Fix outlier re-persisting causing problems with sliding sync tables
Follow-up to https://github.com/element-hq/synapse/pull/17512
When running on `matrix.org`, we discovered that a remote invite is
first persisted as an `outlier` and then re-persisted again where it is
de-outliered. The first the time, the `outlier` is persisted with one
`stream_ordering` but when persisted again and de-outliered, it is
assigned a different `stream_ordering` that won't end up being used.
Since we call `_calculate_sliding_sync_table_changes()` before
`_update_outliers_txn()` which fixes this discrepancy (always use the
`stream_ordering` from the first time it was persisted), we're working
with an unreliable `stream_ordering` value that will possibly be unused
and not make it into the `events` table.
This is so that we can cache it.
We also move the sliding sync types to
`synapse/types/handlers/sliding_sync.py`. This is mainly in-prep for
#17599 to avoid circular imports.
The only change in behaviour is that
`RoomSyncConfig.combine_sync_config(..)` now returns a new room sync
config rather than mutating in-place.
Reviewable commit-by-commit.
---------
Co-authored-by: Eric Eastwood <eric.eastwood@beta.gouv.fr>
Pre-populate room data for quick filtering/sorting in the Sliding Sync
API
Spawning from
https://github.com/element-hq/synapse/pull/17450#discussion_r1697335578
This PR is acting as the Synapse version `N+1` step in the gradual
migration being tracked by
https://github.com/element-hq/synapse/issues/17623
Adding two new database tables:
- `sliding_sync_joined_rooms`: A table for storing room meta data that
the local server is still participating in. The info here can be shared
across all `Membership.JOIN`. Keyed on `(room_id)` and updated when the
relevant room current state changes or a new event is sent in the room.
- `sliding_sync_membership_snapshots`: A table for storing a snapshot of
room meta data at the time of the local user's membership. Keyed on
`(room_id, user_id)` and only updated when a user's membership in a room
changes.
Also adds background updates to populate these tables with all of the
existing data.
We want to have the guarantee that if a row exists in the sliding sync
tables, we are able to rely on it (accurate data). And if a row doesn't
exist, we use a fallback to get the same info until the background
updates fill in the rows or a new event comes in triggering it to be
fully inserted. This means we need a couple extra things in place until
we bump `SCHEMA_COMPAT_VERSION` and run the foreground update in the
`N+2` part of the gradual migration. For context on why we can't rely on
the tables without these things see [1].
1. On start-up, block until we clear out any rows for the rooms that
have had events since the max-`stream_ordering` of the
`sliding_sync_joined_rooms` table (compare to max-`stream_ordering` of
the `events` table). For `sliding_sync_membership_snapshots`, we can
compare to the max-`stream_ordering` of `local_current_membership`
- This accounts for when someone downgrades their Synapse version and
then upgrades it again. This will ensure that we don't have any
stale/out-of-date data in the
`sliding_sync_joined_rooms`/`sliding_sync_membership_snapshots` tables
since any new events sent in rooms would have also needed to be written
to the sliding sync tables. For example a new event needs to bump
`event_stream_ordering` in `sliding_sync_joined_rooms` table or some
state in the room changing (like the room name). Or another example of
someone's membership changing in a room affecting
`sliding_sync_membership_snapshots`.
1. Add another background update that will catch-up with any rows that
were just deleted from the sliding sync tables (based on the activity in
the `events`/`local_current_membership`). The rooms that need
recalculating are added to the
`sliding_sync_joined_rooms_to_recalculate` table.
1. Making sure rows are fully inserted. Instead of partially inserting,
we need to check if the row already exists and fully insert all data if
not.
All of this extra functionality can be removed once the
`SCHEMA_COMPAT_VERSION` is bumped with support for the new sliding sync
tables so people can no longer downgrade (the `N+2` part of the gradual
migration).
<details>
<summary><sup>[1]</sup></summary>
For `sliding_sync_joined_rooms`, since we partially insert rows as state
comes in, we can't rely on the existence of the row for a given
`room_id`. We can't even rely on looking at whether the background
update has finished. There could still be partial rows from when someone
reverted their Synapse version after the background update finished, had
some state changes (or new rooms), then upgraded again and more state
changes happen leaving a partial row.
For `sliding_sync_membership_snapshots`, we insert items as a whole
except for the `forgotten` column ~~so we can rely on rows existing and
just need to always use a fallback for the `forgotten` data. We can't
use the `forgotten` column in the table for the same reasons above about
`sliding_sync_joined_rooms`.~~ We could have an out-of-date membership
from when someone reverted their Synapse version. (same problems as
outlined for `sliding_sync_joined_rooms` above)
Discussed in an [internal
meeting](https://docs.google.com/document/d/1MnuvPkaCkT_wviSQZ6YKBjiWciCBFMd-7hxyCO-OCbQ/edit#bookmark=id.dz5x6ef4mxz7)
</details>
### TODO
- [x] Update `stream_ordering`/`bump_stamp`
- [x] Handle remote invites
- [x] Handle state resets
- [x] Consider adding `sender` so we can filter `LEAVE` memberships and
distinguish from kicks.
- [x] We should add it to be able to tell leaves from kicks
- [x] Consider adding `tombstone` state to help address
https://github.com/element-hq/synapse/issues/17540
- [x] We should add it `tombstone_successor_room_id`
- [x] Consider adding `forgotten` status to avoid extra
lookup/table-join on `room_memberships`
- [x] We should add it
- [x] Background update to fill in values for all joined rooms and
non-join membership
- [x] Clean-up tables when room is deleted
- [ ] Make sure tables are useful to our use case
- First explored in
https://github.com/element-hq/synapse/compare/erikj/ss_use_new_tables
- Also explored in
76b5a576eb
- [x] Plan for how can we use this with a fallback
- See plan discussed above in main area of the issue description
- Discussed in an [internal
meeting](https://docs.google.com/document/d/1MnuvPkaCkT_wviSQZ6YKBjiWciCBFMd-7hxyCO-OCbQ/edit#bookmark=id.dz5x6ef4mxz7)
- [x] Plan for how we can rely on this new table without a fallback
- Synapse version `N+1`: (this PR) Bump `SCHEMA_VERSION` to `87`. Add
new tables and background update to backfill all rows. Since this is a
new table, we don't have to add any `NOT VALID` constraints and validate
them when the background update completes. Read from new tables with a
fallback in cases where the rows aren't filled in yet.
- Synapse version `N+2`: Bump `SCHEMA_VERSION` to `88` and bump
`SCHEMA_COMPAT_VERSION` to `87` because we don't want people to
downgrade and miss writes while they are on an older version. Add a
foreground update to finish off the backfill so we can read from new
tables without the fallback. Application code can now rely on the new
tables being populated.
- Discussed in an [internal
meeting](https://docs.google.com/document/d/1MnuvPkaCkT_wviSQZ6YKBjiWciCBFMd-7hxyCO-OCbQ/edit#bookmark=id.hh7shg4cxdhj)
### Dev notes
```
SYNAPSE_TEST_LOG_LEVEL=INFO poetry run trial tests.storage.test_events.SlidingSyncPrePopulatedTablesTestCase
SYNAPSE_POSTGRES=1 SYNAPSE_POSTGRES_USER=postgres SYNAPSE_TEST_LOG_LEVEL=INFO poetry run trial tests.storage.test_events.SlidingSyncPrePopulatedTablesTestCase
```
```
SYNAPSE_TEST_LOG_LEVEL=INFO poetry run trial tests.handlers.test_sliding_sync.FilterRoomsTestCase
```
Reference:
- [Development docs on background updates and worked examples of gradual
migrations
](1dfa59b238/docs/development/database_schema.md (background-updates))
- A real example of a gradual migration:
https://github.com/matrix-org/synapse/pull/15649#discussion_r1213779514
- Adding `rooms.creator` field that needed a background update to
backfill data, https://github.com/matrix-org/synapse/pull/10697
- Adding `rooms.room_version` that needed a background update to
backfill data, https://github.com/matrix-org/synapse/pull/6729
- Adding `room_stats_state.room_type` that needed a background update to
backfill data, https://github.com/matrix-org/synapse/pull/13031
- Tables from MSC2716: `insertion_events`, `insertion_event_edges`,
`insertion_event_extremities`, `batch_events`
- `current_state_events` updated in
`synapse/storage/databases/main/events.py`
---
```
persist_event (adds to queue)
_persist_event_batch
_persist_events_and_state_updates (assigns `stream_ordering` to events)
_persist_events_txn
_store_event_txn
_update_metadata_tables_txn
_store_room_members_txn
_update_current_state_txn
```
---
> Concatenated Indexes [...] (also known as multi-column, composite or
combined index)
>
> [...] key consists of multiple columns.
>
> We can take advantage of the fact that the first index column is
always usable for searching
>
> *--
https://use-the-index-luke.com/sql/where-clause/the-equals-operator/concatenated-keys*
---
Dealing with `portdb` (`synapse/_scripts/synapse_port_db.py`),
https://github.com/element-hq/synapse/pull/17512#discussion_r1725998219
---
<details>
<summary>SQL queries:</summary>
Both of these are equivalent and work in SQLite and Postgres
Options 1:
```sql
WITH data_table (room_id, user_id, membership_event_id, membership, event_stream_ordering, {", ".join(insert_keys)}) AS (
VALUES (
?, ?, ?,
(SELECT membership FROM room_memberships WHERE event_id = ?),
(SELECT stream_ordering FROM events WHERE event_id = ?),
{", ".join("?" for _ in insert_values)}
)
)
INSERT INTO sliding_sync_non_join_memberships
(room_id, user_id, membership_event_id, membership, event_stream_ordering, {", ".join(insert_keys)})
SELECT * FROM data_table
WHERE membership != ?
ON CONFLICT (room_id, user_id)
DO UPDATE SET
membership_event_id = EXCLUDED.membership_event_id,
membership = EXCLUDED.membership,
event_stream_ordering = EXCLUDED.event_stream_ordering,
{", ".join(f"{key} = EXCLUDED.{key}" for key in insert_keys)}
```
Option 2:
```sql
INSERT INTO sliding_sync_non_join_memberships
(room_id, user_id, membership_event_id, membership, event_stream_ordering, {", ".join(insert_keys)})
SELECT
column1 as room_id,
column2 as user_id,
column3 as membership_event_id,
column4 as membership,
column5 as event_stream_ordering,
{", ".join("column" + str(i) for i in range(6, 6 + len(insert_keys)))}
FROM (
VALUES (
?, ?, ?,
(SELECT membership FROM room_memberships WHERE event_id = ?),
(SELECT stream_ordering FROM events WHERE event_id = ?),
{", ".join("?" for _ in insert_values)}
)
) as v
WHERE membership != ?
ON CONFLICT (room_id, user_id)
DO UPDATE SET
membership_event_id = EXCLUDED.membership_event_id,
membership = EXCLUDED.membership,
event_stream_ordering = EXCLUDED.event_stream_ordering,
{", ".join(f"{key} = EXCLUDED.{key}" for key in insert_keys)}
```
If we don't need the `membership` condition, we could use:
```sql
INSERT INTO sliding_sync_non_join_memberships
(room_id, membership_event_id, user_id, membership, event_stream_ordering, {", ".join(insert_keys)})
VALUES (
?, ?, ?,
(SELECT membership FROM room_memberships WHERE event_id = ?),
(SELECT stream_ordering FROM events WHERE event_id = ?),
{", ".join("?" for _ in insert_values)}
)
ON CONFLICT (room_id, user_id)
DO UPDATE SET
membership_event_id = EXCLUDED.membership_event_id,
membership = EXCLUDED.membership,
event_stream_ordering = EXCLUDED.event_stream_ordering,
{", ".join(f"{key} = EXCLUDED.{key}" for key in insert_keys)}
```
</details>
### Pull Request Checklist
<!-- Please read
https://element-hq.github.io/synapse/latest/development/contributing_guide.html
before submitting your pull request -->
* [x] Pull request is based on the develop branch
* [x] Pull request includes a [changelog
file](https://element-hq.github.io/synapse/latest/development/contributing_guide.html#changelog).
The entry should:
- Be a short description of your change which makes sense to users.
"Fixed a bug that prevented receiving messages from other servers."
instead of "Moved X method from `EventStore` to `EventWorkerStore`.".
- Use markdown where necessary, mostly for `code blocks`.
- End with either a period (.) or an exclamation mark (!).
- Start with a capital letter.
- Feel free to credit yourself, by adding a sentence "Contributed by
@github_username." or "Contributed by [Your Name]." to the end of the
entry.
* [x] [Code
style](https://element-hq.github.io/synapse/latest/code_style.html) is
correct
(run the
[linters](https://element-hq.github.io/synapse/latest/development/contributing_guide.html#run-the-linters))
---------
Co-authored-by: Erik Johnston <erik@matrix.org>
Regressed in #17543.
The `max_download_size` config is not available on workers that don't
load the media repo.
Besides, we should honour the max_size param that was passed into the
function.
When returning receipts in sliding sync for initial rooms we should
always include our own receipts in the room (even if they don't match
any timeline events).
Reviewable commit-by-commit.
---------
Co-authored-by: Eric Eastwood <eric.eastwood@beta.gouv.fr>
Move calculating of the room lists out of the core handler. This should
make it easier to switch things around to start using the tables in
#17512.
This is just moving code between files and methods.
Reviewable commit-by-commit
`hash_password` now actually accepts password from stdin. The `getpass`
reads from TTY, and does NOT accept stdin in any way.
The manpage has been updated to reflect that.
That file was getting long.
The changes are non functional, and simply split things up into:
- the main class
- the connection store
- the extensions
- the types
This supersedes #17503, given the per-connection state is being heavily
rewritten it felt easier to recreate the PR on top of that work.
This correctly handles the case of timeline limits going up and down.
This does not handle changes in `required_state`, but that can be done
as a separate PR.
Based on #17575.
---------
Co-authored-by: Eric Eastwood <eric.eastwood@beta.gouv.fr>
This is some prep work ahead of correctly tracking receipts, where we
will also want to track the room status in terms of last receipt we had
sent down.
Essentially, we add two classes `PerConnectionState` and a mutable
version, and then operate on those.
---------
Co-authored-by: Eric Eastwood <eric.eastwood@beta.gouv.fr>
Results in:
```
AssertionError: null
File "synapse/http/server.py", line 332, in _async_render_wrapper
callback_return = await self._async_render(request)
File "synapse/http/server.py", line 544, in _async_render
callback_return = await raw_callback_return
File "synapse/federation/transport/server/_base.py", line 369, in new_func
response = await func(
File "synapse/federation/transport/server/federation.py", line 826, in on_GET
await self.media_repo.get_local_media(
File "synapse/media/media_repository.py", line 473, in get_local_media
await respond_with_multipart_responder(
File "synapse/media/_base.py", line 353, in respond_with_multipart_responder
assert content_length is not None
```
So that clients can check for support. Note that if the feature is only
enabled for some users, the `/versions` request must be authenticated to
pick up that SSS is enabled for the user
`old_verify_keys` isn't marked as required in
https://spec.matrix.org/v1.11/server-server-api/#get_matrixkeyv2server
and there's no functional difference between an empty object and
omitting the object, so I don't think there's any reason synapse should
explode when the field is omitted.
Follow on from #17558
Basically, we want to reduce the number of threads we want to use at a
time, i.e. reduce the number of threads that are paused/blocked. We do
this by returning from the thread when the consumer pauses the producer,
rather than pausing in the thread.
---------
Co-authored-by: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com>
Previously, we just had very basic partial room exclusion based on
whether we were lazy-loading room members. Now with this PR, we added
`must_await_full_state(...)` with rules to check if we have a we're only
requesting `required_state` which is completely satisfied even with
partial state.
Partially-stated rooms should have all state events except for remote
membership events so if we require a remote membership event anywhere,
then we need to return `True`.
We do this by reusing the code from sync v2.
Reviewable commit-by-commit. The function `get_user_ids_changed` has
been rewritten entirely, so I would recommend not looking at the diff.
This is in response to issue #17473.
Not all the necessary handlers to deal with media requests are started
now when configuring synapse to use a media worker as per the [example
config](https://element-hq.github.io/synapse/latest/workers.html#synapseappmedia_repository).
The new media endpoints introduced with authenticated media fall under
the `client` & `federation` handlers in synapse.
This PR starts up handlers for the new media endpoints if a worker has
been configured with only the `media` resource type.
### Pull Request Checklist
<!-- Please read
https://element-hq.github.io/synapse/latest/development/contributing_guide.html
before submitting your pull request -->
* [X] Pull request is based on the develop branch
* [x] Pull request includes a [changelog
file](https://element-hq.github.io/synapse/latest/development/contributing_guide.html#changelog).
The entry should:
- Be a short description of your change which makes sense to users.
"Fixed a bug that prevented receiving messages from other servers."
instead of "Moved X method from `EventStore` to `EventWorkerStore`.".
- Use markdown where necessary, mostly for `code blocks`.
- End with either a period (.) or an exclamation mark (!).
- Start with a capital letter.
- Feel free to credit yourself, by adding a sentence "Contributed by
@github_username." or "Contributed by [Your Name]." to the end of the
entry.
* [X] [Code
style](https://element-hq.github.io/synapse/latest/code_style.html) is
correct
(run the
[linters](https://element-hq.github.io/synapse/latest/development/contributing_guide.html#run-the-linters))
---------
Co-authored-by: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com>
Spawning from looking at a couple traces and wanting a little more info.
Follow-up to github.com/element-hq/synapse/pull/17501
The changes in this PR allow you to find slow Sliding Sync traces ignoring the
`wait_for_events` time. In Jaeger, you can now filter for the `current_sync_for_user`
operation with `RESULT.result=true` indicating that it actually returned non-empty results.
If you want to find traces for your own user, you can use
`RESULT.result=true ARG.sync_config.user="@madlittlemods:matrix.org"`
This triggers the client to start a new sliding sync connection. If we
don't do this and the client asks for the full range of rooms, we end up
sending down all rooms and their state from scratch (which can be very
slow)
This causes things like
https://github.com/element-hq/element-x-ios/issues/3115 after we restart
the server
---------
Co-authored-by: Eric Eastwood <eric.eastwood@beta.gouv.fr>
Update `filters.is_encrypted` and `filters.types`/`filters.not_types` to
be robust when dealing with remote invite rooms in Sliding Sync.
Part of
[MSC3575](https://github.com/matrix-org/matrix-spec-proposals/pull/3575):
Sliding Sync
Follow-up to https://github.com/element-hq/synapse/pull/17434
We now take into account current state, fallback to stripped state
for invite/knock rooms, then historical state. If we can't determine
the info needed to filter a room (either from state or stripped state),
it is filtered out.
Rather than always including all rooms in range.
Also adds a pre-filter to rooms that checks the stream change cache to
see if anything might have happened.
Based on #17447
---------
Co-authored-by: Eric Eastwood <eric.eastwood@beta.gouv.fr>
The basic idea is that we introduce a new token for a sliding sync
connection, which stores the mapping of room to room "status" (i.e. have
we sent the room down?). This token allows us to handle duplicate
requests properly. In future it can be used to store more
"per-connection" information safely.
In future this should be migrated into the DB, so its important that we
try to reduce the number of syncs where we need to update the
per-connection information. In this PoC this only happens when we: a)
send down a set of room for the first time, or b) we have previously
sent down a room and there are updates but we are not sending the room
down the sync (due to not falling in a list range)
Co-authored-by: Eric Eastwood <eric.eastwood@beta.gouv.fr>
Backfill events have a negative stream ordering, and so its not useful
to use to compare with other (positive) stream orderings.
Plus, the Rust SDK currently assumes `bump_stamp` is positive.
Introduced in: #17215
This caused us a minor bit of grief as the volume of logs produced was
much higher than normal
---------
Signed-off-by: Olivier 'reivilibre <oliverw@matrix.org>
This is to address an issue in which `m.presence` results on initial
sync are not returning entries of users who are currently offline.
The original behaviour was from
https://github.com/element-hq/synapse/issues/1535
This change is useful for applications that use the
presence system for tracking user profile information/updates (e.g.
https://github.com/element-hq/synapse/pull/16992 or for profile status
messages).
This is gated behind a new configuration option to avoid performance
impact for applications that don't need this, as a pragmatic solution
for now.
As it gets used in sliding sync.
We basically invalidate it in all the same places as
`get_rooms_for_user`. Most of the changes are due to needing the
arguments you pass in to be hashable (which lists aren't)
Prior to this PR, remote downloads which did not provide a
`content-length` were decremented from the remote download ratelimiter
at the max allowable size, leading to excessive ratelimiting - see
https://github.com/element-hq/synapse/issues/17394.
This PR adds a linearizer to limit concurrent remote downloads to 6 per
IP address, and decrements remote downloads without a `content-length`
from the ratelimiter *after* the download is complete and the response
length is known.
Also adds logic to ensure that responses with a known length respect the
`max_download_size`.
This removes the `enable_media_repo` attribute on the server config in
favour of always using the `can_load_media_repo` in the media config.
This should avoid issues like in #17420 in the future
Add room subscriptions to Sliding Sync `/sync`
Based on
[MSC3575](https://github.com/matrix-org/matrix-spec-proposals/pull/3575):
Sliding Sync
Currently, you can only subscribe to rooms you have had *any* membership
in before.
In the future, we will allow `world_readable` rooms to be subscribed to
without joining.
We can only fetch room types for rooms the server is in, so we need to
only filter rooms that we're joined to.
Also includes a perf fix to bulk fetch room types.
This looks like a copy/paste error: the function doesn't reject
anything, but instead allows the action count to go through regardless.
The remainder of the function's documentation appears correct.
Added RHEL/Rocky install instructions (PyPI). Instructions cover
versions 8 and 9 which are the only supported ones - except for RHEL7
which is now on extended life cycle support phase.
Large part of the guide is for installing Python 3.11 or 3.12. RHEL8
ships with Python 3.6 and RHEL9 ships with 3.9. Newer Python versions
can be installed easily as they don't interfere with OS software that
still relies on the default Python version.
I was first planning to add prerequisites part to the prerequisites
section and then install instructions on the top of the page but that
section is for pre-built packages so it just didn't sound right. So I
just dumped everything to the PyPI section of the page. But suggestions
to change are welcome.
I also didn't combine these with Fedora section. I haven't tested those
packages on RHEL and Fedora ships with Python 3.12 out-of-box.
