def test_rwlock(self):
rwlock = ReadWriteLock()
key = object()
ds = [
rwlock.read(key), # 0
rwlock.read(key), # 1
rwlock.write(key), # 2
rwlock.write(key), # 3
rwlock.read(key), # 4
rwlock.read(key), # 5
rwlock.write(key), # 6
]
ds = [defer.ensureDeferred(d) for d in ds]
self._assert_called_before_not_after(ds, 2)
with ds[0].result:
self._assert_called_before_not_after(ds, 2)
self._assert_called_before_not_after(ds, 2)
with ds[1].result:
self._assert_called_before_not_after(ds, 2)
self._assert_called_before_not_after(ds, 3)
with ds[2].result:
self._assert_called_before_not_after(ds, 3)
self._assert_called_before_not_after(ds, 4)
with ds[3].result:
self._assert_called_before_not_after(ds, 4)
self._assert_called_before_not_after(ds, 6)
with ds[5].result:
self._assert_called_before_not_after(ds, 6)
self._assert_called_before_not_after(ds, 6)
with ds[4].result:
self._assert_called_before_not_after(ds, 6)
self._assert_called_before_not_after(ds, 7)
with ds[6].result:
pass
d = defer.ensureDeferred(rwlock.write(key))
self.assertTrue(d.called)
with d.result:
pass
d = defer.ensureDeferred(rwlock.read(key))
self.assertTrue(d.called)
with d.result:
pass
def test_rwlock(self):
rwlock = ReadWriteLock()
key = object()
ds = [
rwlock.read(key), # 0
rwlock.read(key), # 1
rwlock.write(key), # 2
rwlock.write(key), # 3
rwlock.read(key), # 4
rwlock.read(key), # 5
rwlock.write(key), # 6
]
self._assert_called_before_not_after(ds, 2)
with ds[0].result:
self._assert_called_before_not_after(ds, 2)
self._assert_called_before_not_after(ds, 2)
with ds[1].result:
self._assert_called_before_not_after(ds, 2)
self._assert_called_before_not_after(ds, 3)
with ds[2].result:
self._assert_called_before_not_after(ds, 3)
self._assert_called_before_not_after(ds, 4)
with ds[3].result:
self._assert_called_before_not_after(ds, 4)
self._assert_called_before_not_after(ds, 6)
with ds[5].result:
self._assert_called_before_not_after(ds, 6)
self._assert_called_before_not_after(ds, 6)
with ds[4].result:
self._assert_called_before_not_after(ds, 6)
self._assert_called_before_not_after(ds, 7)
with ds[6].result:
pass
d = rwlock.write(key)
self.assertTrue(d.called)
with d.result:
pass
d = rwlock.read(key)
self.assertTrue(d.called)
with d.result:
pass
class PaginationHandler(object):
"""Handles pagination and purge history requests.
These are in the same handler due to the fact we need to block clients
paginating during a purge.
"""
def __init__(self, hs):
self.hs = hs
self.auth = hs.get_auth()
self.store = hs.get_datastore()
self.clock = hs.get_clock()
self.pagination_lock = ReadWriteLock()
self._purges_in_progress_by_room = set()
# map from purge id to PurgeStatus
self._purges_by_id = {}
def start_purge_history(self, room_id, token, delete_local_events=False):
"""Start off a history purge on a room.
Args:
room_id (str): The room to purge from
token (str): topological token to delete events before
delete_local_events (bool): True to delete local events as well as
remote ones
Returns:
str: unique ID for this purge transaction.
"""
if room_id in self._purges_in_progress_by_room:
raise SynapseError(
400,
"History purge already in progress for %s" % (room_id, ),
)
purge_id = random_string(16)
# we log the purge_id here so that it can be tied back to the
# request id in the log lines.
logger.info("[purge] starting purge_id %s", purge_id)
self._purges_by_id[purge_id] = PurgeStatus()
run_in_background(
self._purge_history,
purge_id,
room_id,
token,
delete_local_events,
)
return purge_id
@defer.inlineCallbacks
def _purge_history(self, purge_id, room_id, token, delete_local_events):
"""Carry out a history purge on a room.
Args:
purge_id (str): The id for this purge
room_id (str): The room to purge from
token (str): topological token to delete events before
delete_local_events (bool): True to delete local events as well as
remote ones
Returns:
Deferred
"""
self._purges_in_progress_by_room.add(room_id)
try:
with (yield self.pagination_lock.write(room_id)):
yield self.store.purge_history(
room_id,
token,
delete_local_events,
)
logger.info("[purge] complete")
self._purges_by_id[purge_id].status = PurgeStatus.STATUS_COMPLETE
except Exception:
f = Failure()
logger.error(
"[purge] failed",
exc_info=(f.type, f.value, f.getTracebackObject()),
)
self._purges_by_id[purge_id].status = PurgeStatus.STATUS_FAILED
finally:
self._purges_in_progress_by_room.discard(room_id)
# remove the purge from the list 24 hours after it completes
def clear_purge():
del self._purges_by_id[purge_id]
self.hs.get_reactor().callLater(24 * 3600, clear_purge)
def get_purge_status(self, purge_id):
"""Get the current status of an active purge
Args:
purge_id (str): purge_id returned by start_purge_history
Returns:
PurgeStatus|None
"""
return self._purges_by_id.get(purge_id)
@defer.inlineCallbacks
def get_messages(self,
requester,
room_id=None,
pagin_config=None,
as_client_event=True,
event_filter=None):
"""Get messages in a room.
Args:
requester (Requester): The user requesting messages.
room_id (str): The room they want messages from.
pagin_config (synapse.api.streams.PaginationConfig): The pagination
config rules to apply, if any.
as_client_event (bool): True to get events in client-server format.
event_filter (Filter): Filter to apply to results or None
Returns:
dict: Pagination API results
"""
user_id = requester.user.to_string()
if pagin_config.from_token:
room_token = pagin_config.from_token.room_key
else:
pagin_config.from_token = (
yield self.hs.get_event_sources().get_current_token_for_room(
room_id=room_id))
room_token = pagin_config.from_token.room_key
room_token = RoomStreamToken.parse(room_token)
pagin_config.from_token = pagin_config.from_token.copy_and_replace(
"room_key", str(room_token))
source_config = pagin_config.get_source_config("room")
with (yield self.pagination_lock.read(room_id)):
membership, member_event_id = yield self.auth.check_in_room_or_world_readable(
room_id, user_id)
if source_config.direction == 'b':
# if we're going backwards, we might need to backfill. This
# requires that we have a topo token.
if room_token.topological:
max_topo = room_token.topological
else:
max_topo = yield self.store.get_max_topological_token(
room_id, room_token.stream)
if membership == Membership.LEAVE:
# If they have left the room then clamp the token to be before
# they left the room, to save the effort of loading from the
# database.
leave_token = yield self.store.get_topological_token_for_event(
member_event_id)
leave_token = RoomStreamToken.parse(leave_token)
if leave_token.topological < max_topo:
source_config.from_key = str(leave_token)
yield self.hs.get_handlers().federation_handler.maybe_backfill(
room_id, max_topo)
events, next_key = yield self.store.paginate_room_events(
room_id=room_id,
from_key=source_config.from_key,
to_key=source_config.to_key,
direction=source_config.direction,
limit=source_config.limit,
event_filter=event_filter,
)
next_token = pagin_config.from_token.copy_and_replace(
"room_key", next_key)
if events:
if event_filter:
events = event_filter.filter(events)
events = yield filter_events_for_client(
self.store,
user_id,
events,
is_peeking=(member_event_id is None),
)
if not events:
defer.returnValue({
"chunk": [],
"start": pagin_config.from_token.to_string(),
"end": next_token.to_string(),
})
state = None
if event_filter and event_filter.lazy_load_members(
) and len(events) > 0:
# TODO: remove redundant members
# FIXME: we also care about invite targets etc.
state_filter = StateFilter.from_types(
(EventTypes.Member, event.sender) for event in events)
state_ids = yield self.store.get_state_ids_for_event(
events[0].event_id,
state_filter=state_filter,
)
if state_ids:
state = yield self.store.get_events(list(state_ids.values()))
state = state.values()
time_now = self.clock.time_msec()
chunk = {
"chunk":
[serialize_event(e, time_now, as_client_event) for e in events],
"start":
pagin_config.from_token.to_string(),
"end":
next_token.to_string(),
}
if state:
chunk["state"] = [
serialize_event(e, time_now, as_client_event) for e in state
]
defer.returnValue(chunk)
class PaginationHandler(object):
"""Handles pagination and purge history requests.
These are in the same handler due to the fact we need to block clients
paginating during a purge.
"""
def __init__(self, hs):
self.hs = hs
self.auth = hs.get_auth()
self.store = hs.get_datastore()
self.clock = hs.get_clock()
self.pagination_lock = ReadWriteLock()
self._purges_in_progress_by_room = set()
# map from purge id to PurgeStatus
self._purges_by_id = {}
def start_purge_history(self, room_id, token,
delete_local_events=False):
"""Start off a history purge on a room.
Args:
room_id (str): The room to purge from
token (str): topological token to delete events before
delete_local_events (bool): True to delete local events as well as
remote ones
Returns:
str: unique ID for this purge transaction.
"""
if room_id in self._purges_in_progress_by_room:
raise SynapseError(
400,
"History purge already in progress for %s" % (room_id, ),
)
purge_id = random_string(16)
# we log the purge_id here so that it can be tied back to the
# request id in the log lines.
logger.info("[purge] starting purge_id %s", purge_id)
self._purges_by_id[purge_id] = PurgeStatus()
run_in_background(
self._purge_history,
purge_id, room_id, token, delete_local_events,
)
return purge_id
@defer.inlineCallbacks
def _purge_history(self, purge_id, room_id, token,
delete_local_events):
"""Carry out a history purge on a room.
Args:
purge_id (str): The id for this purge
room_id (str): The room to purge from
token (str): topological token to delete events before
delete_local_events (bool): True to delete local events as well as
remote ones
Returns:
Deferred
"""
self._purges_in_progress_by_room.add(room_id)
try:
with (yield self.pagination_lock.write(room_id)):
yield self.store.purge_history(
room_id, token, delete_local_events,
)
logger.info("[purge] complete")
self._purges_by_id[purge_id].status = PurgeStatus.STATUS_COMPLETE
except Exception:
logger.error("[purge] failed: %s", Failure().getTraceback().rstrip())
self._purges_by_id[purge_id].status = PurgeStatus.STATUS_FAILED
finally:
self._purges_in_progress_by_room.discard(room_id)
# remove the purge from the list 24 hours after it completes
def clear_purge():
del self._purges_by_id[purge_id]
self.hs.get_reactor().callLater(24 * 3600, clear_purge)
def get_purge_status(self, purge_id):
"""Get the current status of an active purge
Args:
purge_id (str): purge_id returned by start_purge_history
Returns:
PurgeStatus|None
"""
return self._purges_by_id.get(purge_id)
@defer.inlineCallbacks
def get_messages(self, requester, room_id=None, pagin_config=None,
as_client_event=True, event_filter=None):
"""Get messages in a room.
Args:
requester (Requester): The user requesting messages.
room_id (str): The room they want messages from.
pagin_config (synapse.api.streams.PaginationConfig): The pagination
config rules to apply, if any.
as_client_event (bool): True to get events in client-server format.
event_filter (Filter): Filter to apply to results or None
Returns:
dict: Pagination API results
"""
user_id = requester.user.to_string()
if pagin_config.from_token:
room_token = pagin_config.from_token.room_key
else:
pagin_config.from_token = (
yield self.hs.get_event_sources().get_current_token_for_room(
room_id=room_id
)
)
room_token = pagin_config.from_token.room_key
room_token = RoomStreamToken.parse(room_token)
pagin_config.from_token = pagin_config.from_token.copy_and_replace(
"room_key", str(room_token)
)
source_config = pagin_config.get_source_config("room")
with (yield self.pagination_lock.read(room_id)):
membership, member_event_id = yield self.auth.check_in_room_or_world_readable(
room_id, user_id
)
if source_config.direction == 'b':
# if we're going backwards, we might need to backfill. This
# requires that we have a topo token.
if room_token.topological:
max_topo = room_token.topological
else:
max_topo = yield self.store.get_max_topological_token(
room_id, room_token.stream
)
if membership == Membership.LEAVE:
# If they have left the room then clamp the token to be before
# they left the room, to save the effort of loading from the
# database.
leave_token = yield self.store.get_topological_token_for_event(
member_event_id
)
leave_token = RoomStreamToken.parse(leave_token)
if leave_token.topological < max_topo:
source_config.from_key = str(leave_token)
yield self.hs.get_handlers().federation_handler.maybe_backfill(
room_id, max_topo
)
events, next_key = yield self.store.paginate_room_events(
room_id=room_id,
from_key=source_config.from_key,
to_key=source_config.to_key,
direction=source_config.direction,
limit=source_config.limit,
event_filter=event_filter,
)
next_token = pagin_config.from_token.copy_and_replace(
"room_key", next_key
)
if events:
if event_filter:
events = event_filter.filter(events)
events = yield filter_events_for_client(
self.store,
user_id,
events,
is_peeking=(member_event_id is None),
)
if not events:
defer.returnValue({
"chunk": [],
"start": pagin_config.from_token.to_string(),
"end": next_token.to_string(),
})
state = None
if event_filter and event_filter.lazy_load_members():
# TODO: remove redundant members
# FIXME: we also care about invite targets etc.
state_filter = StateFilter.from_types(
(EventTypes.Member, event.sender)
for event in events
)
state_ids = yield self.store.get_state_ids_for_event(
events[0].event_id, state_filter=state_filter,
)
if state_ids:
state = yield self.store.get_events(list(state_ids.values()))
state = state.values()
time_now = self.clock.time_msec()
chunk = {
"chunk": [
serialize_event(e, time_now, as_client_event)
for e in events
],
"start": pagin_config.from_token.to_string(),
"end": next_token.to_string(),
}
if state:
chunk["state"] = [
serialize_event(e, time_now, as_client_event)
for e in state
]
defer.returnValue(chunk)
class PaginationHandler(object):
"""Handles pagination and purge history requests.
These are in the same handler due to the fact we need to block clients
paginating during a purge.
"""
def __init__(self, hs):
self.hs = hs
self.auth = hs.get_auth()
self.store = hs.get_datastore()
self.storage = hs.get_storage()
self.state_store = self.storage.state
self.clock = hs.get_clock()
self._server_name = hs.hostname
self.pagination_lock = ReadWriteLock()
self._purges_in_progress_by_room = set()
# map from purge id to PurgeStatus
self._purges_by_id = {}
self._event_serializer = hs.get_event_client_serializer()
self._retention_default_max_lifetime = hs.config.retention_default_max_lifetime
if hs.config.retention_enabled:
# Run the purge jobs described in the configuration file.
for job in hs.config.retention_purge_jobs:
logger.info("Setting up purge job with config: %s", job)
self.clock.looping_call(
run_as_background_process,
job["interval"],
"purge_history_for_rooms_in_range",
self.purge_history_for_rooms_in_range,
job["shortest_max_lifetime"],
job["longest_max_lifetime"],
)
@defer.inlineCallbacks
def purge_history_for_rooms_in_range(self, min_ms, max_ms):
"""Purge outdated events from rooms within the given retention range.
If a default retention policy is defined in the server's configuration and its
'max_lifetime' is within this range, also targets rooms which don't have a
retention policy.
Args:
min_ms (int|None): Duration in milliseconds that define the lower limit of
the range to handle (exclusive). If None, it means that the range has no
lower limit.
max_ms (int|None): Duration in milliseconds that define the upper limit of
the range to handle (inclusive). If None, it means that the range has no
upper limit.
"""
# We want the storage layer to to include rooms with no retention policy in its
# return value only if a default retention policy is defined in the server's
# configuration and that policy's 'max_lifetime' is either lower (or equal) than
# max_ms or higher than min_ms (or both).
if self._retention_default_max_lifetime is not None:
include_null = True
if min_ms is not None and min_ms >= self._retention_default_max_lifetime:
# The default max_lifetime is lower than (or equal to) min_ms.
include_null = False
if max_ms is not None and max_ms < self._retention_default_max_lifetime:
# The default max_lifetime is higher than max_ms.
include_null = False
else:
include_null = False
logger.info(
"[purge] Running purge job for %d < max_lifetime <= %d (include NULLs = %s)",
min_ms,
max_ms,
include_null,
)
rooms = yield self.store.get_rooms_for_retention_period_in_range(
min_ms, max_ms, include_null)
logger.debug("[purge] Rooms to purge: %s", rooms)
for room_id, retention_policy in iteritems(rooms):
logger.info("[purge] Attempting to purge messages in room %s",
room_id)
if room_id in self._purges_in_progress_by_room:
logger.warning(
"[purge] not purging room %s as there's an ongoing purge running"
" for this room",
room_id,
)
continue
max_lifetime = retention_policy["max_lifetime"]
if max_lifetime is None:
# If max_lifetime is None, it means that include_null equals True,
# therefore we can safely assume that there is a default policy defined
# in the server's configuration.
max_lifetime = self._retention_default_max_lifetime
# Figure out what token we should start purging at.
ts = self.clock.time_msec() - max_lifetime
stream_ordering = yield self.store.find_first_stream_ordering_after_ts(
ts)
r = yield self.store.get_room_event_before_stream_ordering(
room_id,
stream_ordering,
)
if not r:
logger.warning(
"[purge] purging events not possible: No event found "
"(ts %i => stream_ordering %i)",
ts,
stream_ordering,
)
continue
(stream, topo, _event_id) = r
token = "t%d-%d" % (topo, stream)
purge_id = random_string(16)
self._purges_by_id[purge_id] = PurgeStatus()
logger.info("Starting purging events in room %s (purge_id %s)" %
(room_id, purge_id))
# We want to purge everything, including local events, and to run the purge in
# the background so that it's not blocking any other operation apart from
# other purges in the same room.
run_as_background_process(
"_purge_history",
self._purge_history,
purge_id,
room_id,
token,
True,
)
def start_purge_history(self, room_id, token, delete_local_events=False):
"""Start off a history purge on a room.
Args:
room_id (str): The room to purge from
token (str): topological token to delete events before
delete_local_events (bool): True to delete local events as well as
remote ones
Returns:
str: unique ID for this purge transaction.
"""
if room_id in self._purges_in_progress_by_room:
raise SynapseError(
400, "History purge already in progress for %s" % (room_id, ))
purge_id = random_string(16)
# we log the purge_id here so that it can be tied back to the
# request id in the log lines.
logger.info("[purge] starting purge_id %s", purge_id)
self._purges_by_id[purge_id] = PurgeStatus()
run_in_background(self._purge_history, purge_id, room_id, token,
delete_local_events)
return purge_id
@defer.inlineCallbacks
def _purge_history(self, purge_id, room_id, token, delete_local_events):
"""Carry out a history purge on a room.
Args:
purge_id (str): The id for this purge
room_id (str): The room to purge from
token (str): topological token to delete events before
delete_local_events (bool): True to delete local events as well as
remote ones
Returns:
Deferred
"""
self._purges_in_progress_by_room.add(room_id)
try:
with (yield self.pagination_lock.write(room_id)):
yield self.storage.purge_events.purge_history(
room_id, token, delete_local_events)
logger.info("[purge] complete")
self._purges_by_id[purge_id].status = PurgeStatus.STATUS_COMPLETE
except Exception:
f = Failure()
logger.error("[purge] failed",
exc_info=(f.type, f.value, f.getTracebackObject()))
self._purges_by_id[purge_id].status = PurgeStatus.STATUS_FAILED
finally:
self._purges_in_progress_by_room.discard(room_id)
# remove the purge from the list 24 hours after it completes
def clear_purge():
del self._purges_by_id[purge_id]
self.hs.get_reactor().callLater(24 * 3600, clear_purge)
def get_purge_status(self, purge_id):
"""Get the current status of an active purge
Args:
purge_id (str): purge_id returned by start_purge_history
Returns:
PurgeStatus|None
"""
return self._purges_by_id.get(purge_id)
async def purge_room(self, room_id):
"""Purge the given room from the database"""
with (await self.pagination_lock.write(room_id)):
# check we know about the room
await self.store.get_room_version(room_id)
# first check that we have no users in this room
joined = await defer.maybeDeferred(self.store.is_host_joined,
room_id, self._server_name)
if joined:
raise SynapseError(400, "Users are still joined to this room")
await self.storage.purge_events.purge_room(room_id)
async def get_messages(
self,
requester,
room_id=None,
pagin_config=None,
as_client_event=True,
event_filter=None,
):
"""Get messages in a room.
Args:
requester (Requester): The user requesting messages.
room_id (str): The room they want messages from.
pagin_config (synapse.api.streams.PaginationConfig): The pagination
config rules to apply, if any.
as_client_event (bool): True to get events in client-server format.
event_filter (Filter): Filter to apply to results or None
Returns:
dict: Pagination API results
"""
user_id = requester.user.to_string()
if pagin_config.from_token:
room_token = pagin_config.from_token.room_key
else:
pagin_config.from_token = (
await
self.hs.get_event_sources().get_current_token_for_pagination())
room_token = pagin_config.from_token.room_key
room_token = RoomStreamToken.parse(room_token)
pagin_config.from_token = pagin_config.from_token.copy_and_replace(
"room_key", str(room_token))
source_config = pagin_config.get_source_config("room")
with (await self.pagination_lock.read(room_id)):
(
membership,
member_event_id,
) = await self.auth.check_in_room_or_world_readable(
room_id, user_id)
if source_config.direction == "b":
# if we're going backwards, we might need to backfill. This
# requires that we have a topo token.
if room_token.topological:
max_topo = room_token.topological
else:
max_topo = await self.store.get_max_topological_token(
room_id, room_token.stream)
if membership == Membership.LEAVE:
# If they have left the room then clamp the token to be before
# they left the room, to save the effort of loading from the
# database.
leave_token = await self.store.get_topological_token_for_event(
member_event_id)
leave_token = RoomStreamToken.parse(leave_token)
if leave_token.topological < max_topo:
source_config.from_key = str(leave_token)
await self.hs.get_handlers().federation_handler.maybe_backfill(
room_id, max_topo)
events, next_key = await self.store.paginate_room_events(
room_id=room_id,
from_key=source_config.from_key,
to_key=source_config.to_key,
direction=source_config.direction,
limit=source_config.limit,
event_filter=event_filter,
)
next_token = pagin_config.from_token.copy_and_replace(
"room_key", next_key)
if events:
if event_filter:
events = event_filter.filter(events)
events = await filter_events_for_client(
self.storage,
user_id,
events,
is_peeking=(member_event_id is None))
if not events:
return {
"chunk": [],
"start": pagin_config.from_token.to_string(),
"end": next_token.to_string(),
}
state = None
if event_filter and event_filter.lazy_load_members(
) and len(events) > 0:
# TODO: remove redundant members
# FIXME: we also care about invite targets etc.
state_filter = StateFilter.from_types(
(EventTypes.Member, event.sender) for event in events)
state_ids = await self.state_store.get_state_ids_for_event(
events[0].event_id, state_filter=state_filter)
if state_ids:
state = await self.store.get_events(list(state_ids.values()))
state = state.values()
time_now = self.clock.time_msec()
chunk = {
"chunk": (await self._event_serializer.serialize_events(
events, time_now, as_client_event=as_client_event)),
"start":
pagin_config.from_token.to_string(),
"end":
next_token.to_string(),
}
if state:
chunk["state"] = await self._event_serializer.serialize_events(
state, time_now, as_client_event=as_client_event)
return chunk
class PaginationHandler:
"""Handles pagination and purge history requests.
These are in the same handler due to the fact we need to block clients
paginating during a purge.
"""
# when to remove a completed deletion/purge from the results map
CLEAR_PURGE_AFTER_MS = 1000 * 3600 * 24 # 24 hours
def __init__(self, hs: "HomeServer"):
self.hs = hs
self.auth = hs.get_auth()
self.store = hs.get_datastores().main
self._storage_controllers = hs.get_storage_controllers()
self._state_storage_controller = self._storage_controllers.state
self.clock = hs.get_clock()
self._server_name = hs.hostname
self._room_shutdown_handler = hs.get_room_shutdown_handler()
self._relations_handler = hs.get_relations_handler()
self.pagination_lock = ReadWriteLock()
# IDs of rooms in which there currently an active purge *or delete* operation.
self._purges_in_progress_by_room: Set[str] = set()
# map from purge id to PurgeStatus
self._purges_by_id: Dict[str, PurgeStatus] = {}
# map from purge id to DeleteStatus
self._delete_by_id: Dict[str, DeleteStatus] = {}
# map from room id to delete ids
# Dict[`room_id`, List[`delete_id`]]
self._delete_by_room: Dict[str, List[str]] = {}
self._event_serializer = hs.get_event_client_serializer()
self._retention_default_max_lifetime = (
hs.config.retention.retention_default_max_lifetime)
self._retention_allowed_lifetime_min = (
hs.config.retention.retention_allowed_lifetime_min)
self._retention_allowed_lifetime_max = (
hs.config.retention.retention_allowed_lifetime_max)
if (hs.config.worker.run_background_tasks
and hs.config.retention.retention_enabled):
# Run the purge jobs described in the configuration file.
for job in hs.config.retention.retention_purge_jobs:
logger.info("Setting up purge job with config: %s", job)
self.clock.looping_call(
run_as_background_process,
job.interval,
"purge_history_for_rooms_in_range",
self.purge_history_for_rooms_in_range,
job.shortest_max_lifetime,
job.longest_max_lifetime,
)
async def purge_history_for_rooms_in_range(self, min_ms: Optional[int],
max_ms: Optional[int]) -> None:
"""Purge outdated events from rooms within the given retention range.
If a default retention policy is defined in the server's configuration and its
'max_lifetime' is within this range, also targets rooms which don't have a
retention policy.
Args:
min_ms: Duration in milliseconds that define the lower limit of
the range to handle (exclusive). If None, it means that the range has no
lower limit.
max_ms: Duration in milliseconds that define the upper limit of
the range to handle (inclusive). If None, it means that the range has no
upper limit.
"""
# We want the storage layer to include rooms with no retention policy in its
# return value only if a default retention policy is defined in the server's
# configuration and that policy's 'max_lifetime' is either lower (or equal) than
# max_ms or higher than min_ms (or both).
if self._retention_default_max_lifetime is not None:
include_null = True
if min_ms is not None and min_ms >= self._retention_default_max_lifetime:
# The default max_lifetime is lower than (or equal to) min_ms.
include_null = False
if max_ms is not None and max_ms < self._retention_default_max_lifetime:
# The default max_lifetime is higher than max_ms.
include_null = False
else:
include_null = False
logger.info(
"[purge] Running purge job for %s < max_lifetime <= %s (include NULLs = %s)",
min_ms,
max_ms,
include_null,
)
rooms = await self.store.get_rooms_for_retention_period_in_range(
min_ms, max_ms, include_null)
logger.debug("[purge] Rooms to purge: %s", rooms)
for room_id, retention_policy in rooms.items():
logger.info("[purge] Attempting to purge messages in room %s",
room_id)
if room_id in self._purges_in_progress_by_room:
logger.warning(
"[purge] not purging room %s as there's an ongoing purge running"
" for this room",
room_id,
)
continue
# If max_lifetime is None, it means that the room has no retention policy.
# Given we only retrieve such rooms when there's a default retention policy
# defined in the server's configuration, we can safely assume that's the
# case and use it for this room.
max_lifetime = (retention_policy.max_lifetime
or self._retention_default_max_lifetime)
# Cap the effective max_lifetime to be within the range allowed in the
# config.
# We do this in two steps:
# 1. Make sure it's higher or equal to the minimum allowed value, and if
# it's not replace it with that value. This is because the server
# operator can be required to not delete information before a given
# time, e.g. to comply with freedom of information laws.
# 2. Make sure the resulting value is lower or equal to the maximum allowed
# value, and if it's not replace it with that value. This is because the
# server operator can be required to delete any data after a specific
# amount of time.
if self._retention_allowed_lifetime_min is not None:
max_lifetime = max(self._retention_allowed_lifetime_min,
max_lifetime)
if self._retention_allowed_lifetime_max is not None:
max_lifetime = min(max_lifetime,
self._retention_allowed_lifetime_max)
logger.debug("[purge] max_lifetime for room %s: %s", room_id,
max_lifetime)
# Figure out what token we should start purging at.
ts = self.clock.time_msec() - max_lifetime
stream_ordering = await self.store.find_first_stream_ordering_after_ts(
ts)
r = await self.store.get_room_event_before_stream_ordering(
room_id,
stream_ordering,
)
if not r:
logger.warning(
"[purge] purging events not possible: No event found "
"(ts %i => stream_ordering %i)",
ts,
stream_ordering,
)
continue
(stream, topo, _event_id) = r
token = "t%d-%d" % (topo, stream)
purge_id = random_string(16)
self._purges_by_id[purge_id] = PurgeStatus()
logger.info("Starting purging events in room %s (purge_id %s)" %
(room_id, purge_id))
# We want to purge everything, including local events, and to run the purge in
# the background so that it's not blocking any other operation apart from
# other purges in the same room.
run_as_background_process(
"_purge_history",
self._purge_history,
purge_id,
room_id,
token,
True,
)
def start_purge_history(self,
room_id: str,
token: str,
delete_local_events: bool = False) -> str:
"""Start off a history purge on a room.
Args:
room_id: The room to purge from
token: topological token to delete events before
delete_local_events: True to delete local events as well as
remote ones
Returns:
unique ID for this purge transaction.
"""
if room_id in self._purges_in_progress_by_room:
raise SynapseError(
400, "History purge already in progress for %s" % (room_id, ))
purge_id = random_string(16)
# we log the purge_id here so that it can be tied back to the
# request id in the log lines.
logger.info("[purge] starting purge_id %s", purge_id)
self._purges_by_id[purge_id] = PurgeStatus()
run_as_background_process(
"purge_history",
self._purge_history,
purge_id,
room_id,
token,
delete_local_events,
)
return purge_id
async def _purge_history(self, purge_id: str, room_id: str, token: str,
delete_local_events: bool) -> None:
"""Carry out a history purge on a room.
Args:
purge_id: The ID for this purge.
room_id: The room to purge from
token: topological token to delete events before
delete_local_events: True to delete local events as well as remote ones
"""
self._purges_in_progress_by_room.add(room_id)
try:
async with self.pagination_lock.write(room_id):
await self._storage_controllers.purge_events.purge_history(
room_id, token, delete_local_events)
logger.info("[purge] complete")
self._purges_by_id[purge_id].status = PurgeStatus.STATUS_COMPLETE
except Exception:
f = Failure()
logger.error(
"[purge] failed",
exc_info=(f.type, f.value, f.getTracebackObject()
) # type: ignore
)
self._purges_by_id[purge_id].status = PurgeStatus.STATUS_FAILED
self._purges_by_id[purge_id].error = f.getErrorMessage()
finally:
self._purges_in_progress_by_room.discard(room_id)
# remove the purge from the list 24 hours after it completes
def clear_purge() -> None:
del self._purges_by_id[purge_id]
self.hs.get_reactor().callLater(
PaginationHandler.CLEAR_PURGE_AFTER_MS / 1000, clear_purge)
def get_purge_status(self, purge_id: str) -> Optional[PurgeStatus]:
"""Get the current status of an active purge
Args:
purge_id: purge_id returned by start_purge_history
"""
return self._purges_by_id.get(purge_id)
def get_delete_status(self, delete_id: str) -> Optional[DeleteStatus]:
"""Get the current status of an active deleting
Args:
delete_id: delete_id returned by start_shutdown_and_purge_room
"""
return self._delete_by_id.get(delete_id)
def get_delete_ids_by_room(self,
room_id: str) -> Optional[Collection[str]]:
"""Get all active delete ids by room
Args:
room_id: room_id that is deleted
"""
return self._delete_by_room.get(room_id)
async def purge_room(self, room_id: str, force: bool = False) -> None:
"""Purge the given room from the database.
This function is part the delete room v1 API.
Args:
room_id: room to be purged
force: set true to skip checking for joined users.
"""
async with self.pagination_lock.write(room_id):
# first check that we have no users in this room
if not force:
joined = await self.store.is_host_joined(
room_id, self._server_name)
if joined:
raise SynapseError(400,
"Users are still joined to this room")
await self._storage_controllers.purge_events.purge_room(room_id)
async def get_messages(
self,
requester: Requester,
room_id: str,
pagin_config: PaginationConfig,
as_client_event: bool = True,
event_filter: Optional[Filter] = None,
) -> JsonDict:
"""Get messages in a room.
Args:
requester: The user requesting messages.
room_id: The room they want messages from.
pagin_config: The pagination config rules to apply, if any.
as_client_event: True to get events in client-server format.
event_filter: Filter to apply to results or None
Returns:
Pagination API results
"""
user_id = requester.user.to_string()
if pagin_config.from_token:
from_token = pagin_config.from_token
else:
from_token = (await self.hs.get_event_sources().
get_current_token_for_pagination(room_id))
# We expect `/messages` to use historic pagination tokens by default but
# `/messages` should still works with live tokens when manually provided.
assert from_token.room_key.topological is not None
if pagin_config.limit is None:
# This shouldn't happen as we've set a default limit before this
# gets called.
raise Exception("limit not set")
room_token = from_token.room_key
async with self.pagination_lock.read(room_id):
(
membership,
member_event_id,
) = await self.auth.check_user_in_room_or_world_readable(
room_id, user_id, allow_departed_users=True)
if pagin_config.direction == "b":
# if we're going backwards, we might need to backfill. This
# requires that we have a topo token.
if room_token.topological:
curr_topo = room_token.topological
else:
curr_topo = await self.store.get_current_topological_token(
room_id, room_token.stream)
if membership == Membership.LEAVE:
# If they have left the room then clamp the token to be before
# they left the room, to save the effort of loading from the
# database.
# This is only None if the room is world_readable, in which
# case "JOIN" would have been returned.
assert member_event_id
leave_token = await self.store.get_topological_token_for_event(
member_event_id)
assert leave_token.topological is not None
if leave_token.topological < curr_topo:
from_token = from_token.copy_and_replace(
StreamKeyType.ROOM, leave_token)
await self.hs.get_federation_handler().maybe_backfill(
room_id,
curr_topo,
limit=pagin_config.limit,
)
to_room_key = None
if pagin_config.to_token:
to_room_key = pagin_config.to_token.room_key
events, next_key = await self.store.paginate_room_events(
room_id=room_id,
from_key=from_token.room_key,
to_key=to_room_key,
direction=pagin_config.direction,
limit=pagin_config.limit,
event_filter=event_filter,
)
next_token = from_token.copy_and_replace(StreamKeyType.ROOM,
next_key)
# if no events are returned from pagination, that implies
# we have reached the end of the available events.
# In that case we do not return end, to tell the client
# there is no need for further queries.
if not events:
return {
"chunk": [],
"start": await from_token.to_string(self.store),
}
if event_filter:
events = await event_filter.filter(events)
events = await filter_events_for_client(
self._storage_controllers,
user_id,
events,
is_peeking=(member_event_id is None),
)
# if after the filter applied there are no more events
# return immediately - but there might be more in next_token batch
if not events:
return {
"chunk": [],
"start": await from_token.to_string(self.store),
"end": await next_token.to_string(self.store),
}
state = None
if event_filter and event_filter.lazy_load_members and len(events) > 0:
# TODO: remove redundant members
# FIXME: we also care about invite targets etc.
state_filter = StateFilter.from_types(
(EventTypes.Member, event.sender) for event in events)
state_ids = await self._state_storage_controller.get_state_ids_for_event(
events[0].event_id, state_filter=state_filter)
if state_ids:
state_dict = await self.store.get_events(
list(state_ids.values()))
state = state_dict.values()
aggregations = await self._relations_handler.get_bundled_aggregations(
events, user_id)
time_now = self.clock.time_msec()
serialize_options = SerializeEventConfig(
as_client_event=as_client_event)
chunk = {
"chunk": (self._event_serializer.serialize_events(
events,
time_now,
config=serialize_options,
bundle_aggregations=aggregations,
)),
"start":
await from_token.to_string(self.store),
"end":
await next_token.to_string(self.store),
}
if state:
chunk["state"] = self._event_serializer.serialize_events(
state, time_now, config=serialize_options)
return chunk
async def _shutdown_and_purge_room(
self,
delete_id: str,
room_id: str,
requester_user_id: str,
new_room_user_id: Optional[str] = None,
new_room_name: Optional[str] = None,
message: Optional[str] = None,
block: bool = False,
purge: bool = True,
force_purge: bool = False,
) -> None:
"""
Shuts down and purges a room.
See `RoomShutdownHandler.shutdown_room` for details of creation of the new room
Args:
delete_id: The ID for this delete.
room_id: The ID of the room to shut down.
requester_user_id:
User who requested the action. Will be recorded as putting the room on the
blocking list.
new_room_user_id:
If set, a new room will be created with this user ID
as the creator and admin, and all users in the old room will be
moved into that room. If not set, no new room will be created
and the users will just be removed from the old room.
new_room_name:
A string representing the name of the room that new users will
be invited to. Defaults to `Content Violation Notification`
message:
A string containing the first message that will be sent as
`new_room_user_id` in the new room. Ideally this will clearly
convey why the original room was shut down.
Defaults to `Sharing illegal content on this server is not
permitted and rooms in violation will be blocked.`
block:
If set to `true`, this room will be added to a blocking list,
preventing future attempts to join the room. Defaults to `false`.
purge:
If set to `true`, purge the given room from the database.
force_purge:
If set to `true`, the room will be purged from database
also if it fails to remove some users from room.
Saves a `RoomShutdownHandler.ShutdownRoomResponse` in `DeleteStatus`:
"""
self._purges_in_progress_by_room.add(room_id)
try:
async with self.pagination_lock.write(room_id):
self._delete_by_id[
delete_id].status = DeleteStatus.STATUS_SHUTTING_DOWN
self._delete_by_id[
delete_id].shutdown_room = await self._room_shutdown_handler.shutdown_room(
room_id=room_id,
requester_user_id=requester_user_id,
new_room_user_id=new_room_user_id,
new_room_name=new_room_name,
message=message,
block=block,
)
self._delete_by_id[
delete_id].status = DeleteStatus.STATUS_PURGING
if purge:
logger.info("starting purge room_id %s", room_id)
# first check that we have no users in this room
if not force_purge:
joined = await self.store.is_host_joined(
room_id, self._server_name)
if joined:
raise SynapseError(
400, "Users are still joined to this room")
await self._storage_controllers.purge_events.purge_room(
room_id)
logger.info("complete")
self._delete_by_id[delete_id].status = DeleteStatus.STATUS_COMPLETE
except Exception:
f = Failure()
logger.error(
"failed",
exc_info=(f.type, f.value,
f.getTracebackObject()), # type: ignore
)
self._delete_by_id[delete_id].status = DeleteStatus.STATUS_FAILED
self._delete_by_id[delete_id].error = f.getErrorMessage()
finally:
self._purges_in_progress_by_room.discard(room_id)
# remove the delete from the list 24 hours after it completes
def clear_delete() -> None:
del self._delete_by_id[delete_id]
self._delete_by_room[room_id].remove(delete_id)
if not self._delete_by_room[room_id]:
del self._delete_by_room[room_id]
self.hs.get_reactor().callLater(
PaginationHandler.CLEAR_PURGE_AFTER_MS / 1000, clear_delete)
def start_shutdown_and_purge_room(
self,
room_id: str,
requester_user_id: str,
new_room_user_id: Optional[str] = None,
new_room_name: Optional[str] = None,
message: Optional[str] = None,
block: bool = False,
purge: bool = True,
force_purge: bool = False,
) -> str:
"""Start off shut down and purge on a room.
Args:
room_id: The ID of the room to shut down.
requester_user_id:
User who requested the action and put the room on the
blocking list.
new_room_user_id:
If set, a new room will be created with this user ID
as the creator and admin, and all users in the old room will be
moved into that room. If not set, no new room will be created
and the users will just be removed from the old room.
new_room_name:
A string representing the name of the room that new users will
be invited to. Defaults to `Content Violation Notification`
message:
A string containing the first message that will be sent as
`new_room_user_id` in the new room. Ideally this will clearly
convey why the original room was shut down.
Defaults to `Sharing illegal content on this server is not
permitted and rooms in violation will be blocked.`
block:
If set to `true`, this room will be added to a blocking list,
preventing future attempts to join the room. Defaults to `false`.
purge:
If set to `true`, purge the given room from the database.
force_purge:
If set to `true`, the room will be purged from database
also if it fails to remove some users from room.
Returns:
unique ID for this delete transaction.
"""
if room_id in self._purges_in_progress_by_room:
raise SynapseError(
400, "History purge already in progress for %s" % (room_id, ))
# This check is double to `RoomShutdownHandler.shutdown_room`
# But here the requester get a direct response / error with HTTP request
# and do not have to check the purge status
if new_room_user_id is not None:
if not self.hs.is_mine_id(new_room_user_id):
raise SynapseError(
400, "User must be our own: %s" % (new_room_user_id, ))
delete_id = random_string(16)
# we log the delete_id here so that it can be tied back to the
# request id in the log lines.
logger.info(
"starting shutdown room_id %s with delete_id %s",
room_id,
delete_id,
)
self._delete_by_id[delete_id] = DeleteStatus()
self._delete_by_room.setdefault(room_id, []).append(delete_id)
run_as_background_process(
"shutdown_and_purge_room",
self._shutdown_and_purge_room,
delete_id,
room_id,
requester_user_id,
new_room_user_id,
new_room_name,
message,
block,
purge,
force_purge,
)
return delete_id