This PR aims to fix#16895, caused by a regression in #7 and not fixed
by #16903. The PR #16903 only fixes a starvation issue, where the CPU
isn't released. There is a second issue, where the execution is blocked.
This theory is supported by the flame graphs provided in #16895 and the
fact that I see the CPU usage reducing and far below the limit.
Since the changes in #7, the method `check_state_independent_auth_rules`
is called with the additional parameter `batched_auth_events`:
6fa13b4f92/synapse/handlers/federation_event.py (L1741-L1743)
It makes the execution enter this if clause, introduced with #151956fa13b4f92/synapse/event_auth.py (L178-L189)
There are two issues in the above code snippet.
First, there is the blocking issue. I'm not entirely sure if this is a
deadlock, starvation, or something different. In the beginning, I
thought the copy operation was responsible. It wasn't. Then I
investigated the nested `store.get_events` inside the function `update`.
This was also not causing the blocking issue. Only when I replaced the
set difference operation (`-` ) with a list comprehension, the blocking
was resolved. Creating and comparing sets with a very large amount of
events seems to be problematic.
This is how the flamegraph looks now while persisting outliers. As you
can see, the execution no longer locks up in the above function.
Second, the copying here doesn't serve any purpose, because only a
shallow copy is created. This means the same objects from the original
dict are referenced. This fails the intention of protecting these
objects from mutation. The review of the original PR
https://github.com/matrix-org/synapse/pull/15195 had an extensive
discussion about this matter.
Various approaches to copying the auth_events were attempted:
1) Implementing a deepcopy caused issues due to
builtins.EventInternalMetadata not being pickleable.
2) Creating a dict with new objects akin to a deepcopy.
3) Creating a dict with new objects containing only necessary
attributes.
Concluding, there is no easy way to create an actual copy of the
objects. Opting for a deepcopy can significantly strain memory and CPU
resources, making it an inefficient choice. I don't see why the copy is
necessary in the first place. Therefore I'm proposing to remove it
altogether.
After these changes, I was able to successfully join these rooms,
without the main worker locking up:
- #synapse:matrix.org
- #element-android:matrix.org
- #element-web:matrix.org
- #ecips:matrix.org
- #ipfs-chatter:ipfs.io
- #python:matrix.org
- #matrix:matrix.org
During the migration the automated script to update the copyright
headers accidentally got rid of some of the existing copyright lines.
Reinstate them.
And fix a bug in the implementation of the updated redaction
format (MSC2174) where the top-level redacts field was not
properly added for backwards-compatibility.
This moves `redacts` from being a top-level property to
a `content` property in a new room version.
MSC2176 (which was previously implemented) states to not
`redact` this property.
* Better test for bad values in power levels events
The previous test only checked that Synapse didn't raise an exception,
but didn't check that we had correctly interpreted the value of the
dodgy power level.
It also conflated two things: bad room notification levels, and bad user
levels. There _is_ logic for converting the latter to integers, but we
should test it separately.
* Check we ignore types that don't convert to int
* Handle `None` values in `notifications.room`
* Changelog
* Also test that bad values are rejected by event auth
* Docstring
* linter scripttttttttt
* Test boolean values in PL content
* Reject boolean power levels
* Changelog
* Extend the auth rule checks for `m.room.create` events
... and move them up to the top of the function. Since the no auth_events are
allowed for m.room.create events, we may as well get the m.room.create event
checks out of the way first.
* Add a test for create events with prev_events
* Add auth events to events used in tests
* Move some event auth checks out to a different method
Some of the event auth checks apply to an event's auth_events, rather than the
state at the event - which means they can play no part in state
resolution. Move them out to a separate method.
* Rename check_auth_rules_for_event
Now it only checks the state-dependent auth rules, it needs a better name.
Instead, use the `room_version` property of the event we're checking.
The `room_version` was originally added as a parameter somewhere around #4482,
but really it's been redundant since #6875 added a `room_version` field to `EventBase`.
Instead, use the `room_version` property of the event we're validating.
The `room_version` was originally added as a parameter somewhere around #4482,
but really it's been redundant since #6875 added a `room_version` field to `EventBase`.
This fixes a bug where we would accept an event whose `auth_events` include
rejected events, if the rejected event was shadowed by another `auth_event`
with same `(type, state_key)`.
The approach is to pass a list of auth events into
`check_auth_rules_for_event` instead of a dict, which of course means updating
the call sites.
This is an extension of #10956.
This fixes a "Event not signed by authorising server" error when
transition room member from join -> join, e.g. when updating a
display name or avatar URL for restricted rooms.
Broadly, the existing `event_auth.check` function has two parts:
* a validation section: checks that the event isn't too big, that it has the rught signatures, etc.
This bit is independent of the rest of the state in the room, and so need only be done once
for each event.
* an auth section: ensures that the event is allowed, given the rest of the state in the room.
This gets done multiple times, against various sets of room state, because it forms part of
the state res algorithm.
Currently, this is implemented with `do_sig_check` and `do_size_check` parameters, but I think
that makes everything hard to follow. Instead, we split the function in two and call each part
separately where it is needed.
* Make historical messages available to federated servers
Part of MSC2716: https://github.com/matrix-org/matrix-doc/pull/2716
Follow-up to https://github.com/matrix-org/synapse/pull/9247
* Debug message not available on federation
* Add base starting insertion point when no chunk ID is provided
* Fix messages from multiple senders in historical chunk
Follow-up to https://github.com/matrix-org/synapse/pull/9247
Part of MSC2716: https://github.com/matrix-org/matrix-doc/pull/2716
---
Previously, Synapse would throw a 403,
`Cannot force another user to join.`,
because we were trying to use `?user_id` from a single virtual user
which did not match with messages from other users in the chunk.
* Remove debug lines
* Messing with selecting insertion event extremeties
* Move db schema change to new version
* Add more better comments
* Make a fake requester with just what we need
See https://github.com/matrix-org/synapse/pull/10276#discussion_r660999080
* Store insertion events in table
* Make base insertion event float off on its own
See https://github.com/matrix-org/synapse/pull/10250#issuecomment-875711889
Conflicts:
synapse/rest/client/v1/room.py
* Validate that the app service can actually control the given user
See https://github.com/matrix-org/synapse/pull/10276#issuecomment-876316455
Conflicts:
synapse/rest/client/v1/room.py
* Add some better comments on what we're trying to check for
* Continue debugging
* Share validation logic
* Add inserted historical messages to /backfill response
* Remove debug sql queries
* Some marker event implemntation trials
* Clean up PR
* Rename insertion_event_id to just event_id
* Add some better sql comments
* More accurate description
* Add changelog
* Make it clear what MSC the change is part of
* Add more detail on which insertion event came through
* Address review and improve sql queries
* Only use event_id as unique constraint
* Fix test case where insertion event is already in the normal DAG
* Remove debug changes
* Switch to chunk events so we can auth via power_levels
Previously, we were using `content.chunk_id` to connect one
chunk to another. But these events can be from any `sender`
and we can't tell who should be able to send historical events.
We know we only want the application service to do it but these
events have the sender of a real historical message, not the
application service user ID as the sender. Other federated homeservers
also have no indicator which senders are an application service on
the originating homeserver.
So we want to auth all of the MSC2716 events via power_levels
and have them be sent by the application service with proper
PL levels in the room.
* Switch to chunk events for federation
* Add unstable room version to support new historical PL
* Fix federated events being rejected for no state_groups
Add fix from https://github.com/matrix-org/synapse/pull/10439
until it merges.
* Only connect base insertion event to prev_event_ids
Per discussion with @erikjohnston,
https://matrix.to/#/!UytJQHLQYfvYWsGrGY:jki.re/$12bTUiObDFdHLAYtT7E-BvYRp3k_xv8w0dUQHibasJk?via=jki.re&via=matrix.org
* Make it possible to get the room_version with txn
* Allow but ignore historical events in unsupported room version
See https://github.com/matrix-org/synapse/pull/10245#discussion_r675592489
We can't reject historical events on unsupported room versions because homeservers without knowledge of MSC2716 or the new room version don't reject historical events either.
Since we can't rely on the auth check here to stop historical events on unsupported room versions, I've added some additional checks in the processing/persisting code (`synapse/storage/databases/main/events.py` -> `_handle_insertion_event` and `_handle_chunk_event`). I've had to do some refactoring so there is method to fetch the room version by `txn`.
* Move to unique index syntax
See https://github.com/matrix-org/synapse/pull/10245#discussion_r675638509
* High-level document how the insertion->chunk lookup works
* Remove create_event fallback for room_versions
See https://github.com/matrix-org/synapse/pull/10245/files#r677641879
* Use updated method name
