forked-synapse/synapse/handlers/room_batch.py

465 lines
20 KiB
Python
Raw Normal View History

import logging
from typing import TYPE_CHECKING, List, Tuple
from synapse.api.constants import EventContentFields, EventTypes
from synapse.appservice import ApplicationService
from synapse.http.servlet import assert_params_in_dict
from synapse.types import JsonDict, Requester, UserID, create_requester
from synapse.util.stringutils import random_string
if TYPE_CHECKING:
from synapse.server import HomeServer
logger = logging.getLogger(__name__)
class RoomBatchHandler:
def __init__(self, hs: "HomeServer"):
self.hs = hs
self.store = hs.get_datastores().main
2022-05-31 08:17:50 -04:00
self._state_storage_controller = hs.get_storage_controllers().state
self.event_creation_handler = hs.get_event_creation_handler()
self.room_member_handler = hs.get_room_member_handler()
self.auth = hs.get_auth()
async def inherit_depth_from_prev_ids(self, prev_event_ids: List[str]) -> int:
"""Finds the depth which would sort it after the most-recent
prev_event_id but before the successors of those events. If no
successors are found, we assume it's an historical extremity part of the
current batch and use the same depth of the prev_event_ids.
Args:
prev_event_ids: List of prev event IDs
Returns:
Inherited depth
"""
(
most_recent_prev_event_id,
most_recent_prev_event_depth,
) = await self.store.get_max_depth_of(prev_event_ids)
# We want to insert the historical event after the `prev_event` but before the successor event
#
# We inherit depth from the successor event instead of the `prev_event`
# because events returned from `/messages` are first sorted by `topological_ordering`
# which is just the `depth` and then tie-break with `stream_ordering`.
#
# We mark these inserted historical events as "backfilled" which gives them a
# negative `stream_ordering`. If we use the same depth as the `prev_event`,
# then our historical event will tie-break and be sorted before the `prev_event`
# when it should come after.
#
# We want to use the successor event depth so they appear after `prev_event` because
# it has a larger `depth` but before the successor event because the `stream_ordering`
# is negative before the successor event.
assert most_recent_prev_event_id is not None
successor_event_ids = await self.store.get_successor_events(
most_recent_prev_event_id
)
# If we can't find any successor events, then it's a forward extremity of
# historical messages and we can just inherit from the previous historical
# event which we can already assume has the correct depth where we want
# to insert into.
if not successor_event_ids:
depth = most_recent_prev_event_depth
else:
(
_,
oldest_successor_depth,
) = await self.store.get_min_depth_of(successor_event_ids)
depth = oldest_successor_depth
return depth
def create_insertion_event_dict(
self, sender: str, room_id: str, origin_server_ts: int
) -> JsonDict:
"""Creates an event dict for an "insertion" event with the proper fields
and a random batch ID.
Args:
sender: The event author MXID
room_id: The room ID that the event belongs to
origin_server_ts: Timestamp when the event was sent
Returns:
The new event dictionary to insert.
"""
next_batch_id = random_string(8)
insertion_event = {
"type": EventTypes.MSC2716_INSERTION,
"sender": sender,
"room_id": room_id,
"content": {
EventContentFields.MSC2716_NEXT_BATCH_ID: next_batch_id,
EventContentFields.MSC2716_HISTORICAL: True,
},
"origin_server_ts": origin_server_ts,
}
return insertion_event
async def create_requester_for_user_id_from_app_service(
self, user_id: str, app_service: ApplicationService
) -> Requester:
"""Creates a new requester for the given user_id
and validates that the app service is allowed to control
the given user.
Args:
user_id: The author MXID that the app service is controlling
app_service: The app service that controls the user
Returns:
Requester object
"""
await self.auth.validate_appservice_can_control_user_id(app_service, user_id)
return create_requester(user_id, app_service=app_service)
async def get_most_recent_full_state_ids_from_event_id_list(
self, event_ids: List[str]
) -> List[str]:
"""Find the most recent event_id and grab the full state at that event.
We will use this as a base to auth our historical messages against.
Args:
event_ids: List of event ID's to look at
Returns:
List of event ID's
"""
(
most_recent_event_id,
_,
) = await self.store.get_max_depth_of(event_ids)
# mapping from (type, state_key) -> state_event_id
assert most_recent_event_id is not None
2022-05-31 08:17:50 -04:00
prev_state_map = await self._state_storage_controller.get_state_ids_for_event(
most_recent_event_id
)
# List of state event ID's
full_state_ids = list(prev_state_map.values())
return full_state_ids
async def persist_state_events_at_start(
self,
state_events_at_start: List[JsonDict],
room_id: str,
initial_state_event_ids: List[str],
app_service_requester: Requester,
) -> List[str]:
"""Takes all `state_events_at_start` event dictionaries and creates/persists
them in a floating state event chain which don't resolve into the current room
state. They are floating because they reference no prev_events which disconnects
them from the normal DAG.
Args:
state_events_at_start:
room_id: Room where you want the events persisted in.
initial_state_event_ids:
The base set of state for the historical batch which the floating
state chain will derive from. This should probably be the state
from the `prev_event` defined by `/batch_send?prev_event_id=$abc`.
app_service_requester: The requester of an application service.
Returns:
List of state event ID's we just persisted
"""
assert app_service_requester.app_service
state_event_ids_at_start = []
state_event_ids = initial_state_event_ids.copy()
Resolve and share `state_groups` for all historical events in batch (MSC2716) (#10975) Resolve and share `state_groups` for all historical events in batch. This also helps for showing the appropriate avatar/displayname in Element and will work whenever `/messages` has one of the historical messages as the first message in the batch. This does have the flaw where if you just insert a single historical event somewhere, it probably won't resolve the state correctly from `/messages` or `/context` since it will grab a non historical event above or below with resolved state which never included the historical state back then. For the same reasions, this also does not work in Element between the transition from actual messages to historical messages. In the Gitter case, this isn't really a problem since all of the historical messages are in one big lump at the beginning of the room. For a future iteration, might be good to look at `/messages` and `/context` to additionally add the `state` for any historical messages in that batch. --- How are the `state_groups` shared? To illustrate the `state_group` sharing, see this example: **Before** (new `state_group` for every event 😬, very inefficient): ``` # Tests from https://github.com/matrix-org/complement/pull/206 $ COMPLEMENT_ALWAYS_PRINT_SERVER_LOGS=1 COMPLEMENT_DIR=../complement ./scripts-dev/complement.sh TestBackfillingHistory/parallel/should_resolve_member_state_events_for_historical_events create_new_client_event m.room.member event=$_JXfwUDIWS6xKGG4SmZXjSFrizhARM7QblhATVWWUcA state_group=None create_new_client_event org.matrix.msc2716.insertion event=$1ZBfmBKEjg94d-vGYymKrVYeghwBOuGJ3wubU1-I9y0 state_group=9 create_new_client_event org.matrix.msc2716.insertion event=$Mq2JvRetTyclPuozRI682SAjYp3GqRuPc8_cH5-ezPY state_group=10 create_new_client_event m.room.message event=$MfmY4rBQkxrIp8jVwVMTJ4PKnxSigpG9E2cn7S0AtTo state_group=11 create_new_client_event m.room.message event=$uYOv6V8wiF7xHwOMt-60d1AoOIbqLgrDLz6ZIQDdWUI state_group=12 create_new_client_event m.room.message event=$PAbkJRMxb0bX4A6av463faiAhxkE3FEObM1xB4D0UG4 state_group=13 create_new_client_event org.matrix.msc2716.batch event=$Oy_S7AWN7rJQe_MYwGPEy6RtbYklrI-tAhmfiLrCaKI state_group=14 ``` **After** (all events in batch sharing `state_group=10`) (the base insertion event has `state_group=8` which matches the `prev_event` we're inserting next to): ``` # Tests from https://github.com/matrix-org/complement/pull/206 $ COMPLEMENT_ALWAYS_PRINT_SERVER_LOGS=1 COMPLEMENT_DIR=../complement ./scripts-dev/complement.sh TestBackfillingHistory/parallel/should_resolve_member_state_events_for_historical_events create_new_client_event m.room.member event=$PWomJ8PwENYEYuVNoG30gqtybuQQSZ55eldBUSs0i0U state_group=None create_new_client_event org.matrix.msc2716.insertion event=$e_mCU7Eah9ABF6nQU7lu4E1RxIWccNF05AKaTT5m3lw state_group=9 create_new_client_event org.matrix.msc2716.insertion event=$ui7A3_GdXIcJq0C8GpyrF8X7B3DTjMd_WGCjogax7xU state_group=10 create_new_client_event m.room.message event=$EnTIM5rEGVezQJiYl62uFBl6kJ7B-sMxWqe2D_4FX1I state_group=10 create_new_client_event m.room.message event=$LGx5jGONnBPuNhAuZqHeEoXChd9ryVkuTZatGisOPjk state_group=10 create_new_client_event m.room.message event=$wW0zwoN50lbLu1KoKbybVMxLbKUj7GV_olozIc5i3M0 state_group=10 create_new_client_event org.matrix.msc2716.batch event=$5ZB6dtzqFBCEuMRgpkU201Qhx3WtXZGTz_YgldL6JrQ state_group=10 ```
2021-10-13 18:44:00 -04:00
Fix historical messages backfilling in random order on remote homeservers (MSC2716) (#11114) Fix https://github.com/matrix-org/synapse/issues/11091 Fix https://github.com/matrix-org/synapse/issues/10764 (side-stepping the issue because we no longer have to deal with `fake_prev_event_id`) 1. Made the `/backfill` response return messages in `(depth, stream_ordering)` order (previously only sorted by `depth`) - Technically, it shouldn't really matter how `/backfill` returns things but I'm just trying to make the `stream_ordering` a little more consistent from the origin to the remote homeservers in order to get the order of messages from `/messages` consistent ([sorted by `(topological_ordering, stream_ordering)`](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)). - Even now that we return backfilled messages in order, it still doesn't guarantee the same `stream_ordering` (and more importantly the [`/messages` order](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)) on the other server. For example, if a room has a bunch of history imported and someone visits a permalink to a historical message back in time, their homeserver will skip over the historical messages in between and insert the permalink as the next message in the `stream_order` and totally throw off the sort. - This will be even more the case when we add the [MSC3030 jump to date API endpoint](https://github.com/matrix-org/matrix-doc/pull/3030) so the static archives can navigate and jump to a certain date. - We're solving this in the future by switching to [online topological ordering](https://github.com/matrix-org/gomatrixserverlib/issues/187) and [chunking](https://github.com/matrix-org/synapse/issues/3785) which by its nature will apply retroactively to fix any inconsistencies introduced by people permalinking 2. As we're navigating `prev_events` to return in `/backfill`, we order by `depth` first (newest -> oldest) and now also tie-break based on the `stream_ordering` (newest -> oldest). This is technically important because MSC2716 inserts a bunch of historical messages at the same `depth` so it's best to be prescriptive about which ones we should process first. In reality, I think the code already looped over the historical messages as expected because the database is already in order. 3. Making the historical state chain and historical event chain float on their own by having no `prev_events` instead of a fake `prev_event` which caused backfill to get clogged with an unresolvable event. Fixes https://github.com/matrix-org/synapse/issues/11091 and https://github.com/matrix-org/synapse/issues/10764 4. We no longer find connected insertion events by finding a potential `prev_event` connection to the current event we're iterating over. We now solely rely on marker events which when processed, add the insertion event as an extremity and the federating homeserver can ask about it when time calls. - Related discussion, https://github.com/matrix-org/synapse/pull/11114#discussion_r741514793 Before | After --- | --- ![](https://user-images.githubusercontent.com/558581/139218681-b465c862-5c49-4702-a59e-466733b0cf45.png) | ![](https://user-images.githubusercontent.com/558581/146453159-a1609e0a-8324-439d-ae44-e4bce43ac6d1.png) #### Why aren't we sorting topologically when receiving backfill events? > The main reason we're going to opt to not sort topologically when receiving backfill events is because it's probably best to do whatever is easiest to make it just work. People will probably have opinions once they look at [MSC2716](https://github.com/matrix-org/matrix-doc/pull/2716) which could change whatever implementation anyway. > > As mentioned, ideally we would do this but code necessary to make the fake edges but it gets confusing and gives an impression of “just whyyyy” (feels icky). This problem also dissolves with online topological ordering. > > -- https://github.com/matrix-org/synapse/pull/11114#discussion_r741517138 See https://github.com/matrix-org/synapse/pull/11114#discussion_r739610091 for the technical difficulties
2022-02-07 16:54:13 -05:00
# Make the state events float off on their own by specifying no
# prev_events for the first one in the chain so we don't have a bunch of
# `@mxid joined the room` noise between each batch.
prev_event_ids_for_state_chain: List[str] = []
Resolve and share `state_groups` for all historical events in batch (MSC2716) (#10975) Resolve and share `state_groups` for all historical events in batch. This also helps for showing the appropriate avatar/displayname in Element and will work whenever `/messages` has one of the historical messages as the first message in the batch. This does have the flaw where if you just insert a single historical event somewhere, it probably won't resolve the state correctly from `/messages` or `/context` since it will grab a non historical event above or below with resolved state which never included the historical state back then. For the same reasions, this also does not work in Element between the transition from actual messages to historical messages. In the Gitter case, this isn't really a problem since all of the historical messages are in one big lump at the beginning of the room. For a future iteration, might be good to look at `/messages` and `/context` to additionally add the `state` for any historical messages in that batch. --- How are the `state_groups` shared? To illustrate the `state_group` sharing, see this example: **Before** (new `state_group` for every event 😬, very inefficient): ``` # Tests from https://github.com/matrix-org/complement/pull/206 $ COMPLEMENT_ALWAYS_PRINT_SERVER_LOGS=1 COMPLEMENT_DIR=../complement ./scripts-dev/complement.sh TestBackfillingHistory/parallel/should_resolve_member_state_events_for_historical_events create_new_client_event m.room.member event=$_JXfwUDIWS6xKGG4SmZXjSFrizhARM7QblhATVWWUcA state_group=None create_new_client_event org.matrix.msc2716.insertion event=$1ZBfmBKEjg94d-vGYymKrVYeghwBOuGJ3wubU1-I9y0 state_group=9 create_new_client_event org.matrix.msc2716.insertion event=$Mq2JvRetTyclPuozRI682SAjYp3GqRuPc8_cH5-ezPY state_group=10 create_new_client_event m.room.message event=$MfmY4rBQkxrIp8jVwVMTJ4PKnxSigpG9E2cn7S0AtTo state_group=11 create_new_client_event m.room.message event=$uYOv6V8wiF7xHwOMt-60d1AoOIbqLgrDLz6ZIQDdWUI state_group=12 create_new_client_event m.room.message event=$PAbkJRMxb0bX4A6av463faiAhxkE3FEObM1xB4D0UG4 state_group=13 create_new_client_event org.matrix.msc2716.batch event=$Oy_S7AWN7rJQe_MYwGPEy6RtbYklrI-tAhmfiLrCaKI state_group=14 ``` **After** (all events in batch sharing `state_group=10`) (the base insertion event has `state_group=8` which matches the `prev_event` we're inserting next to): ``` # Tests from https://github.com/matrix-org/complement/pull/206 $ COMPLEMENT_ALWAYS_PRINT_SERVER_LOGS=1 COMPLEMENT_DIR=../complement ./scripts-dev/complement.sh TestBackfillingHistory/parallel/should_resolve_member_state_events_for_historical_events create_new_client_event m.room.member event=$PWomJ8PwENYEYuVNoG30gqtybuQQSZ55eldBUSs0i0U state_group=None create_new_client_event org.matrix.msc2716.insertion event=$e_mCU7Eah9ABF6nQU7lu4E1RxIWccNF05AKaTT5m3lw state_group=9 create_new_client_event org.matrix.msc2716.insertion event=$ui7A3_GdXIcJq0C8GpyrF8X7B3DTjMd_WGCjogax7xU state_group=10 create_new_client_event m.room.message event=$EnTIM5rEGVezQJiYl62uFBl6kJ7B-sMxWqe2D_4FX1I state_group=10 create_new_client_event m.room.message event=$LGx5jGONnBPuNhAuZqHeEoXChd9ryVkuTZatGisOPjk state_group=10 create_new_client_event m.room.message event=$wW0zwoN50lbLu1KoKbybVMxLbKUj7GV_olozIc5i3M0 state_group=10 create_new_client_event org.matrix.msc2716.batch event=$5ZB6dtzqFBCEuMRgpkU201Qhx3WtXZGTz_YgldL6JrQ state_group=10 ```
2021-10-13 18:44:00 -04:00
Fix historical messages backfilling in random order on remote homeservers (MSC2716) (#11114) Fix https://github.com/matrix-org/synapse/issues/11091 Fix https://github.com/matrix-org/synapse/issues/10764 (side-stepping the issue because we no longer have to deal with `fake_prev_event_id`) 1. Made the `/backfill` response return messages in `(depth, stream_ordering)` order (previously only sorted by `depth`) - Technically, it shouldn't really matter how `/backfill` returns things but I'm just trying to make the `stream_ordering` a little more consistent from the origin to the remote homeservers in order to get the order of messages from `/messages` consistent ([sorted by `(topological_ordering, stream_ordering)`](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)). - Even now that we return backfilled messages in order, it still doesn't guarantee the same `stream_ordering` (and more importantly the [`/messages` order](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)) on the other server. For example, if a room has a bunch of history imported and someone visits a permalink to a historical message back in time, their homeserver will skip over the historical messages in between and insert the permalink as the next message in the `stream_order` and totally throw off the sort. - This will be even more the case when we add the [MSC3030 jump to date API endpoint](https://github.com/matrix-org/matrix-doc/pull/3030) so the static archives can navigate and jump to a certain date. - We're solving this in the future by switching to [online topological ordering](https://github.com/matrix-org/gomatrixserverlib/issues/187) and [chunking](https://github.com/matrix-org/synapse/issues/3785) which by its nature will apply retroactively to fix any inconsistencies introduced by people permalinking 2. As we're navigating `prev_events` to return in `/backfill`, we order by `depth` first (newest -> oldest) and now also tie-break based on the `stream_ordering` (newest -> oldest). This is technically important because MSC2716 inserts a bunch of historical messages at the same `depth` so it's best to be prescriptive about which ones we should process first. In reality, I think the code already looped over the historical messages as expected because the database is already in order. 3. Making the historical state chain and historical event chain float on their own by having no `prev_events` instead of a fake `prev_event` which caused backfill to get clogged with an unresolvable event. Fixes https://github.com/matrix-org/synapse/issues/11091 and https://github.com/matrix-org/synapse/issues/10764 4. We no longer find connected insertion events by finding a potential `prev_event` connection to the current event we're iterating over. We now solely rely on marker events which when processed, add the insertion event as an extremity and the federating homeserver can ask about it when time calls. - Related discussion, https://github.com/matrix-org/synapse/pull/11114#discussion_r741514793 Before | After --- | --- ![](https://user-images.githubusercontent.com/558581/139218681-b465c862-5c49-4702-a59e-466733b0cf45.png) | ![](https://user-images.githubusercontent.com/558581/146453159-a1609e0a-8324-439d-ae44-e4bce43ac6d1.png) #### Why aren't we sorting topologically when receiving backfill events? > The main reason we're going to opt to not sort topologically when receiving backfill events is because it's probably best to do whatever is easiest to make it just work. People will probably have opinions once they look at [MSC2716](https://github.com/matrix-org/matrix-doc/pull/2716) which could change whatever implementation anyway. > > As mentioned, ideally we would do this but code necessary to make the fake edges but it gets confusing and gives an impression of “just whyyyy” (feels icky). This problem also dissolves with online topological ordering. > > -- https://github.com/matrix-org/synapse/pull/11114#discussion_r741517138 See https://github.com/matrix-org/synapse/pull/11114#discussion_r739610091 for the technical difficulties
2022-02-07 16:54:13 -05:00
for index, state_event in enumerate(state_events_at_start):
assert_params_in_dict(
state_event, ["type", "origin_server_ts", "content", "sender"]
)
logger.debug(
"RoomBatchSendEventRestServlet inserting state_event=%s", state_event
)
event_dict = {
"type": state_event["type"],
"origin_server_ts": state_event["origin_server_ts"],
"content": state_event["content"],
"room_id": room_id,
"sender": state_event["sender"],
"state_key": state_event["state_key"],
}
# Mark all events as historical
event_dict["content"][EventContentFields.MSC2716_HISTORICAL] = True
# TODO: This is pretty much the same as some other code to handle inserting state in this file
if event_dict["type"] == EventTypes.Member:
membership = event_dict["content"].get("membership", None)
event_id, _ = await self.room_member_handler.update_membership(
await self.create_requester_for_user_id_from_app_service(
state_event["sender"], app_service_requester.app_service
),
target=UserID.from_string(event_dict["state_key"]),
room_id=room_id,
action=membership,
content=event_dict["content"],
historical=True,
# Only the first event in the state chain should be floating.
Fix historical messages backfilling in random order on remote homeservers (MSC2716) (#11114) Fix https://github.com/matrix-org/synapse/issues/11091 Fix https://github.com/matrix-org/synapse/issues/10764 (side-stepping the issue because we no longer have to deal with `fake_prev_event_id`) 1. Made the `/backfill` response return messages in `(depth, stream_ordering)` order (previously only sorted by `depth`) - Technically, it shouldn't really matter how `/backfill` returns things but I'm just trying to make the `stream_ordering` a little more consistent from the origin to the remote homeservers in order to get the order of messages from `/messages` consistent ([sorted by `(topological_ordering, stream_ordering)`](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)). - Even now that we return backfilled messages in order, it still doesn't guarantee the same `stream_ordering` (and more importantly the [`/messages` order](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)) on the other server. For example, if a room has a bunch of history imported and someone visits a permalink to a historical message back in time, their homeserver will skip over the historical messages in between and insert the permalink as the next message in the `stream_order` and totally throw off the sort. - This will be even more the case when we add the [MSC3030 jump to date API endpoint](https://github.com/matrix-org/matrix-doc/pull/3030) so the static archives can navigate and jump to a certain date. - We're solving this in the future by switching to [online topological ordering](https://github.com/matrix-org/gomatrixserverlib/issues/187) and [chunking](https://github.com/matrix-org/synapse/issues/3785) which by its nature will apply retroactively to fix any inconsistencies introduced by people permalinking 2. As we're navigating `prev_events` to return in `/backfill`, we order by `depth` first (newest -> oldest) and now also tie-break based on the `stream_ordering` (newest -> oldest). This is technically important because MSC2716 inserts a bunch of historical messages at the same `depth` so it's best to be prescriptive about which ones we should process first. In reality, I think the code already looped over the historical messages as expected because the database is already in order. 3. Making the historical state chain and historical event chain float on their own by having no `prev_events` instead of a fake `prev_event` which caused backfill to get clogged with an unresolvable event. Fixes https://github.com/matrix-org/synapse/issues/11091 and https://github.com/matrix-org/synapse/issues/10764 4. We no longer find connected insertion events by finding a potential `prev_event` connection to the current event we're iterating over. We now solely rely on marker events which when processed, add the insertion event as an extremity and the federating homeserver can ask about it when time calls. - Related discussion, https://github.com/matrix-org/synapse/pull/11114#discussion_r741514793 Before | After --- | --- ![](https://user-images.githubusercontent.com/558581/139218681-b465c862-5c49-4702-a59e-466733b0cf45.png) | ![](https://user-images.githubusercontent.com/558581/146453159-a1609e0a-8324-439d-ae44-e4bce43ac6d1.png) #### Why aren't we sorting topologically when receiving backfill events? > The main reason we're going to opt to not sort topologically when receiving backfill events is because it's probably best to do whatever is easiest to make it just work. People will probably have opinions once they look at [MSC2716](https://github.com/matrix-org/matrix-doc/pull/2716) which could change whatever implementation anyway. > > As mentioned, ideally we would do this but code necessary to make the fake edges but it gets confusing and gives an impression of “just whyyyy” (feels icky). This problem also dissolves with online topological ordering. > > -- https://github.com/matrix-org/synapse/pull/11114#discussion_r741517138 See https://github.com/matrix-org/synapse/pull/11114#discussion_r739610091 for the technical difficulties
2022-02-07 16:54:13 -05:00
# The rest should hang off each other in a chain.
allow_no_prev_events=index == 0,
prev_event_ids=prev_event_ids_for_state_chain,
# The first event in the state chain is floating with no
# `prev_events` which means it can't derive state from
# anywhere automatically. So we need to set some state
# explicitly.
#
# Make sure to use a copy of this list because we modify it
# later in the loop here. Otherwise it will be the same
# reference and also update in the event when we append
# later.
state_event_ids=state_event_ids.copy(),
)
else:
(
event,
_,
) = await self.event_creation_handler.create_and_send_nonmember_event(
await self.create_requester_for_user_id_from_app_service(
state_event["sender"], app_service_requester.app_service
),
event_dict,
historical=True,
# Only the first event in the state chain should be floating.
Fix historical messages backfilling in random order on remote homeservers (MSC2716) (#11114) Fix https://github.com/matrix-org/synapse/issues/11091 Fix https://github.com/matrix-org/synapse/issues/10764 (side-stepping the issue because we no longer have to deal with `fake_prev_event_id`) 1. Made the `/backfill` response return messages in `(depth, stream_ordering)` order (previously only sorted by `depth`) - Technically, it shouldn't really matter how `/backfill` returns things but I'm just trying to make the `stream_ordering` a little more consistent from the origin to the remote homeservers in order to get the order of messages from `/messages` consistent ([sorted by `(topological_ordering, stream_ordering)`](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)). - Even now that we return backfilled messages in order, it still doesn't guarantee the same `stream_ordering` (and more importantly the [`/messages` order](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)) on the other server. For example, if a room has a bunch of history imported and someone visits a permalink to a historical message back in time, their homeserver will skip over the historical messages in between and insert the permalink as the next message in the `stream_order` and totally throw off the sort. - This will be even more the case when we add the [MSC3030 jump to date API endpoint](https://github.com/matrix-org/matrix-doc/pull/3030) so the static archives can navigate and jump to a certain date. - We're solving this in the future by switching to [online topological ordering](https://github.com/matrix-org/gomatrixserverlib/issues/187) and [chunking](https://github.com/matrix-org/synapse/issues/3785) which by its nature will apply retroactively to fix any inconsistencies introduced by people permalinking 2. As we're navigating `prev_events` to return in `/backfill`, we order by `depth` first (newest -> oldest) and now also tie-break based on the `stream_ordering` (newest -> oldest). This is technically important because MSC2716 inserts a bunch of historical messages at the same `depth` so it's best to be prescriptive about which ones we should process first. In reality, I think the code already looped over the historical messages as expected because the database is already in order. 3. Making the historical state chain and historical event chain float on their own by having no `prev_events` instead of a fake `prev_event` which caused backfill to get clogged with an unresolvable event. Fixes https://github.com/matrix-org/synapse/issues/11091 and https://github.com/matrix-org/synapse/issues/10764 4. We no longer find connected insertion events by finding a potential `prev_event` connection to the current event we're iterating over. We now solely rely on marker events which when processed, add the insertion event as an extremity and the federating homeserver can ask about it when time calls. - Related discussion, https://github.com/matrix-org/synapse/pull/11114#discussion_r741514793 Before | After --- | --- ![](https://user-images.githubusercontent.com/558581/139218681-b465c862-5c49-4702-a59e-466733b0cf45.png) | ![](https://user-images.githubusercontent.com/558581/146453159-a1609e0a-8324-439d-ae44-e4bce43ac6d1.png) #### Why aren't we sorting topologically when receiving backfill events? > The main reason we're going to opt to not sort topologically when receiving backfill events is because it's probably best to do whatever is easiest to make it just work. People will probably have opinions once they look at [MSC2716](https://github.com/matrix-org/matrix-doc/pull/2716) which could change whatever implementation anyway. > > As mentioned, ideally we would do this but code necessary to make the fake edges but it gets confusing and gives an impression of “just whyyyy” (feels icky). This problem also dissolves with online topological ordering. > > -- https://github.com/matrix-org/synapse/pull/11114#discussion_r741517138 See https://github.com/matrix-org/synapse/pull/11114#discussion_r739610091 for the technical difficulties
2022-02-07 16:54:13 -05:00
# The rest should hang off each other in a chain.
allow_no_prev_events=index == 0,
prev_event_ids=prev_event_ids_for_state_chain,
# The first event in the state chain is floating with no
# `prev_events` which means it can't derive state from
# anywhere automatically. So we need to set some state
# explicitly.
#
# Make sure to use a copy of this list because we modify it
# later in the loop here. Otherwise it will be the same
# reference and also update in the event when we append later.
state_event_ids=state_event_ids.copy(),
)
event_id = event.event_id
state_event_ids_at_start.append(event_id)
state_event_ids.append(event_id)
Resolve and share `state_groups` for all historical events in batch (MSC2716) (#10975) Resolve and share `state_groups` for all historical events in batch. This also helps for showing the appropriate avatar/displayname in Element and will work whenever `/messages` has one of the historical messages as the first message in the batch. This does have the flaw where if you just insert a single historical event somewhere, it probably won't resolve the state correctly from `/messages` or `/context` since it will grab a non historical event above or below with resolved state which never included the historical state back then. For the same reasions, this also does not work in Element between the transition from actual messages to historical messages. In the Gitter case, this isn't really a problem since all of the historical messages are in one big lump at the beginning of the room. For a future iteration, might be good to look at `/messages` and `/context` to additionally add the `state` for any historical messages in that batch. --- How are the `state_groups` shared? To illustrate the `state_group` sharing, see this example: **Before** (new `state_group` for every event 😬, very inefficient): ``` # Tests from https://github.com/matrix-org/complement/pull/206 $ COMPLEMENT_ALWAYS_PRINT_SERVER_LOGS=1 COMPLEMENT_DIR=../complement ./scripts-dev/complement.sh TestBackfillingHistory/parallel/should_resolve_member_state_events_for_historical_events create_new_client_event m.room.member event=$_JXfwUDIWS6xKGG4SmZXjSFrizhARM7QblhATVWWUcA state_group=None create_new_client_event org.matrix.msc2716.insertion event=$1ZBfmBKEjg94d-vGYymKrVYeghwBOuGJ3wubU1-I9y0 state_group=9 create_new_client_event org.matrix.msc2716.insertion event=$Mq2JvRetTyclPuozRI682SAjYp3GqRuPc8_cH5-ezPY state_group=10 create_new_client_event m.room.message event=$MfmY4rBQkxrIp8jVwVMTJ4PKnxSigpG9E2cn7S0AtTo state_group=11 create_new_client_event m.room.message event=$uYOv6V8wiF7xHwOMt-60d1AoOIbqLgrDLz6ZIQDdWUI state_group=12 create_new_client_event m.room.message event=$PAbkJRMxb0bX4A6av463faiAhxkE3FEObM1xB4D0UG4 state_group=13 create_new_client_event org.matrix.msc2716.batch event=$Oy_S7AWN7rJQe_MYwGPEy6RtbYklrI-tAhmfiLrCaKI state_group=14 ``` **After** (all events in batch sharing `state_group=10`) (the base insertion event has `state_group=8` which matches the `prev_event` we're inserting next to): ``` # Tests from https://github.com/matrix-org/complement/pull/206 $ COMPLEMENT_ALWAYS_PRINT_SERVER_LOGS=1 COMPLEMENT_DIR=../complement ./scripts-dev/complement.sh TestBackfillingHistory/parallel/should_resolve_member_state_events_for_historical_events create_new_client_event m.room.member event=$PWomJ8PwENYEYuVNoG30gqtybuQQSZ55eldBUSs0i0U state_group=None create_new_client_event org.matrix.msc2716.insertion event=$e_mCU7Eah9ABF6nQU7lu4E1RxIWccNF05AKaTT5m3lw state_group=9 create_new_client_event org.matrix.msc2716.insertion event=$ui7A3_GdXIcJq0C8GpyrF8X7B3DTjMd_WGCjogax7xU state_group=10 create_new_client_event m.room.message event=$EnTIM5rEGVezQJiYl62uFBl6kJ7B-sMxWqe2D_4FX1I state_group=10 create_new_client_event m.room.message event=$LGx5jGONnBPuNhAuZqHeEoXChd9ryVkuTZatGisOPjk state_group=10 create_new_client_event m.room.message event=$wW0zwoN50lbLu1KoKbybVMxLbKUj7GV_olozIc5i3M0 state_group=10 create_new_client_event org.matrix.msc2716.batch event=$5ZB6dtzqFBCEuMRgpkU201Qhx3WtXZGTz_YgldL6JrQ state_group=10 ```
2021-10-13 18:44:00 -04:00
# Connect all the state in a floating chain
Fix historical messages backfilling in random order on remote homeservers (MSC2716) (#11114) Fix https://github.com/matrix-org/synapse/issues/11091 Fix https://github.com/matrix-org/synapse/issues/10764 (side-stepping the issue because we no longer have to deal with `fake_prev_event_id`) 1. Made the `/backfill` response return messages in `(depth, stream_ordering)` order (previously only sorted by `depth`) - Technically, it shouldn't really matter how `/backfill` returns things but I'm just trying to make the `stream_ordering` a little more consistent from the origin to the remote homeservers in order to get the order of messages from `/messages` consistent ([sorted by `(topological_ordering, stream_ordering)`](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)). - Even now that we return backfilled messages in order, it still doesn't guarantee the same `stream_ordering` (and more importantly the [`/messages` order](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)) on the other server. For example, if a room has a bunch of history imported and someone visits a permalink to a historical message back in time, their homeserver will skip over the historical messages in between and insert the permalink as the next message in the `stream_order` and totally throw off the sort. - This will be even more the case when we add the [MSC3030 jump to date API endpoint](https://github.com/matrix-org/matrix-doc/pull/3030) so the static archives can navigate and jump to a certain date. - We're solving this in the future by switching to [online topological ordering](https://github.com/matrix-org/gomatrixserverlib/issues/187) and [chunking](https://github.com/matrix-org/synapse/issues/3785) which by its nature will apply retroactively to fix any inconsistencies introduced by people permalinking 2. As we're navigating `prev_events` to return in `/backfill`, we order by `depth` first (newest -> oldest) and now also tie-break based on the `stream_ordering` (newest -> oldest). This is technically important because MSC2716 inserts a bunch of historical messages at the same `depth` so it's best to be prescriptive about which ones we should process first. In reality, I think the code already looped over the historical messages as expected because the database is already in order. 3. Making the historical state chain and historical event chain float on their own by having no `prev_events` instead of a fake `prev_event` which caused backfill to get clogged with an unresolvable event. Fixes https://github.com/matrix-org/synapse/issues/11091 and https://github.com/matrix-org/synapse/issues/10764 4. We no longer find connected insertion events by finding a potential `prev_event` connection to the current event we're iterating over. We now solely rely on marker events which when processed, add the insertion event as an extremity and the federating homeserver can ask about it when time calls. - Related discussion, https://github.com/matrix-org/synapse/pull/11114#discussion_r741514793 Before | After --- | --- ![](https://user-images.githubusercontent.com/558581/139218681-b465c862-5c49-4702-a59e-466733b0cf45.png) | ![](https://user-images.githubusercontent.com/558581/146453159-a1609e0a-8324-439d-ae44-e4bce43ac6d1.png) #### Why aren't we sorting topologically when receiving backfill events? > The main reason we're going to opt to not sort topologically when receiving backfill events is because it's probably best to do whatever is easiest to make it just work. People will probably have opinions once they look at [MSC2716](https://github.com/matrix-org/matrix-doc/pull/2716) which could change whatever implementation anyway. > > As mentioned, ideally we would do this but code necessary to make the fake edges but it gets confusing and gives an impression of “just whyyyy” (feels icky). This problem also dissolves with online topological ordering. > > -- https://github.com/matrix-org/synapse/pull/11114#discussion_r741517138 See https://github.com/matrix-org/synapse/pull/11114#discussion_r739610091 for the technical difficulties
2022-02-07 16:54:13 -05:00
prev_event_ids_for_state_chain = [event_id]
return state_event_ids_at_start
async def persist_historical_events(
self,
events_to_create: List[JsonDict],
room_id: str,
inherited_depth: int,
initial_state_event_ids: List[str],
app_service_requester: Requester,
) -> List[str]:
"""Create and persists all events provided sequentially. Handles the
complexity of creating events in chronological order so they can
reference each other by prev_event but still persists in
reverse-chronoloical order so they have the correct
(topological_ordering, stream_ordering) and sort correctly from
/messages.
Args:
events_to_create: List of historical events to create in JSON
dictionary format.
room_id: Room where you want the events persisted in.
inherited_depth: The depth to create the events at (you will
probably by calling inherit_depth_from_prev_ids(...)).
initial_state_event_ids:
This is used to set explicit state for the insertion event at
the start of the historical batch since it's floating with no
prev_events to derive state from automatically.
app_service_requester: The requester of an application service.
Returns:
List of persisted event IDs
"""
assert app_service_requester.app_service
# We expect the first event in a historical batch to be an insertion event
assert events_to_create[0]["type"] == EventTypes.MSC2716_INSERTION
# We expect the last event in a historical batch to be an batch event
assert events_to_create[-1]["type"] == EventTypes.MSC2716_BATCH
Fix historical messages backfilling in random order on remote homeservers (MSC2716) (#11114) Fix https://github.com/matrix-org/synapse/issues/11091 Fix https://github.com/matrix-org/synapse/issues/10764 (side-stepping the issue because we no longer have to deal with `fake_prev_event_id`) 1. Made the `/backfill` response return messages in `(depth, stream_ordering)` order (previously only sorted by `depth`) - Technically, it shouldn't really matter how `/backfill` returns things but I'm just trying to make the `stream_ordering` a little more consistent from the origin to the remote homeservers in order to get the order of messages from `/messages` consistent ([sorted by `(topological_ordering, stream_ordering)`](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)). - Even now that we return backfilled messages in order, it still doesn't guarantee the same `stream_ordering` (and more importantly the [`/messages` order](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)) on the other server. For example, if a room has a bunch of history imported and someone visits a permalink to a historical message back in time, their homeserver will skip over the historical messages in between and insert the permalink as the next message in the `stream_order` and totally throw off the sort. - This will be even more the case when we add the [MSC3030 jump to date API endpoint](https://github.com/matrix-org/matrix-doc/pull/3030) so the static archives can navigate and jump to a certain date. - We're solving this in the future by switching to [online topological ordering](https://github.com/matrix-org/gomatrixserverlib/issues/187) and [chunking](https://github.com/matrix-org/synapse/issues/3785) which by its nature will apply retroactively to fix any inconsistencies introduced by people permalinking 2. As we're navigating `prev_events` to return in `/backfill`, we order by `depth` first (newest -> oldest) and now also tie-break based on the `stream_ordering` (newest -> oldest). This is technically important because MSC2716 inserts a bunch of historical messages at the same `depth` so it's best to be prescriptive about which ones we should process first. In reality, I think the code already looped over the historical messages as expected because the database is already in order. 3. Making the historical state chain and historical event chain float on their own by having no `prev_events` instead of a fake `prev_event` which caused backfill to get clogged with an unresolvable event. Fixes https://github.com/matrix-org/synapse/issues/11091 and https://github.com/matrix-org/synapse/issues/10764 4. We no longer find connected insertion events by finding a potential `prev_event` connection to the current event we're iterating over. We now solely rely on marker events which when processed, add the insertion event as an extremity and the federating homeserver can ask about it when time calls. - Related discussion, https://github.com/matrix-org/synapse/pull/11114#discussion_r741514793 Before | After --- | --- ![](https://user-images.githubusercontent.com/558581/139218681-b465c862-5c49-4702-a59e-466733b0cf45.png) | ![](https://user-images.githubusercontent.com/558581/146453159-a1609e0a-8324-439d-ae44-e4bce43ac6d1.png) #### Why aren't we sorting topologically when receiving backfill events? > The main reason we're going to opt to not sort topologically when receiving backfill events is because it's probably best to do whatever is easiest to make it just work. People will probably have opinions once they look at [MSC2716](https://github.com/matrix-org/matrix-doc/pull/2716) which could change whatever implementation anyway. > > As mentioned, ideally we would do this but code necessary to make the fake edges but it gets confusing and gives an impression of “just whyyyy” (feels icky). This problem also dissolves with online topological ordering. > > -- https://github.com/matrix-org/synapse/pull/11114#discussion_r741517138 See https://github.com/matrix-org/synapse/pull/11114#discussion_r739610091 for the technical difficulties
2022-02-07 16:54:13 -05:00
# Make the historical event chain float off on its own by specifying no
# prev_events for the first event in the chain which causes the HS to
# ask for the state at the start of the batch later.
prev_event_ids: List[str] = []
event_ids = []
events_to_persist = []
Fix historical messages backfilling in random order on remote homeservers (MSC2716) (#11114) Fix https://github.com/matrix-org/synapse/issues/11091 Fix https://github.com/matrix-org/synapse/issues/10764 (side-stepping the issue because we no longer have to deal with `fake_prev_event_id`) 1. Made the `/backfill` response return messages in `(depth, stream_ordering)` order (previously only sorted by `depth`) - Technically, it shouldn't really matter how `/backfill` returns things but I'm just trying to make the `stream_ordering` a little more consistent from the origin to the remote homeservers in order to get the order of messages from `/messages` consistent ([sorted by `(topological_ordering, stream_ordering)`](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)). - Even now that we return backfilled messages in order, it still doesn't guarantee the same `stream_ordering` (and more importantly the [`/messages` order](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)) on the other server. For example, if a room has a bunch of history imported and someone visits a permalink to a historical message back in time, their homeserver will skip over the historical messages in between and insert the permalink as the next message in the `stream_order` and totally throw off the sort. - This will be even more the case when we add the [MSC3030 jump to date API endpoint](https://github.com/matrix-org/matrix-doc/pull/3030) so the static archives can navigate and jump to a certain date. - We're solving this in the future by switching to [online topological ordering](https://github.com/matrix-org/gomatrixserverlib/issues/187) and [chunking](https://github.com/matrix-org/synapse/issues/3785) which by its nature will apply retroactively to fix any inconsistencies introduced by people permalinking 2. As we're navigating `prev_events` to return in `/backfill`, we order by `depth` first (newest -> oldest) and now also tie-break based on the `stream_ordering` (newest -> oldest). This is technically important because MSC2716 inserts a bunch of historical messages at the same `depth` so it's best to be prescriptive about which ones we should process first. In reality, I think the code already looped over the historical messages as expected because the database is already in order. 3. Making the historical state chain and historical event chain float on their own by having no `prev_events` instead of a fake `prev_event` which caused backfill to get clogged with an unresolvable event. Fixes https://github.com/matrix-org/synapse/issues/11091 and https://github.com/matrix-org/synapse/issues/10764 4. We no longer find connected insertion events by finding a potential `prev_event` connection to the current event we're iterating over. We now solely rely on marker events which when processed, add the insertion event as an extremity and the federating homeserver can ask about it when time calls. - Related discussion, https://github.com/matrix-org/synapse/pull/11114#discussion_r741514793 Before | After --- | --- ![](https://user-images.githubusercontent.com/558581/139218681-b465c862-5c49-4702-a59e-466733b0cf45.png) | ![](https://user-images.githubusercontent.com/558581/146453159-a1609e0a-8324-439d-ae44-e4bce43ac6d1.png) #### Why aren't we sorting topologically when receiving backfill events? > The main reason we're going to opt to not sort topologically when receiving backfill events is because it's probably best to do whatever is easiest to make it just work. People will probably have opinions once they look at [MSC2716](https://github.com/matrix-org/matrix-doc/pull/2716) which could change whatever implementation anyway. > > As mentioned, ideally we would do this but code necessary to make the fake edges but it gets confusing and gives an impression of “just whyyyy” (feels icky). This problem also dissolves with online topological ordering. > > -- https://github.com/matrix-org/synapse/pull/11114#discussion_r741517138 See https://github.com/matrix-org/synapse/pull/11114#discussion_r739610091 for the technical difficulties
2022-02-07 16:54:13 -05:00
for index, ev in enumerate(events_to_create):
assert_params_in_dict(ev, ["type", "origin_server_ts", "content", "sender"])
Resolve and share `state_groups` for all historical events in batch (MSC2716) (#10975) Resolve and share `state_groups` for all historical events in batch. This also helps for showing the appropriate avatar/displayname in Element and will work whenever `/messages` has one of the historical messages as the first message in the batch. This does have the flaw where if you just insert a single historical event somewhere, it probably won't resolve the state correctly from `/messages` or `/context` since it will grab a non historical event above or below with resolved state which never included the historical state back then. For the same reasions, this also does not work in Element between the transition from actual messages to historical messages. In the Gitter case, this isn't really a problem since all of the historical messages are in one big lump at the beginning of the room. For a future iteration, might be good to look at `/messages` and `/context` to additionally add the `state` for any historical messages in that batch. --- How are the `state_groups` shared? To illustrate the `state_group` sharing, see this example: **Before** (new `state_group` for every event 😬, very inefficient): ``` # Tests from https://github.com/matrix-org/complement/pull/206 $ COMPLEMENT_ALWAYS_PRINT_SERVER_LOGS=1 COMPLEMENT_DIR=../complement ./scripts-dev/complement.sh TestBackfillingHistory/parallel/should_resolve_member_state_events_for_historical_events create_new_client_event m.room.member event=$_JXfwUDIWS6xKGG4SmZXjSFrizhARM7QblhATVWWUcA state_group=None create_new_client_event org.matrix.msc2716.insertion event=$1ZBfmBKEjg94d-vGYymKrVYeghwBOuGJ3wubU1-I9y0 state_group=9 create_new_client_event org.matrix.msc2716.insertion event=$Mq2JvRetTyclPuozRI682SAjYp3GqRuPc8_cH5-ezPY state_group=10 create_new_client_event m.room.message event=$MfmY4rBQkxrIp8jVwVMTJ4PKnxSigpG9E2cn7S0AtTo state_group=11 create_new_client_event m.room.message event=$uYOv6V8wiF7xHwOMt-60d1AoOIbqLgrDLz6ZIQDdWUI state_group=12 create_new_client_event m.room.message event=$PAbkJRMxb0bX4A6av463faiAhxkE3FEObM1xB4D0UG4 state_group=13 create_new_client_event org.matrix.msc2716.batch event=$Oy_S7AWN7rJQe_MYwGPEy6RtbYklrI-tAhmfiLrCaKI state_group=14 ``` **After** (all events in batch sharing `state_group=10`) (the base insertion event has `state_group=8` which matches the `prev_event` we're inserting next to): ``` # Tests from https://github.com/matrix-org/complement/pull/206 $ COMPLEMENT_ALWAYS_PRINT_SERVER_LOGS=1 COMPLEMENT_DIR=../complement ./scripts-dev/complement.sh TestBackfillingHistory/parallel/should_resolve_member_state_events_for_historical_events create_new_client_event m.room.member event=$PWomJ8PwENYEYuVNoG30gqtybuQQSZ55eldBUSs0i0U state_group=None create_new_client_event org.matrix.msc2716.insertion event=$e_mCU7Eah9ABF6nQU7lu4E1RxIWccNF05AKaTT5m3lw state_group=9 create_new_client_event org.matrix.msc2716.insertion event=$ui7A3_GdXIcJq0C8GpyrF8X7B3DTjMd_WGCjogax7xU state_group=10 create_new_client_event m.room.message event=$EnTIM5rEGVezQJiYl62uFBl6kJ7B-sMxWqe2D_4FX1I state_group=10 create_new_client_event m.room.message event=$LGx5jGONnBPuNhAuZqHeEoXChd9ryVkuTZatGisOPjk state_group=10 create_new_client_event m.room.message event=$wW0zwoN50lbLu1KoKbybVMxLbKUj7GV_olozIc5i3M0 state_group=10 create_new_client_event org.matrix.msc2716.batch event=$5ZB6dtzqFBCEuMRgpkU201Qhx3WtXZGTz_YgldL6JrQ state_group=10 ```
2021-10-13 18:44:00 -04:00
assert self.hs.is_mine_id(ev["sender"]), "User must be our own: %s" % (
ev["sender"],
)
event_dict = {
"type": ev["type"],
"origin_server_ts": ev["origin_server_ts"],
"content": ev["content"],
"room_id": room_id,
"sender": ev["sender"], # requester.user.to_string(),
"prev_events": prev_event_ids.copy(),
}
# Mark all events as historical
event_dict["content"][EventContentFields.MSC2716_HISTORICAL] = True
event, context = await self.event_creation_handler.create_event(
await self.create_requester_for_user_id_from_app_service(
ev["sender"], app_service_requester.app_service
),
event_dict,
# Only the first event (which is the insertion event) in the
# chain should be floating. The rest should hang off each other
# in a chain.
Fix historical messages backfilling in random order on remote homeservers (MSC2716) (#11114) Fix https://github.com/matrix-org/synapse/issues/11091 Fix https://github.com/matrix-org/synapse/issues/10764 (side-stepping the issue because we no longer have to deal with `fake_prev_event_id`) 1. Made the `/backfill` response return messages in `(depth, stream_ordering)` order (previously only sorted by `depth`) - Technically, it shouldn't really matter how `/backfill` returns things but I'm just trying to make the `stream_ordering` a little more consistent from the origin to the remote homeservers in order to get the order of messages from `/messages` consistent ([sorted by `(topological_ordering, stream_ordering)`](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)). - Even now that we return backfilled messages in order, it still doesn't guarantee the same `stream_ordering` (and more importantly the [`/messages` order](https://github.com/matrix-org/synapse/blob/develop/docs/development/room-dag-concepts.md#depth-and-stream-ordering)) on the other server. For example, if a room has a bunch of history imported and someone visits a permalink to a historical message back in time, their homeserver will skip over the historical messages in between and insert the permalink as the next message in the `stream_order` and totally throw off the sort. - This will be even more the case when we add the [MSC3030 jump to date API endpoint](https://github.com/matrix-org/matrix-doc/pull/3030) so the static archives can navigate and jump to a certain date. - We're solving this in the future by switching to [online topological ordering](https://github.com/matrix-org/gomatrixserverlib/issues/187) and [chunking](https://github.com/matrix-org/synapse/issues/3785) which by its nature will apply retroactively to fix any inconsistencies introduced by people permalinking 2. As we're navigating `prev_events` to return in `/backfill`, we order by `depth` first (newest -> oldest) and now also tie-break based on the `stream_ordering` (newest -> oldest). This is technically important because MSC2716 inserts a bunch of historical messages at the same `depth` so it's best to be prescriptive about which ones we should process first. In reality, I think the code already looped over the historical messages as expected because the database is already in order. 3. Making the historical state chain and historical event chain float on their own by having no `prev_events` instead of a fake `prev_event` which caused backfill to get clogged with an unresolvable event. Fixes https://github.com/matrix-org/synapse/issues/11091 and https://github.com/matrix-org/synapse/issues/10764 4. We no longer find connected insertion events by finding a potential `prev_event` connection to the current event we're iterating over. We now solely rely on marker events which when processed, add the insertion event as an extremity and the federating homeserver can ask about it when time calls. - Related discussion, https://github.com/matrix-org/synapse/pull/11114#discussion_r741514793 Before | After --- | --- ![](https://user-images.githubusercontent.com/558581/139218681-b465c862-5c49-4702-a59e-466733b0cf45.png) | ![](https://user-images.githubusercontent.com/558581/146453159-a1609e0a-8324-439d-ae44-e4bce43ac6d1.png) #### Why aren't we sorting topologically when receiving backfill events? > The main reason we're going to opt to not sort topologically when receiving backfill events is because it's probably best to do whatever is easiest to make it just work. People will probably have opinions once they look at [MSC2716](https://github.com/matrix-org/matrix-doc/pull/2716) which could change whatever implementation anyway. > > As mentioned, ideally we would do this but code necessary to make the fake edges but it gets confusing and gives an impression of “just whyyyy” (feels icky). This problem also dissolves with online topological ordering. > > -- https://github.com/matrix-org/synapse/pull/11114#discussion_r741517138 See https://github.com/matrix-org/synapse/pull/11114#discussion_r739610091 for the technical difficulties
2022-02-07 16:54:13 -05:00
allow_no_prev_events=index == 0,
prev_event_ids=event_dict.get("prev_events"),
# Since the first event (which is the insertion event) in the
# chain is floating with no `prev_events`, it can't derive state
# from anywhere automatically. So we need to set some state
# explicitly.
state_event_ids=initial_state_event_ids if index == 0 else None,
historical=True,
depth=inherited_depth,
)
Resolve and share `state_groups` for all historical events in batch (MSC2716) (#10975) Resolve and share `state_groups` for all historical events in batch. This also helps for showing the appropriate avatar/displayname in Element and will work whenever `/messages` has one of the historical messages as the first message in the batch. This does have the flaw where if you just insert a single historical event somewhere, it probably won't resolve the state correctly from `/messages` or `/context` since it will grab a non historical event above or below with resolved state which never included the historical state back then. For the same reasions, this also does not work in Element between the transition from actual messages to historical messages. In the Gitter case, this isn't really a problem since all of the historical messages are in one big lump at the beginning of the room. For a future iteration, might be good to look at `/messages` and `/context` to additionally add the `state` for any historical messages in that batch. --- How are the `state_groups` shared? To illustrate the `state_group` sharing, see this example: **Before** (new `state_group` for every event 😬, very inefficient): ``` # Tests from https://github.com/matrix-org/complement/pull/206 $ COMPLEMENT_ALWAYS_PRINT_SERVER_LOGS=1 COMPLEMENT_DIR=../complement ./scripts-dev/complement.sh TestBackfillingHistory/parallel/should_resolve_member_state_events_for_historical_events create_new_client_event m.room.member event=$_JXfwUDIWS6xKGG4SmZXjSFrizhARM7QblhATVWWUcA state_group=None create_new_client_event org.matrix.msc2716.insertion event=$1ZBfmBKEjg94d-vGYymKrVYeghwBOuGJ3wubU1-I9y0 state_group=9 create_new_client_event org.matrix.msc2716.insertion event=$Mq2JvRetTyclPuozRI682SAjYp3GqRuPc8_cH5-ezPY state_group=10 create_new_client_event m.room.message event=$MfmY4rBQkxrIp8jVwVMTJ4PKnxSigpG9E2cn7S0AtTo state_group=11 create_new_client_event m.room.message event=$uYOv6V8wiF7xHwOMt-60d1AoOIbqLgrDLz6ZIQDdWUI state_group=12 create_new_client_event m.room.message event=$PAbkJRMxb0bX4A6av463faiAhxkE3FEObM1xB4D0UG4 state_group=13 create_new_client_event org.matrix.msc2716.batch event=$Oy_S7AWN7rJQe_MYwGPEy6RtbYklrI-tAhmfiLrCaKI state_group=14 ``` **After** (all events in batch sharing `state_group=10`) (the base insertion event has `state_group=8` which matches the `prev_event` we're inserting next to): ``` # Tests from https://github.com/matrix-org/complement/pull/206 $ COMPLEMENT_ALWAYS_PRINT_SERVER_LOGS=1 COMPLEMENT_DIR=../complement ./scripts-dev/complement.sh TestBackfillingHistory/parallel/should_resolve_member_state_events_for_historical_events create_new_client_event m.room.member event=$PWomJ8PwENYEYuVNoG30gqtybuQQSZ55eldBUSs0i0U state_group=None create_new_client_event org.matrix.msc2716.insertion event=$e_mCU7Eah9ABF6nQU7lu4E1RxIWccNF05AKaTT5m3lw state_group=9 create_new_client_event org.matrix.msc2716.insertion event=$ui7A3_GdXIcJq0C8GpyrF8X7B3DTjMd_WGCjogax7xU state_group=10 create_new_client_event m.room.message event=$EnTIM5rEGVezQJiYl62uFBl6kJ7B-sMxWqe2D_4FX1I state_group=10 create_new_client_event m.room.message event=$LGx5jGONnBPuNhAuZqHeEoXChd9ryVkuTZatGisOPjk state_group=10 create_new_client_event m.room.message event=$wW0zwoN50lbLu1KoKbybVMxLbKUj7GV_olozIc5i3M0 state_group=10 create_new_client_event org.matrix.msc2716.batch event=$5ZB6dtzqFBCEuMRgpkU201Qhx3WtXZGTz_YgldL6JrQ state_group=10 ```
2021-10-13 18:44:00 -04:00
assert context._state_group
# Normally this is done when persisting the event but we have to
# pre-emptively do it here because we create all the events first,
# then persist them in another pass below. And we want to share
# state_groups across the whole batch so this lookup needs to work
# for the next event in the batch in this loop.
await self.store.store_state_group_id_for_event_id(
event_id=event.event_id,
state_group_id=context._state_group,
)
logger.debug(
"RoomBatchSendEventRestServlet inserting event=%s, prev_event_ids=%s",
event,
prev_event_ids,
)
events_to_persist.append((event, context))
event_id = event.event_id
event_ids.append(event_id)
prev_event_ids = [event_id]
# Persist events in reverse-chronological order so they have the
# correct stream_ordering as they are backfilled (which decrements).
# Events are sorted by (topological_ordering, stream_ordering)
# where topological_ordering is just depth.
for (event, context) in reversed(events_to_persist):
await self.event_creation_handler.handle_new_client_event(
await self.create_requester_for_user_id_from_app_service(
event.sender, app_service_requester.app_service
),
events_and_context=[(event, context)],
)
return event_ids
async def handle_batch_of_events(
self,
events_to_create: List[JsonDict],
room_id: str,
batch_id_to_connect_to: str,
inherited_depth: int,
initial_state_event_ids: List[str],
app_service_requester: Requester,
) -> Tuple[List[str], str]:
"""
Handles creating and persisting all of the historical events as well as
insertion and batch meta events to make the batch navigable in the DAG.
Args:
events_to_create: List of historical events to create in JSON
dictionary format.
room_id: Room where you want the events created in.
batch_id_to_connect_to: The batch_id from the insertion event you
want this batch to connect to.
inherited_depth: The depth to create the events at (you will
probably by calling inherit_depth_from_prev_ids(...)).
initial_state_event_ids:
This is used to set explicit state for the insertion event at
the start of the historical batch since it's floating with no
prev_events to derive state from automatically. This should
probably be the state from the `prev_event` defined by
`/batch_send?prev_event_id=$abc` plus the outcome of
`persist_state_events_at_start`
app_service_requester: The requester of an application service.
Returns:
Tuple containing a list of created events and the next_batch_id
"""
# Connect this current batch to the insertion event from the previous batch
last_event_in_batch = events_to_create[-1]
batch_event = {
"type": EventTypes.MSC2716_BATCH,
"sender": app_service_requester.user.to_string(),
"room_id": room_id,
"content": {
EventContentFields.MSC2716_BATCH_ID: batch_id_to_connect_to,
EventContentFields.MSC2716_HISTORICAL: True,
},
# Since the batch event is put at the end of the batch,
# where the newest-in-time event is, copy the origin_server_ts from
# the last event we're inserting
"origin_server_ts": last_event_in_batch["origin_server_ts"],
}
# Add the batch event to the end of the batch (newest-in-time)
events_to_create.append(batch_event)
# Add an "insertion" event to the start of each batch (next to the oldest-in-time
# event in the batch) so the next batch can be connected to this one.
insertion_event = self.create_insertion_event_dict(
sender=app_service_requester.user.to_string(),
room_id=room_id,
# Since the insertion event is put at the start of the batch,
# where the oldest-in-time event is, copy the origin_server_ts from
# the first event we're inserting
origin_server_ts=events_to_create[0]["origin_server_ts"],
)
next_batch_id = insertion_event["content"][
EventContentFields.MSC2716_NEXT_BATCH_ID
]
# Prepend the insertion event to the start of the batch (oldest-in-time)
events_to_create = [insertion_event] + events_to_create
# Create and persist all of the historical events
event_ids = await self.persist_historical_events(
events_to_create=events_to_create,
room_id=room_id,
inherited_depth=inherited_depth,
initial_state_event_ids=initial_state_event_ids,
app_service_requester=app_service_requester,
)
return event_ids, next_batch_id