get_current_token()
- limit = pagin_config.limit if pagin_config else None
- if limit is None:
- limit = 10
-
room_members = [
m
for m in current_state.values()
@@ -467,7 +448,7 @@ class InitialSyncHandler:
run_in_background(
self.store.get_recent_events_for_room,
room_id,
- limit=limit,
+ limit=pagin_config.limit,
end_token=now_token.room_key,
),
),
diff --git a/synapse/handlers/message.py b/synapse/handlers/message.py
index da1acea275..4e55ebba0b 100644
--- a/synapse/handlers/message.py
+++ b/synapse/handlers/message.py
@@ -1390,7 +1390,7 @@ class EventCreationHandler:
extra_users=extra_users,
),
run_in_background(
- self.cache_joined_hosts_for_event, event, context
+ self.cache_joined_hosts_for_events, events_and_context
).addErrback(
log_failure, "cache_joined_hosts_for_event failed"
),
@@ -1491,62 +1491,65 @@ class EventCreationHandler:
await self.store.remove_push_actions_from_staging(event.event_id)
raise
- async def cache_joined_hosts_for_event(
- self, event: EventBase, context: EventContext
+ async def cache_joined_hosts_for_events(
+ self, events_and_context: List[Tuple[EventBase, EventContext]]
) -> None:
- """Precalculate the joined hosts at the event, when using Redis, so that
+ """Precalculate the joined hosts at each of the given events, when using Redis, so that
external federation senders don't have to recalculate it themselves.
"""
- if not self._external_cache.is_enabled():
- return
-
- # If external cache is enabled we should always have this.
- assert self._external_cache_joined_hosts_updates is not None
+ for event, _ in events_and_context:
+ if not self._external_cache.is_enabled():
+ return
- # We actually store two mappings, event ID -> prev state group,
- # state group -> joined hosts, which is much more space efficient
- # than event ID -> joined hosts.
- #
- # Note: We have to cache event ID -> prev state group, as we don't
- # store that in the DB.
- #
- # Note: We set the state group -> joined hosts cache if it hasn't been
- # set for a while, so that the expiry time is reset.
+ # If external cache is enabled we should always have this.
+ assert self._external_cache_joined_hosts_updates is not None
- state_entry = await self.state.resolve_state_groups_for_events(
- event.room_id, event_ids=event.prev_event_ids()
- )
+ # We actually store two mappings, event ID -> prev state group,
+ # state group -> joined hosts, which is much more space efficient
+ # than event ID -> joined hosts.
+ #
+ # Note: We have to cache event ID -> prev state group, as we don't
+ # store that in the DB.
+ #
+ # Note: We set the state group -> joined hosts cache if it hasn't been
+ # set for a while, so that the expiry time is reset.
- if state_entry.state_group:
- await self._external_cache.set(
- "event_to_prev_state_group",
- event.event_id,
- state_entry.state_group,
- expiry_ms=60 * 60 * 1000,
+ state_entry = await self.state.resolve_state_groups_for_events(
+ event.room_id, event_ids=event.prev_event_ids()
)
- if state_entry.state_group in self._external_cache_joined_hosts_updates:
- return
+ if state_entry.state_group:
+ await self._external_cache.set(
+ "event_to_prev_state_group",
+ event.event_id,
+ state_entry.state_group,
+ expiry_ms=60 * 60 * 1000,
+ )
- state = await state_entry.get_state(
- self._storage_controllers.state, StateFilter.all()
- )
- with opentracing.start_active_span("get_joined_hosts"):
- joined_hosts = await self.store.get_joined_hosts(
- event.room_id, state, state_entry
+ if state_entry.state_group in self._external_cache_joined_hosts_updates:
+ return
+
+ state = await state_entry.get_state(
+ self._storage_controllers.state, StateFilter.all()
)
+ with opentracing.start_active_span("get_joined_hosts"):
+ joined_hosts = await self.store.get_joined_hosts(
+ event.room_id, state, state_entry
+ )
- # Note that the expiry times must be larger than the expiry time in
- # _external_cache_joined_hosts_updates.
- await self._external_cache.set(
- "get_joined_hosts",
- str(state_entry.state_group),
- list(joined_hosts),
- expiry_ms=60 * 60 * 1000,
- )
+ # Note that the expiry times must be larger than the expiry time in
+ # _external_cache_joined_hosts_updates.
+ await self._external_cache.set(
+ "get_joined_hosts",
+ str(state_entry.state_group),
+ list(joined_hosts),
+ expiry_ms=60 * 60 * 1000,
+ )
- self._external_cache_joined_hosts_updates[state_entry.state_group] = None
+ self._external_cache_joined_hosts_updates[
+ state_entry.state_group
+ ] = None
async def _validate_canonical_alias(
self,
diff --git a/synapse/handlers/pagination.py b/synapse/handlers/pagination.py
index 1f83bab836..a4ca9cb8b4 100644
--- a/synapse/handlers/pagination.py
+++ b/synapse/handlers/pagination.py
@@ -458,11 +458,6 @@ class PaginationHandler:
# `/messages` should still works with live tokens when manually provided.
assert from_token.room_key.topological is not None
- if pagin_config.limit is None:
- # This shouldn't happen as we've set a default limit before this
- # gets called.
- raise Exception("limit not set")
-
room_token = from_token.room_key
async with self.pagination_lock.read(room_id):
diff --git a/synapse/handlers/presence.py b/synapse/handlers/presence.py
index 4e575ffbaa..2670e561d7 100644
--- a/synapse/handlers/presence.py
+++ b/synapse/handlers/presence.py
@@ -1596,7 +1596,9 @@ class PresenceEventSource(EventSource[int, UserPresenceState]):
self,
user: UserID,
from_key: Optional[int],
- limit: Optional[int] = None,
+ # Having a default limit doesn't match the EventSource API, but some
+ # callers do not provide it. It is unused in this class.
+ limit: int = 0,
room_ids: Optional[Collection[str]] = None,
is_guest: bool = False,
explicit_room_id: Optional[str] = None,
diff --git a/synapse/handlers/receipts.py b/synapse/handlers/receipts.py
index 4a7ec9e426..ac01582442 100644
--- a/synapse/handlers/receipts.py
+++ b/synapse/handlers/receipts.py
@@ -257,7 +257,7 @@ class ReceiptEventSource(EventSource[int, JsonDict]):
self,
user: UserID,
from_key: int,
- limit: Optional[int],
+ limit: int,
room_ids: Iterable[str],
is_guest: bool,
explicit_room_id: Optional[str] = None,
diff --git a/synapse/handlers/relations.py b/synapse/handlers/relations.py
index cc5e45c241..0a0c6d938e 100644
--- a/synapse/handlers/relations.py
+++ b/synapse/handlers/relations.py
@@ -11,6 +11,7 @@
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
+import enum
import logging
from typing import TYPE_CHECKING, Dict, FrozenSet, Iterable, List, Optional, Tuple
@@ -20,7 +21,7 @@ from synapse.api.constants import RelationTypes
from synapse.api.errors import SynapseError
from synapse.events import EventBase, relation_from_event
from synapse.logging.opentracing import trace
-from synapse.storage.databases.main.relations import _RelatedEvent
+from synapse.storage.databases.main.relations import ThreadsNextBatch, _RelatedEvent
from synapse.streams.config import PaginationConfig
from synapse.types import JsonDict, Requester, StreamToken, UserID
from synapse.visibility import filter_events_for_client
@@ -32,6 +33,13 @@ if TYPE_CHECKING:
logger = logging.getLogger(__name__)
+class ThreadsListInclude(str, enum.Enum):
+ """Valid values for the 'include' flag of /threads."""
+
+ all = "all"
+ participated = "participated"
+
+
@attr.s(slots=True, frozen=True, auto_attribs=True)
class _ThreadAggregation:
# The latest event in the thread.
@@ -108,9 +116,6 @@ class RelationsHandler:
if event is None:
raise SynapseError(404, "Unknown parent event.")
- # TODO Update pagination config to not allow None limits.
- assert pagin_config.limit is not None
-
# Note that ignored users are not passed into get_relations_for_event
# below. Ignored users are handled in filter_events_for_client (and by
# not passing them in here we should get a better cache hit rate).
@@ -482,3 +487,79 @@ class RelationsHandler:
results.setdefault(event_id, BundledAggregations()).replace = edit
return results
+
+ async def get_threads(
+ self,
+ requester: Requester,
+ room_id: str,
+ include: ThreadsListInclude,
+ limit: int = 5,
+ from_token: Optional[ThreadsNextBatch] = None,
+ ) -> JsonDict:
+ """Get related events of a event, ordered by topological ordering.
+
+ Args:
+ requester: The user requesting the relations.
+ room_id: The room the event belongs to.
+ include: One of "all" or "participated" to indicate which threads should
+ be returned.
+ limit: Only fetch the most recent `limit` events.
+ from_token: Fetch rows from the given token, or from the start if None.
+
+ Returns:
+ The pagination chunk.
+ """
+
+ user_id = requester.user.to_string()
+
+ # TODO Properly handle a user leaving a room.
+ (_, member_event_id) = await self._auth.check_user_in_room_or_world_readable(
+ room_id, requester, allow_departed_users=True
+ )
+
+ # Note that ignored users are not passed into get_relations_for_event
+ # below. Ignored users are handled in filter_events_for_client (and by
+ # not passing them in here we should get a better cache hit rate).
+ thread_roots, next_batch = await self._main_store.get_threads(
+ room_id=room_id, limit=limit, from_token=from_token
+ )
+
+ events = await self._main_store.get_events_as_list(thread_roots)
+
+ if include == ThreadsListInclude.participated:
+ # Pre-seed thread participation with whether the requester sent the event.
+ participated = {event.event_id: event.sender == user_id for event in events}
+ # For events the requester did not send, check the database for whether
+ # the requester sent a threaded reply.
+ participated.update(
+ await self._main_store.get_threads_participated(
+ [eid for eid, p in participated.items() if not p],
+ user_id,
+ )
+ )
+
+ # Limit the returned threads to those the user has participated in.
+ events = [event for event in events if participated[event.event_id]]
+
+ events = await filter_events_for_client(
+ self._storage_controllers,
+ user_id,
+ events,
+ is_peeking=(member_event_id is None),
+ )
+
+ aggregations = await self.get_bundled_aggregations(
+ events, requester.user.to_string()
+ )
+
+ now = self._clock.time_msec()
+ serialized_events = self._event_serializer.serialize_events(
+ events, now, bundle_aggregations=aggregations
+ )
+
+ return_value: JsonDict = {"chunk": serialized_events}
+
+ if next_batch:
+ return_value["next_batch"] = str(next_batch)
+
+ return return_value
diff --git a/synapse/handlers/room.py b/synapse/handlers/room.py
index 57ab05ad25..4e1aacb408 100644
--- a/synapse/handlers/room.py
+++ b/synapse/handlers/room.py
@@ -1646,7 +1646,7 @@ class RoomEventSource(EventSource[RoomStreamToken, EventBase]):
self,
user: UserID,
from_key: RoomStreamToken,
- limit: Optional[int],
+ limit: int,
room_ids: Collection[str],
is_guest: bool,
explicit_room_id: Optional[str] = None,
diff --git a/synapse/handlers/typing.py b/synapse/handlers/typing.py
index f953691669..a0ea719430 100644
--- a/synapse/handlers/typing.py
+++ b/synapse/handlers/typing.py
@@ -513,7 +513,7 @@ class TypingNotificationEventSource(EventSource[int, JsonDict]):
self,
user: UserID,
from_key: int,
- limit: Optional[int],
+ limit: int,
room_ids: Iterable[str],
is_guest: bool,
explicit_room_id: Optional[str] = None,
|
