This repository has been archived by the owner on Apr 26, 2024. It is now read-only.
/
federation.py
2031 lines (1767 loc) · 84.1 KB
/
federation.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# Copyright 2014-2022 The Matrix.org Foundation C.I.C.
# Copyright 2020 Sorunome
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# 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.
"""Contains handlers for federation events."""
import enum
import itertools
import logging
from enum import Enum
from http import HTTPStatus
from typing import (
TYPE_CHECKING,
AbstractSet,
Dict,
Iterable,
List,
Optional,
Set,
Tuple,
Union,
)
import attr
from prometheus_client import Histogram
from signedjson.key import decode_verify_key_bytes
from signedjson.sign import verify_signed_json
from unpaddedbase64 import decode_base64
from synapse import event_auth
from synapse.api.constants import MAX_DEPTH, EventContentFields, EventTypes, Membership
from synapse.api.errors import (
AuthError,
CodeMessageException,
Codes,
FederationDeniedError,
FederationError,
FederationPullAttemptBackoffError,
HttpResponseException,
NotFoundError,
PartialStateConflictError,
RequestSendFailed,
SynapseError,
)
from synapse.api.room_versions import KNOWN_ROOM_VERSIONS, RoomVersion
from synapse.crypto.event_signing import compute_event_signature
from synapse.event_auth import validate_event_for_room_version
from synapse.events import EventBase
from synapse.events.snapshot import EventContext, UnpersistedEventContextBase
from synapse.events.validator import EventValidator
from synapse.federation.federation_client import InvalidResponseError
from synapse.http.servlet import assert_params_in_dict
from synapse.logging.context import nested_logging_context
from synapse.logging.opentracing import SynapseTags, set_tag, tag_args, trace
from synapse.metrics.background_process_metrics import run_as_background_process
from synapse.module_api import NOT_SPAM
from synapse.replication.http.federation import (
ReplicationCleanRoomRestServlet,
ReplicationStoreRoomOnOutlierMembershipRestServlet,
)
from synapse.storage.databases.main.events_worker import EventRedactBehaviour
from synapse.types import JsonDict, StrCollection, get_domain_from_id
from synapse.types.state import StateFilter
from synapse.util.async_helpers import Linearizer
from synapse.util.retryutils import NotRetryingDestination
from synapse.visibility import filter_events_for_server
if TYPE_CHECKING:
from synapse.server import HomeServer
logger = logging.getLogger(__name__)
# Added to debug performance and track progress on optimizations
backfill_processing_before_timer = Histogram(
"synapse_federation_backfill_processing_before_time_seconds",
"sec",
[],
buckets=(
0.1,
0.5,
1.0,
2.5,
5.0,
7.5,
10.0,
15.0,
20.0,
30.0,
40.0,
60.0,
80.0,
"+Inf",
),
)
# TODO: We can refactor this away now that there is only one backfill point again
class _BackfillPointType(Enum):
# a regular backwards extremity (ie, an event which we don't yet have, but which
# is referred to by other events in the DAG)
BACKWARDS_EXTREMITY = enum.auto()
@attr.s(slots=True, auto_attribs=True, frozen=True)
class _BackfillPoint:
"""A potential point we might backfill from"""
event_id: str
depth: int
type: _BackfillPointType
class FederationHandler:
"""Handles general incoming federation requests
Incoming events are *not* handled here, for which see FederationEventHandler.
"""
def __init__(self, hs: "HomeServer"):
self.hs = hs
self.clock = hs.get_clock()
self.store = hs.get_datastores().main
self._storage_controllers = hs.get_storage_controllers()
self._state_storage_controller = self._storage_controllers.state
self.federation_client = hs.get_federation_client()
self.state_handler = hs.get_state_handler()
self.server_name = hs.hostname
self.keyring = hs.get_keyring()
self.is_mine_id = hs.is_mine_id
self.is_mine_server_name = hs.is_mine_server_name
self._spam_checker_module_callbacks = hs.get_module_api_callbacks().spam_checker
self.event_creation_handler = hs.get_event_creation_handler()
self.event_builder_factory = hs.get_event_builder_factory()
self._event_auth_handler = hs.get_event_auth_handler()
self._server_notices_mxid = hs.config.servernotices.server_notices_mxid
self.config = hs.config
self.http_client = hs.get_proxied_blocklisted_http_client()
self._replication = hs.get_replication_data_handler()
self._federation_event_handler = hs.get_federation_event_handler()
self._device_handler = hs.get_device_handler()
self._bulk_push_rule_evaluator = hs.get_bulk_push_rule_evaluator()
self._notifier = hs.get_notifier()
self._clean_room_for_join_client = ReplicationCleanRoomRestServlet.make_client(
hs
)
if hs.config.worker.worker_app:
self._maybe_store_room_on_outlier_membership = (
ReplicationStoreRoomOnOutlierMembershipRestServlet.make_client(hs)
)
else:
self._maybe_store_room_on_outlier_membership = (
self.store.maybe_store_room_on_outlier_membership
)
self._room_backfill = Linearizer("room_backfill")
self._third_party_event_rules = (
hs.get_module_api_callbacks().third_party_event_rules
)
# Tracks running partial state syncs by room ID.
# Partial state syncs currently only run on the main process, so it's okay to
# track them in-memory for now.
self._active_partial_state_syncs: Set[str] = set()
# Tracks partial state syncs we may want to restart.
# A dictionary mapping room IDs to (initial destination, other destinations)
# tuples.
self._partial_state_syncs_maybe_needing_restart: Dict[
str, Tuple[Optional[str], AbstractSet[str]]
] = {}
# A lock guarding the partial state flag for rooms.
# When the lock is held for a given room, no other concurrent code may
# partial state or un-partial state the room.
self._is_partial_state_room_linearizer = Linearizer(
name="_is_partial_state_room_linearizer"
)
# if this is the main process, fire off a background process to resume
# any partial-state-resync operations which were in flight when we
# were shut down.
if not hs.config.worker.worker_app:
run_as_background_process(
"resume_sync_partial_state_room", self._resume_partial_state_room_sync
)
@trace
@tag_args
async def maybe_backfill(
self, room_id: str, current_depth: int, limit: int
) -> bool:
"""Checks the database to see if we should backfill before paginating,
and if so do.
Args:
room_id
current_depth: The depth from which we're paginating from. This is
used to decide if we should backfill and what extremities to
use.
limit: The number of events that the pagination request will
return. This is used as part of the heuristic to decide if we
should back paginate.
Returns:
True if we actually tried to backfill something, otherwise False.
"""
# Starting the processing time here so we can include the room backfill
# linearizer lock queue in the timing
processing_start_time = self.clock.time_msec()
async with self._room_backfill.queue(room_id):
return await self._maybe_backfill_inner(
room_id,
current_depth,
limit,
processing_start_time=processing_start_time,
)
@trace
@tag_args
async def _maybe_backfill_inner(
self,
room_id: str,
current_depth: int,
limit: int,
*,
processing_start_time: Optional[int],
) -> bool:
"""
Checks whether the `current_depth` is at or approaching any backfill
points in the room and if so, will backfill. We only care about
checking backfill points that happened before the `current_depth`
(meaning less than or equal to the `current_depth`).
Args:
room_id: The room to backfill in.
current_depth: The depth to check at for any upcoming backfill points.
limit: The max number of events to request from the remote federated server.
processing_start_time: The time when `maybe_backfill` started processing.
Only used for timing. If `None`, no timing observation will be made.
Returns:
True if we actually tried to backfill something, otherwise False.
"""
backwards_extremities = [
_BackfillPoint(event_id, depth, _BackfillPointType.BACKWARDS_EXTREMITY)
for event_id, depth in await self.store.get_backfill_points_in_room(
room_id=room_id,
current_depth=current_depth,
# We only need to end up with 5 extremities combined with the
# insertion event extremities to make the `/backfill` request
# but fetch an order of magnitude more to make sure there is
# enough even after we filter them by whether visible in the
# history. This isn't fool-proof as all backfill points within
# our limit could be filtered out but seems like a good amount
# to try with at least.
limit=50,
)
]
# we now have a list of potential places to backpaginate from. We prefer to
# start with the most recent (ie, max depth), so let's sort the list.
sorted_backfill_points: List[_BackfillPoint] = sorted(
backwards_extremities,
key=lambda e: -int(e.depth),
)
logger.debug(
"_maybe_backfill_inner: room_id: %s: current_depth: %s, limit: %s, "
"backfill points (%d): %s",
room_id,
current_depth,
limit,
len(sorted_backfill_points),
sorted_backfill_points,
)
set_tag(
SynapseTags.RESULT_PREFIX + "sorted_backfill_points",
str(sorted_backfill_points),
)
set_tag(
SynapseTags.RESULT_PREFIX + "sorted_backfill_points.length",
str(len(sorted_backfill_points)),
)
# If we have no backfill points lower than the `current_depth` then either we
# can a) bail or b) still attempt to backfill. We opt to try backfilling anyway
# just in case we do get relevant events. This is good for eventual consistency
# sake but we don't need to block the client for something that is just as
# likely not to return anything relevant so we backfill in the background. The
# only way, this could return something relevant is if we discover a new branch
# of history that extends all the way back to where we are currently paginating
# and it's within the 100 events that are returned from `/backfill`.
if not sorted_backfill_points and current_depth != MAX_DEPTH:
logger.debug(
"_maybe_backfill_inner: all backfill points are *after* current depth. Trying again with later backfill points."
)
run_as_background_process(
"_maybe_backfill_inner_anyway_with_max_depth",
self._maybe_backfill_inner,
room_id=room_id,
# We use `MAX_DEPTH` so that we find all backfill points next
# time (all events are below the `MAX_DEPTH`)
current_depth=MAX_DEPTH,
limit=limit,
# We don't want to start another timing observation from this
# nested recursive call. The top-most call can record the time
# overall otherwise the smaller one will throw off the results.
processing_start_time=None,
)
# We return `False` because we're backfilling in the background and there is
# no new events immediately for the caller to know about yet.
return False
# Even after recursing with `MAX_DEPTH`, we didn't find any
# backward extremities to backfill from.
if not sorted_backfill_points:
logger.debug(
"_maybe_backfill_inner: Not backfilling as no backward extremeties found."
)
return False
# If we're approaching an extremity we trigger a backfill, otherwise we
# no-op.
#
# We chose twice the limit here as then clients paginating backwards
# will send pagination requests that trigger backfill at least twice
# using the most recent extremity before it gets removed (see below). We
# chose more than one times the limit in case of failure, but choosing a
# much larger factor will result in triggering a backfill request much
# earlier than necessary.
max_depth_of_backfill_points = sorted_backfill_points[0].depth
if current_depth - 2 * limit > max_depth_of_backfill_points:
logger.debug(
"Not backfilling as we don't need to. %d < %d - 2 * %d",
max_depth_of_backfill_points,
current_depth,
limit,
)
return False
# For performance's sake, we only want to paginate from a particular extremity
# if we can actually see the events we'll get. Otherwise, we'd just spend a lot
# of resources to get redacted events. We check each extremity in turn and
# ignore those which users on our server wouldn't be able to see.
#
# Additionally, we limit ourselves to backfilling from at most 5 extremities,
# for two reasons:
#
# - The check which determines if we can see an extremity's events can be
# expensive (we load the full state for the room at each of the backfill
# points, or (worse) their successors)
# - We want to avoid the server-server API request URI becoming too long.
#
# *Note*: the spec wants us to keep backfilling until we reach the start
# of the room in case we are allowed to see some of the history. However,
# in practice that causes more issues than its worth, as (a) it's
# relatively rare for there to be any visible history and (b) even when
# there is it's often sufficiently long ago that clients would stop
# attempting to paginate before backfill reached the visible history.
extremities_to_request: List[str] = []
for bp in sorted_backfill_points:
if len(extremities_to_request) >= 5:
break
# For regular backwards extremities, we don't have the extremity events
# themselves, so we need to actually check the events that reference them -
# their "successor" events.
#
# TODO: Correctly handle the case where we are allowed to see the
# successor event but not the backward extremity, e.g. in the case of
# initial join of the server where we are allowed to see the join
# event but not anything before it. This would require looking at the
# state *before* the event, ignoring the special casing certain event
# types have.
event_ids_to_check = await self.store.get_successor_events(bp.event_id)
events_to_check = await self.store.get_events_as_list(
event_ids_to_check,
redact_behaviour=EventRedactBehaviour.as_is,
get_prev_content=False,
)
# We unset `filter_out_erased_senders` as we might otherwise get false
# positives from users having been erased.
filtered_extremities = await filter_events_for_server(
self._storage_controllers,
self.server_name,
self.server_name,
events_to_check,
redact=False,
filter_out_erased_senders=False,
filter_out_remote_partial_state_events=False,
)
if filtered_extremities:
extremities_to_request.append(bp.event_id)
else:
logger.debug(
"_maybe_backfill_inner: skipping extremity %s as it would not be visible",
bp,
)
if not extremities_to_request:
logger.debug(
"_maybe_backfill_inner: found no extremities which would be visible"
)
return False
logger.debug(
"_maybe_backfill_inner: extremities_to_request %s", extremities_to_request
)
set_tag(
SynapseTags.RESULT_PREFIX + "extremities_to_request",
str(extremities_to_request),
)
set_tag(
SynapseTags.RESULT_PREFIX + "extremities_to_request.length",
str(len(extremities_to_request)),
)
# Now we need to decide which hosts to hit first.
# First we try hosts that are already in the room.
# TODO: HEURISTIC ALERT.
likely_domains = (
await self._storage_controllers.state.get_current_hosts_in_room_ordered(
room_id
)
)
async def try_backfill(domains: StrCollection) -> bool:
# TODO: Should we try multiple of these at a time?
# Number of contacted remote homeservers that have denied our backfill
# request with a 4xx code.
denied_count = 0
# Maximum number of contacted remote homeservers that can deny our
# backfill request with 4xx codes before we give up.
max_denied_count = 5
for dom in domains:
# We don't want to ask our own server for information we don't have
if self.is_mine_server_name(dom):
continue
try:
await self._federation_event_handler.backfill(
dom, room_id, limit=100, extremities=extremities_to_request
)
# If this succeeded then we probably already have the
# appropriate stuff.
# TODO: We can probably do something more intelligent here.
return True
except NotRetryingDestination as e:
logger.info("_maybe_backfill_inner: %s", e)
continue
except FederationDeniedError:
logger.info(
"_maybe_backfill_inner: Not attempting to backfill from %s because the homeserver is not on our federation whitelist",
dom,
)
continue
except (SynapseError, InvalidResponseError) as e:
logger.info("Failed to backfill from %s because %s", dom, e)
continue
except HttpResponseException as e:
if 400 <= e.code < 500:
logger.warning(
"Backfill denied from %s because %s [%d/%d]",
dom,
e,
denied_count,
max_denied_count,
)
denied_count += 1
if denied_count >= max_denied_count:
return False
continue
logger.info("Failed to backfill from %s because %s", dom, e)
continue
except CodeMessageException as e:
if 400 <= e.code < 500:
logger.warning(
"Backfill denied from %s because %s [%d/%d]",
dom,
e,
denied_count,
max_denied_count,
)
denied_count += 1
if denied_count >= max_denied_count:
return False
continue
logger.info("Failed to backfill from %s because %s", dom, e)
continue
except RequestSendFailed as e:
logger.info("Failed to get backfill from %s because %s", dom, e)
continue
except Exception as e:
logger.exception("Failed to backfill from %s because %s", dom, e)
continue
return False
# If we have the `processing_start_time`, then we can make an
# observation. We wouldn't have the `processing_start_time` in the case
# where `_maybe_backfill_inner` is recursively called to find any
# backfill points regardless of `current_depth`.
if processing_start_time is not None:
processing_end_time = self.clock.time_msec()
backfill_processing_before_timer.observe(
(processing_end_time - processing_start_time) / 1000
)
success = await try_backfill(likely_domains)
if success:
return True
# TODO: we could also try servers which were previously in the room, but
# are no longer.
return False
async def send_invite(self, target_host: str, event: EventBase) -> EventBase:
"""Sends the invite to the remote server for signing.
Invites must be signed by the invitee's server before distribution.
"""
try:
pdu = await self.federation_client.send_invite(
destination=target_host,
room_id=event.room_id,
event_id=event.event_id,
pdu=event,
)
except RequestSendFailed:
raise SynapseError(502, f"Can't connect to server {target_host}")
return pdu
async def on_event_auth(self, event_id: str) -> List[EventBase]:
event = await self.store.get_event(event_id)
auth = await self.store.get_auth_chain(
event.room_id, list(event.auth_event_ids()), include_given=True
)
return list(auth)
async def do_invite_join(
self, target_hosts: Iterable[str], room_id: str, joinee: str, content: JsonDict
) -> Tuple[str, int]:
"""Attempts to join the `joinee` to the room `room_id` via the
servers contained in `target_hosts`.
This first triggers a /make_join/ request that returns a partial
event that we can fill out and sign. This is then sent to the
remote server via /send_join/ which responds with the state at that
event and the auth_chains.
We suspend processing of any received events from this room until we
have finished processing the join.
Args:
target_hosts: List of servers to attempt to join the room with.
room_id: The ID of the room to join.
joinee: The User ID of the joining user.
content: The event content to use for the join event.
"""
# TODO: We should be able to call this on workers, but the upgrading of
# room stuff after join currently doesn't work on workers.
# TODO: Before we relax this condition, we need to allow re-syncing of
# partial room state to happen on workers.
assert self.config.worker.worker_app is None
logger.debug("Joining %s to %s", joinee, room_id)
origin, event, room_version_obj = await self._make_and_verify_event(
target_hosts,
room_id,
joinee,
"join",
content,
params={"ver": KNOWN_ROOM_VERSIONS},
)
# This shouldn't happen, because the RoomMemberHandler has a
# linearizer lock which only allows one operation per user per room
# at a time - so this is just paranoia.
assert room_id not in self._federation_event_handler.room_queues
self._federation_event_handler.room_queues[room_id] = []
is_host_joined = await self.store.is_host_joined(room_id, self.server_name)
if not is_host_joined:
# We may have old forward extremities lying around if the homeserver left
# the room completely in the past. Clear them out.
#
# Note that this check-then-clear is subject to races where
# * the homeserver is in the room and stops being in the room just after
# the check. We won't reset the forward extremities, but that's okay,
# since they will be almost up to date.
# * the homeserver is not in the room and starts being in the room just
# after the check. This can't happen, since `RoomMemberHandler` has a
# linearizer lock which prevents concurrent remote joins into the same
# room.
# In short, the races either have an acceptable outcome or should be
# impossible.
await self._clean_room_for_join(room_id)
try:
# Try the host we successfully got a response to /make_join/
# request first.
host_list = list(target_hosts)
try:
host_list.remove(origin)
host_list.insert(0, origin)
except ValueError:
pass
async with self._is_partial_state_room_linearizer.queue(room_id):
already_partial_state_room = await self.store.is_partial_state_room(
room_id
)
ret = await self.federation_client.send_join(
host_list,
event,
room_version_obj,
# Perform a full join when we are already in the room and it is a
# full state room, since we are not allowed to persist a partial
# state join event in a full state room. In the future, we could
# optimize this by always performing a partial state join and
# computing the state ourselves or retrieving it from the remote
# homeserver if necessary.
#
# There's a race where we leave the room, then perform a full join
# anyway. This should end up being fast anyway, since we would
# already have the full room state and auth chain persisted.
partial_state=not is_host_joined or already_partial_state_room,
)
event = ret.event
origin = ret.origin
state = ret.state
auth_chain = ret.auth_chain
auth_chain.sort(key=lambda e: e.depth)
logger.debug("do_invite_join auth_chain: %s", auth_chain)
logger.debug("do_invite_join state: %s", state)
logger.debug("do_invite_join event: %s", event)
# if this is the first time we've joined this room, it's time to add
# a row to `rooms` with the correct room version. If there's already a
# row there, we should override it, since it may have been populated
# based on an invite request which lied about the room version.
#
# federation_client.send_join has already checked that the room
# version in the received create event is the same as room_version_obj,
# so we can rely on it now.
#
await self.store.upsert_room_on_join(
room_id=room_id,
room_version=room_version_obj,
state_events=state,
)
if ret.partial_state and not already_partial_state_room:
# Mark the room as having partial state.
# The background process is responsible for unmarking this flag,
# even if the join fails.
# TODO(faster_joins):
# We may want to reset the partial state info if it's from an
# old, failed partial state join.
# https://github.com/matrix-org/synapse/issues/13000
await self.store.store_partial_state_room(
room_id=room_id,
servers=ret.servers_in_room,
device_lists_stream_id=self.store.get_device_stream_token(),
joined_via=origin,
)
try:
max_stream_id = (
await self._federation_event_handler.process_remote_join(
origin,
room_id,
auth_chain,
state,
event,
room_version_obj,
partial_state=ret.partial_state,
)
)
except PartialStateConflictError:
# This should be impossible, since we hold the lock on the room's
# partial statedness.
logger.error(
"Room %s was un-partial stated while processing remote join.",
room_id,
)
raise
else:
# Record the join event id for future use (when we finish the full
# join). We have to do this after persisting the event to keep
# foreign key constraints intact.
if ret.partial_state and not already_partial_state_room:
# TODO(faster_joins):
# We may want to reset the partial state info if it's from
# an old, failed partial state join.
# https://github.com/matrix-org/synapse/issues/13000
await self.store.write_partial_state_rooms_join_event_id(
room_id, event.event_id
)
finally:
# Always kick off the background process that asynchronously fetches
# state for the room.
# If the join failed, the background process is responsible for
# cleaning up — including unmarking the room as a partial state
# room.
if ret.partial_state:
# Kick off the process of asynchronously fetching the state for
# this room.
self._start_partial_state_room_sync(
initial_destination=origin,
other_destinations=ret.servers_in_room,
room_id=room_id,
)
# We wait here until this instance has seen the events come down
# replication (if we're using replication) as the below uses caches.
await self._replication.wait_for_stream_position(
self.config.worker.events_shard_config.get_instance(room_id),
"events",
max_stream_id,
)
# Check whether this room is the result of an upgrade of a room we already know
# about. If so, migrate over user information
predecessor = await self.store.get_room_predecessor(room_id)
if not predecessor or not isinstance(predecessor.get("room_id"), str):
return event.event_id, max_stream_id
old_room_id = predecessor["room_id"]
logger.debug(
"Found predecessor for %s during remote join: %s", room_id, old_room_id
)
# We retrieve the room member handler here as to not cause a cyclic dependency
member_handler = self.hs.get_room_member_handler()
await member_handler.transfer_room_state_on_room_upgrade(
old_room_id, room_id
)
logger.debug("Finished joining %s to %s", joinee, room_id)
return event.event_id, max_stream_id
finally:
room_queue = self._federation_event_handler.room_queues[room_id]
del self._federation_event_handler.room_queues[room_id]
# we don't need to wait for the queued events to be processed -
# it's just a best-effort thing at this point. We do want to do
# them roughly in order, though, otherwise we'll end up making
# lots of requests for missing prev_events which we do actually
# have. Hence we fire off the background task, but don't wait for it.
run_as_background_process(
"handle_queued_pdus", self._handle_queued_pdus, room_queue
)
async def do_knock(
self,
target_hosts: List[str],
room_id: str,
knockee: str,
content: JsonDict,
) -> Tuple[str, int]:
"""Sends the knock to the remote server.
This first triggers a make_knock request that returns a partial
event that we can fill out and sign. This is then sent to the
remote server via send_knock.
Knock events must be signed by the knockee's server before distributing.
Args:
target_hosts: A list of hosts that we want to try knocking through.
room_id: The ID of the room to knock on.
knockee: The ID of the user who is knocking.
content: The content of the knock event.
Returns:
A tuple of (event ID, stream ID).
Raises:
SynapseError: If the chosen remote server returns a 3xx/4xx code.
RuntimeError: If no servers were reachable.
"""
logger.debug("Knocking on room %s on behalf of user %s", room_id, knockee)
# Inform the remote server of the room versions we support
supported_room_versions = list(KNOWN_ROOM_VERSIONS.keys())
# Ask the remote server to create a valid knock event for us. Once received,
# we sign the event
params: Dict[str, Iterable[str]] = {"ver": supported_room_versions}
origin, event, event_format_version = await self._make_and_verify_event(
target_hosts, room_id, knockee, Membership.KNOCK, content, params=params
)
# Mark the knock as an outlier as we don't yet have the state at this point in
# the DAG.
event.internal_metadata.outlier = True
# ... but tell /sync to send it to clients anyway.
event.internal_metadata.out_of_band_membership = True
# Record the room ID and its version so that we have a record of the room
await self._maybe_store_room_on_outlier_membership(
room_id=event.room_id, room_version=event_format_version
)
# Initially try the host that we successfully called /make_knock on
try:
target_hosts.remove(origin)
target_hosts.insert(0, origin)
except ValueError:
pass
# Send the signed event back to the room, and potentially receive some
# further information about the room in the form of partial state events
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.
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(
event.room_id, [(event, context)]
)
return event.event_id, stream_id
async def _handle_queued_pdus(
self, room_queue: List[Tuple[EventBase, str]]
) -> None:
"""Process PDUs which got queued up while we were busy send_joining.
Args:
room_queue: list of PDUs to be processed and the servers that sent them
"""
for p, origin in room_queue:
try:
logger.info(
"Processing queued PDU %s which was received while we were joining",
p,
)
with nested_logging_context(p.event_id):
await self._federation_event_handler.on_receive_pdu(origin, p)
except Exception as e:
logger.warning(
"Error handling queued PDU %s from %s: %s", p.event_id, origin, e
)
async def on_make_join_request(
self, origin: str, room_id: str, user_id: str
) -> EventBase:
"""We've received a /make_join/ request, so we create a partial
join event for the room and return that. We do *not* persist or
process it until the other server has signed it and sent it back.
Args:
origin: The (verified) server name of the requesting server.
room_id: Room to create join event in
user_id: The user to create the join for
"""
if get_domain_from_id(user_id) != origin:
logger.info(
"Got /make_join request for user %r from different origin %s, ignoring",
user_id,
origin,
)
raise SynapseError(403, "User not from origin", Codes.FORBIDDEN)
# checking the room version will check that we've actually heard of the room
# (and return a 404 otherwise)
room_version = await self.store.get_room_version(room_id)
if await self.store.is_partial_state_room(room_id):
# If our server is still only partially joined, we can't give a complete
# response to /make_join, so return a 404 as we would if we weren't in the
# room at all.
# The main reason we can't respond properly is that we need to know about
# the auth events for the join event that we would return.
# We also should not bother entertaining the /make_join since we cannot
# handle the /send_join.
logger.info(
"Rejecting /make_join to %s because it's a partial state room", room_id
)
raise SynapseError(
404,
"Unable to handle /make_join right now; this server is not fully joined.",
errcode=Codes.NOT_FOUND,
)
# now check that we are *still* in the room
is_in_room = await self._event_auth_handler.is_host_in_room(
room_id, self.server_name
)
if not is_in_room:
logger.info(
"Got /make_join request for room %s we are no longer in",
room_id,
)
raise NotFoundError("Not an active room on this server")
event_content = {"membership": Membership.JOIN}
# If the current room is using restricted join rules, additional information
# may need to be included in the event content in order to efficiently
# validate the event.
#
# Note that this requires the /send_join request to come back to the
# same server.
prev_event_ids = None
if room_version.restricted_join_rule:
# Note that the room's state can change out from under us and render our
# nice join rules-conformant event non-conformant by the time we build the
# event. When this happens, our validation at the end fails and we respond
# to the requesting server with a 403, which is misleading — it indicates
# that the user is not allowed to join the room and the joining server
# should not bother retrying via this homeserver or any others, when
# in fact we've just messed up with building the event.
#
# To reduce the likelihood of this race, we capture the forward extremities
# of the room (prev_event_ids) just before fetching the current state, and
# hope that the state we fetch corresponds to the prev events we chose.
prev_event_ids = await self.store.get_prev_events_for_room(room_id)
state_ids = await self._state_storage_controller.get_current_state_ids(
room_id
)
if await self._event_auth_handler.has_restricted_join_rules(
state_ids, room_version
):
prev_member_event_id = state_ids.get((EventTypes.Member, user_id), None)
# If the user is invited or joined to the room already, then
# no additional info is needed.
include_auth_user_id = True
if prev_member_event_id:
prev_member_event = await self.store.get_event(prev_member_event_id)
include_auth_user_id = prev_member_event.membership not in (
Membership.JOIN,
Membership.INVITE,
)
if include_auth_user_id:
event_content[
EventContentFields.AUTHORISING_USER
] = await self._event_auth_handler.get_user_which_could_invite(
room_id,
state_ids,
)
builder = self.event_builder_factory.for_room_version(
room_version,
{