This repository has been archived by the owner on Apr 26, 2024. It is now read-only.
-
-
Notifications
You must be signed in to change notification settings - Fork 2.1k
/
federation.py
2292 lines (1907 loc) · 80.4 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
# -*- coding: utf-8 -*-
# Copyright 2014-2016 OpenMarket Ltd
#
# 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."""
from signedjson.key import decode_verify_key_bytes
from signedjson.sign import verify_signed_json
from unpaddedbase64 import decode_base64
from ._base import BaseHandler
from synapse.api.errors import (
AuthError, FederationError, StoreError, CodeMessageException, SynapseError,
)
from synapse.api.constants import EventTypes, Membership, RejectedReason
from synapse.events.validator import EventValidator
from synapse.util import unwrapFirstError, logcontext
from synapse.util.metrics import measure_func
from synapse.util.logutils import log_function
from synapse.util.async import run_on_reactor, Linearizer
from synapse.util.frozenutils import unfreeze
from synapse.crypto.event_signing import (
compute_event_signature, add_hashes_and_signatures,
)
from synapse.types import UserID, get_domain_from_id
from synapse.events.utils import prune_event
from synapse.util.retryutils import NotRetryingDestination
from synapse.util.distributor import user_joined_room
from twisted.internet import defer
import itertools
import logging
logger = logging.getLogger(__name__)
class FederationHandler(BaseHandler):
"""Handles events that originated from federation.
Responsible for:
a) handling received Pdus before handing them on as Events to the rest
of the home server (including auth and state conflict resoultion)
b) converting events that were produced by local clients that may need
to be sent to remote home servers.
c) doing the necessary dances to invite remote users and join remote
rooms.
"""
def __init__(self, hs):
super(FederationHandler, self).__init__(hs)
self.hs = hs
self.store = hs.get_datastore()
self.replication_layer = hs.get_replication_layer()
self.state_handler = hs.get_state_handler()
self.server_name = hs.hostname
self.keyring = hs.get_keyring()
self.action_generator = hs.get_action_generator()
self.is_mine_id = hs.is_mine_id
self.pusher_pool = hs.get_pusherpool()
self.spam_checker = hs.get_spam_checker()
self.replication_layer.set_handler(self)
# When joining a room we need to queue any events for that room up
self.room_queues = {}
self._room_pdu_linearizer = Linearizer("fed_room_pdu")
@defer.inlineCallbacks
@log_function
def on_receive_pdu(self, origin, pdu, get_missing=True):
""" Process a PDU received via a federation /send/ transaction, or
via backfill of missing prev_events
Args:
origin (str): server which initiated the /send/ transaction. Will
be used to fetch missing events or state.
pdu (FrozenEvent): received PDU
get_missing (bool): True if we should fetch missing prev_events
Returns (Deferred): completes with None
"""
# We reprocess pdus when we have seen them only as outliers
existing = yield self.get_persisted_pdu(
origin, pdu.event_id, do_auth=False
)
# FIXME: Currently we fetch an event again when we already have it
# if it has been marked as an outlier.
already_seen = (
existing and (
not existing.internal_metadata.is_outlier()
or pdu.internal_metadata.is_outlier()
)
)
if already_seen:
logger.debug("Already seen pdu %s", pdu.event_id)
return
# If we are currently in the process of joining this room, then we
# queue up events for later processing.
if pdu.room_id in self.room_queues:
logger.info("Ignoring PDU %s for room %s from %s for now; join "
"in progress", pdu.event_id, pdu.room_id, origin)
self.room_queues[pdu.room_id].append((pdu, origin))
return
# If we're no longer in the room just ditch the event entirely. This
# is probably an old server that has come back and thinks we're still
# in the room (or we've been rejoined to the room by a state reset).
#
# If we were never in the room then maybe our database got vaped and
# we should check if we *are* in fact in the room. If we are then we
# can magically rejoin the room.
is_in_room = yield self.auth.check_host_in_room(
pdu.room_id,
self.server_name
)
if not is_in_room:
was_in_room = yield self.store.was_host_joined(
pdu.room_id, self.server_name,
)
if was_in_room:
logger.info(
"Ignoring PDU %s for room %s from %s as we've left the room!",
pdu.event_id, pdu.room_id, origin,
)
return
state = None
auth_chain = []
have_seen = yield self.store.have_events(
[ev for ev, _ in pdu.prev_events]
)
fetch_state = False
# Get missing pdus if necessary.
if not pdu.internal_metadata.is_outlier():
# We only backfill backwards to the min depth.
min_depth = yield self.get_min_depth_for_context(
pdu.room_id
)
logger.debug(
"_handle_new_pdu min_depth for %s: %d",
pdu.room_id, min_depth
)
prevs = {e_id for e_id, _ in pdu.prev_events}
seen = set(have_seen.keys())
if min_depth and pdu.depth < min_depth:
# This is so that we don't notify the user about this
# message, to work around the fact that some events will
# reference really really old events we really don't want to
# send to the clients.
pdu.internal_metadata.outlier = True
elif min_depth and pdu.depth > min_depth:
if get_missing and prevs - seen:
# If we're missing stuff, ensure we only fetch stuff one
# at a time.
logger.info(
"Acquiring lock for room %r to fetch %d missing events: %r...",
pdu.room_id, len(prevs - seen), list(prevs - seen)[:5],
)
with (yield self._room_pdu_linearizer.queue(pdu.room_id)):
logger.info(
"Acquired lock for room %r to fetch %d missing events",
pdu.room_id, len(prevs - seen),
)
yield self._get_missing_events_for_pdu(
origin, pdu, prevs, min_depth
)
# Update the set of things we've seen after trying to
# fetch the missing stuff
have_seen = yield self.store.have_events(prevs)
seen = set(have_seen.iterkeys())
if not prevs - seen:
logger.info(
"Found all missing prev events for %s", pdu.event_id
)
elif prevs - seen:
logger.info(
"Not fetching %d missing events for room %r,event %s: %r...",
len(prevs - seen), pdu.room_id, pdu.event_id,
list(prevs - seen)[:5],
)
if prevs - seen:
logger.info(
"Still missing %d events for room %r: %r...",
len(prevs - seen), pdu.room_id, list(prevs - seen)[:5]
)
fetch_state = True
if fetch_state:
# We need to get the state at this event, since we haven't
# processed all the prev events.
logger.debug(
"_handle_new_pdu getting state for %s",
pdu.room_id
)
try:
state, auth_chain = yield self.replication_layer.get_state_for_room(
origin, pdu.room_id, pdu.event_id,
)
except Exception:
logger.exception("Failed to get state for event: %s", pdu.event_id)
yield self._process_received_pdu(
origin,
pdu,
state=state,
auth_chain=auth_chain,
)
@defer.inlineCallbacks
def _get_missing_events_for_pdu(self, origin, pdu, prevs, min_depth):
"""
Args:
origin (str): Origin of the pdu. Will be called to get the missing events
pdu: received pdu
prevs (set(str)): List of event ids which we are missing
min_depth (int): Minimum depth of events to return.
"""
# We recalculate seen, since it may have changed.
have_seen = yield self.store.have_events(prevs)
seen = set(have_seen.keys())
if not prevs - seen:
return
latest = yield self.store.get_latest_event_ids_in_room(
pdu.room_id
)
# We add the prev events that we have seen to the latest
# list to ensure the remote server doesn't give them to us
latest = set(latest)
latest |= seen
logger.info(
"Missing %d events for room %r pdu %s: %r...",
len(prevs - seen), pdu.room_id, pdu.event_id, list(prevs - seen)[:5]
)
# XXX: we set timeout to 10s to help workaround
# https://github.com/matrix-org/synapse/issues/1733.
# The reason is to avoid holding the linearizer lock
# whilst processing inbound /send transactions, causing
# FDs to stack up and block other inbound transactions
# which empirically can currently take up to 30 minutes.
#
# N.B. this explicitly disables retry attempts.
#
# N.B. this also increases our chances of falling back to
# fetching fresh state for the room if the missing event
# can't be found, which slightly reduces our security.
# it may also increase our DAG extremity count for the room,
# causing additional state resolution? See #1760.
# However, fetching state doesn't hold the linearizer lock
# apparently.
#
# see https://github.com/matrix-org/synapse/pull/1744
missing_events = yield self.replication_layer.get_missing_events(
origin,
pdu.room_id,
earliest_events_ids=list(latest),
latest_events=[pdu],
limit=10,
min_depth=min_depth,
timeout=10000,
)
logger.info(
"Got %d events: %r...",
len(missing_events), [e.event_id for e in missing_events[:5]]
)
# We want to sort these by depth so we process them and
# tell clients about them in order.
missing_events.sort(key=lambda x: x.depth)
for e in missing_events:
logger.info("Handling found event %s", e.event_id)
yield self.on_receive_pdu(
origin,
e,
get_missing=False
)
@log_function
@defer.inlineCallbacks
def _process_received_pdu(self, origin, pdu, state, auth_chain):
""" Called when we have a new pdu. We need to do auth checks and put it
through the StateHandler.
"""
event = pdu
logger.debug("Processing event: %s", event)
# FIXME (erikj): Awful hack to make the case where we are not currently
# in the room work
# If state and auth_chain are None, then we don't need to do this check
# as we already know we have enough state in the DB to handle this
# event.
if state and auth_chain and not event.internal_metadata.is_outlier():
is_in_room = yield self.auth.check_host_in_room(
event.room_id,
self.server_name
)
else:
is_in_room = True
if not is_in_room:
logger.info(
"Got event for room we're not in: %r %r",
event.room_id, event.event_id
)
try:
event_stream_id, max_stream_id = yield self._persist_auth_tree(
origin, auth_chain, state, event
)
except AuthError as e:
raise FederationError(
"ERROR",
e.code,
e.msg,
affected=event.event_id,
)
else:
event_ids = set()
if state:
event_ids |= {e.event_id for e in state}
if auth_chain:
event_ids |= {e.event_id for e in auth_chain}
seen_ids = set(
(yield self.store.have_events(event_ids)).keys()
)
if state and auth_chain is not None:
# If we have any state or auth_chain given to us by the replication
# layer, then we should handle them (if we haven't before.)
event_infos = []
for e in itertools.chain(auth_chain, state):
if e.event_id in seen_ids:
continue
e.internal_metadata.outlier = True
auth_ids = [e_id for e_id, _ in e.auth_events]
auth = {
(e.type, e.state_key): e for e in auth_chain
if e.event_id in auth_ids or e.type == EventTypes.Create
}
event_infos.append({
"event": e,
"auth_events": auth,
})
seen_ids.add(e.event_id)
yield self._handle_new_events(origin, event_infos)
try:
context, event_stream_id, max_stream_id = yield self._handle_new_event(
origin,
event,
state=state,
)
except AuthError as e:
raise FederationError(
"ERROR",
e.code,
e.msg,
affected=event.event_id,
)
room = yield self.store.get_room(event.room_id)
if not room:
try:
yield self.store.store_room(
room_id=event.room_id,
room_creator_user_id="",
is_public=False,
)
except StoreError:
logger.exception("Failed to store room.")
extra_users = []
if event.type == EventTypes.Member:
target_user_id = event.state_key
target_user = UserID.from_string(target_user_id)
extra_users.append(target_user)
self.notifier.on_new_room_event(
event, event_stream_id, max_stream_id,
extra_users=extra_users
)
if event.type == EventTypes.Member:
if event.membership == Membership.JOIN:
# Only fire user_joined_room if the user has acutally
# joined the room. Don't bother if the user is just
# changing their profile info.
newly_joined = True
prev_state_id = context.prev_state_ids.get(
(event.type, event.state_key)
)
if prev_state_id:
prev_state = yield self.store.get_event(
prev_state_id, allow_none=True,
)
if prev_state and prev_state.membership == Membership.JOIN:
newly_joined = False
if newly_joined:
user = UserID.from_string(event.state_key)
yield user_joined_room(self.distributor, user, event.room_id)
@measure_func("_filter_events_for_server")
@defer.inlineCallbacks
def _filter_events_for_server(self, server_name, room_id, events):
event_to_state_ids = yield self.store.get_state_ids_for_events(
frozenset(e.event_id for e in events),
types=(
(EventTypes.RoomHistoryVisibility, ""),
(EventTypes.Member, None),
)
)
# We only want to pull out member events that correspond to the
# server's domain.
def check_match(id):
try:
return server_name == get_domain_from_id(id)
except Exception:
return False
# Parses mapping `event_id -> (type, state_key) -> state event_id`
# to get all state ids that we're interested in.
event_map = yield self.store.get_events([
e_id
for key_to_eid in event_to_state_ids.values()
for key, e_id in key_to_eid.items()
if key[0] != EventTypes.Member or check_match(key[1])
])
event_to_state = {
e_id: {
key: event_map[inner_e_id]
for key, inner_e_id in key_to_eid.items()
if inner_e_id in event_map
}
for e_id, key_to_eid in event_to_state_ids.items()
}
def redact_disallowed(event, state):
if not state:
return event
history = state.get((EventTypes.RoomHistoryVisibility, ''), None)
if history:
visibility = history.content.get("history_visibility", "shared")
if visibility in ["invited", "joined"]:
# We now loop through all state events looking for
# membership states for the requesting server to determine
# if the server is either in the room or has been invited
# into the room.
for ev in state.values():
if ev.type != EventTypes.Member:
continue
try:
domain = get_domain_from_id(ev.state_key)
except Exception:
continue
if domain != server_name:
continue
memtype = ev.membership
if memtype == Membership.JOIN:
return event
elif memtype == Membership.INVITE:
if visibility == "invited":
return event
else:
return prune_event(event)
return event
defer.returnValue([
redact_disallowed(e, event_to_state[e.event_id])
for e in events
])
@log_function
@defer.inlineCallbacks
def backfill(self, dest, room_id, limit, extremities):
""" Trigger a backfill request to `dest` for the given `room_id`
This will attempt to get more events from the remote. This may return
be successfull and still return no events if the other side has no new
events to offer.
"""
if dest == self.server_name:
raise SynapseError(400, "Can't backfill from self.")
events = yield self.replication_layer.backfill(
dest,
room_id,
limit=limit,
extremities=extremities,
)
# Don't bother processing events we already have.
seen_events = yield self.store.have_events_in_timeline(
set(e.event_id for e in events)
)
events = [e for e in events if e.event_id not in seen_events]
if not events:
defer.returnValue([])
event_map = {e.event_id: e for e in events}
event_ids = set(e.event_id for e in events)
edges = [
ev.event_id
for ev in events
if set(e_id for e_id, _ in ev.prev_events) - event_ids
]
logger.info(
"backfill: Got %d events with %d edges",
len(events), len(edges),
)
# For each edge get the current state.
auth_events = {}
state_events = {}
events_to_state = {}
for e_id in edges:
state, auth = yield self.replication_layer.get_state_for_room(
destination=dest,
room_id=room_id,
event_id=e_id
)
auth_events.update({a.event_id: a for a in auth})
auth_events.update({s.event_id: s for s in state})
state_events.update({s.event_id: s for s in state})
events_to_state[e_id] = state
required_auth = set(
a_id
for event in events + state_events.values() + auth_events.values()
for a_id, _ in event.auth_events
)
auth_events.update({
e_id: event_map[e_id] for e_id in required_auth if e_id in event_map
})
missing_auth = required_auth - set(auth_events)
failed_to_fetch = set()
# Try and fetch any missing auth events from both DB and remote servers.
# We repeatedly do this until we stop finding new auth events.
while missing_auth - failed_to_fetch:
logger.info("Missing auth for backfill: %r", missing_auth)
ret_events = yield self.store.get_events(missing_auth - failed_to_fetch)
auth_events.update(ret_events)
required_auth.update(
a_id for event in ret_events.values() for a_id, _ in event.auth_events
)
missing_auth = required_auth - set(auth_events)
if missing_auth - failed_to_fetch:
logger.info(
"Fetching missing auth for backfill: %r",
missing_auth - failed_to_fetch
)
results = yield logcontext.make_deferred_yieldable(defer.gatherResults(
[
logcontext.preserve_fn(self.replication_layer.get_pdu)(
[dest],
event_id,
outlier=True,
timeout=10000,
)
for event_id in missing_auth - failed_to_fetch
],
consumeErrors=True
)).addErrback(unwrapFirstError)
auth_events.update({a.event_id: a for a in results if a})
required_auth.update(
a_id
for event in results if event
for a_id, _ in event.auth_events
)
missing_auth = required_auth - set(auth_events)
failed_to_fetch = missing_auth - set(auth_events)
seen_events = yield self.store.have_events(
set(auth_events.keys()) | set(state_events.keys())
)
ev_infos = []
for a in auth_events.values():
if a.event_id in seen_events:
continue
a.internal_metadata.outlier = True
ev_infos.append({
"event": a,
"auth_events": {
(auth_events[a_id].type, auth_events[a_id].state_key):
auth_events[a_id]
for a_id, _ in a.auth_events
if a_id in auth_events
}
})
for e_id in events_to_state:
ev_infos.append({
"event": event_map[e_id],
"state": events_to_state[e_id],
"auth_events": {
(auth_events[a_id].type, auth_events[a_id].state_key):
auth_events[a_id]
for a_id, _ in event_map[e_id].auth_events
if a_id in auth_events
}
})
yield self._handle_new_events(
dest, ev_infos,
backfilled=True,
)
events.sort(key=lambda e: e.depth)
for event in events:
if event in events_to_state:
continue
# We store these one at a time since each event depends on the
# previous to work out the state.
# TODO: We can probably do something more clever here.
yield self._handle_new_event(
dest, event, backfilled=True,
)
defer.returnValue(events)
@defer.inlineCallbacks
def maybe_backfill(self, room_id, current_depth):
"""Checks the database to see if we should backfill before paginating,
and if so do.
"""
extremities = yield self.store.get_oldest_events_with_depth_in_room(
room_id
)
if not extremities:
logger.debug("Not backfilling as no extremeties found.")
return
# Check if we reached a point where we should start backfilling.
sorted_extremeties_tuple = sorted(
extremities.items(),
key=lambda e: -int(e[1])
)
max_depth = sorted_extremeties_tuple[0][1]
# We don't want to specify too many extremities as it causes the backfill
# request URI to be too long.
extremities = dict(sorted_extremeties_tuple[:5])
if current_depth > max_depth:
logger.debug(
"Not backfilling as we don't need to. %d < %d",
max_depth, current_depth,
)
return
# Now we need to decide which hosts to hit first.
# First we try hosts that are already in the room
# TODO: HEURISTIC ALERT.
curr_state = yield self.state_handler.get_current_state(room_id)
def get_domains_from_state(state):
joined_users = [
(state_key, int(event.depth))
for (e_type, state_key), event in state.items()
if e_type == EventTypes.Member
and event.membership == Membership.JOIN
]
joined_domains = {}
for u, d in joined_users:
try:
dom = get_domain_from_id(u)
old_d = joined_domains.get(dom)
if old_d:
joined_domains[dom] = min(d, old_d)
else:
joined_domains[dom] = d
except Exception:
pass
return sorted(joined_domains.items(), key=lambda d: d[1])
curr_domains = get_domains_from_state(curr_state)
likely_domains = [
domain for domain, depth in curr_domains
if domain != self.server_name
]
@defer.inlineCallbacks
def try_backfill(domains):
# TODO: Should we try multiple of these at a time?
for dom in domains:
try:
yield self.backfill(
dom, room_id,
limit=100,
extremities=[e for e in extremities.keys()]
)
# If this succeeded then we probably already have the
# appropriate stuff.
# TODO: We can probably do something more intelligent here.
defer.returnValue(True)
except SynapseError as e:
logger.info(
"Failed to backfill from %s because %s",
dom, e,
)
continue
except CodeMessageException as e:
if 400 <= e.code < 500:
raise
logger.info(
"Failed to backfill from %s because %s",
dom, e,
)
continue
except NotRetryingDestination as e:
logger.info(e.message)
continue
except Exception as e:
logger.exception(
"Failed to backfill from %s because %s",
dom, e,
)
continue
defer.returnValue(False)
success = yield try_backfill(likely_domains)
if success:
defer.returnValue(True)
# Huh, well *those* domains didn't work out. Lets try some domains
# from the time.
tried_domains = set(likely_domains)
tried_domains.add(self.server_name)
event_ids = list(extremities.keys())
logger.debug("calling resolve_state_groups in _maybe_backfill")
states = yield logcontext.make_deferred_yieldable(defer.gatherResults(
[
logcontext.preserve_fn(self.state_handler.resolve_state_groups)(
room_id, [e]
)
for e in event_ids
], consumeErrors=True,
))
states = dict(zip(event_ids, [s.state for s in states]))
state_map = yield self.store.get_events(
[e_id for ids in states.values() for e_id in ids],
get_prev_content=False
)
states = {
key: {
k: state_map[e_id]
for k, e_id in state_dict.items()
if e_id in state_map
} for key, state_dict in states.items()
}
for e_id, _ in sorted_extremeties_tuple:
likely_domains = get_domains_from_state(states[e_id])
success = yield try_backfill([
dom for dom in likely_domains
if dom not in tried_domains
])
if success:
defer.returnValue(True)
tried_domains.update(likely_domains)
defer.returnValue(False)
@defer.inlineCallbacks
def send_invite(self, target_host, event):
""" Sends the invite to the remote server for signing.
Invites must be signed by the invitee's server before distribution.
"""
pdu = yield self.replication_layer.send_invite(
destination=target_host,
room_id=event.room_id,
event_id=event.event_id,
pdu=event
)
defer.returnValue(pdu)
@defer.inlineCallbacks
def on_event_auth(self, event_id):
event = yield self.store.get_event(event_id)
auth = yield self.store.get_auth_chain(
[auth_id for auth_id, _ in event.auth_events],
include_given=True
)
for event in auth:
event.signatures.update(
compute_event_signature(
event,
self.hs.hostname,
self.hs.config.signing_key[0]
)
)
defer.returnValue([e for e in auth])
@log_function
@defer.inlineCallbacks
def do_invite_join(self, target_hosts, room_id, joinee, content):
""" Attempts to join the `joinee` to the room `room_id` via the
server `target_host`.
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.
"""
logger.debug("Joining %s to %s", joinee, room_id)
origin, event = yield self._make_and_verify_event(
target_hosts,
room_id,
joinee,
"join",
content,
)
# 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.room_queues)
self.room_queues[room_id] = []
yield self.store.clean_room_for_join(room_id)
handled_events = set()
try:
event = self._sign_event(event)
# Try the host we successfully got a response to /make_join/
# request first.
try:
target_hosts.remove(origin)
target_hosts.insert(0, origin)
except ValueError:
pass
ret = yield self.replication_layer.send_join(target_hosts, event)
origin = ret["origin"]
state = ret["state"]
auth_chain = ret["auth_chain"]
auth_chain.sort(key=lambda e: e.depth)
handled_events.update([s.event_id for s in state])
handled_events.update([a.event_id for a in auth_chain])
handled_events.add(event.event_id)
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)
try:
yield self.store.store_room(
room_id=room_id,
room_creator_user_id="",
is_public=False
)
except Exception:
# FIXME
pass
event_stream_id, max_stream_id = yield self._persist_auth_tree(
origin, auth_chain, state, event
)
self.notifier.on_new_room_event(
event, event_stream_id, max_stream_id,
extra_users=[joinee]
)
logger.debug("Finished joining %s to %s", joinee, room_id)
finally:
room_queue = self.room_queues[room_id]
del self.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 deferred, but don't wait for it.
logcontext.preserve_fn(self._handle_queued_pdus)(room_queue)
defer.returnValue(True)
@defer.inlineCallbacks
def _handle_queued_pdus(self, room_queue):
"""Process PDUs which got queued up while we were busy send_joining.
Args:
room_queue (list[FrozenEvent, str]): 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 %s", p.event_id, p.room_id)
yield self.on_receive_pdu(origin, p)
except Exception as e:
logger.warn(
"Error handling queued PDU %s from %s: %s",
p.event_id, origin, e)
@defer.inlineCallbacks
@log_function
def on_make_join_request(self, room_id, user_id):
""" 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.
"""
event_content = {"membership": Membership.JOIN}
builder = self.event_builder_factory.new({
"type": EventTypes.Member,
"content": event_content,