Skip to content
This repository has been archived by the owner on Apr 26, 2024. It is now read-only.

Await un-partial-stating after a partial-state join #12399

Merged
merged 7 commits into from
Apr 21, 2022

Conversation

richvdh
Copy link
Member

@richvdh richvdh commented Apr 6, 2022

When we join a room via the faster-joins mechanism, we end up with "partial state" at some points on the event DAG. Many parts of the codebase need to wait for the full state to load. So, we implement a mechanism to keep track of which events have partial state, and wait for them to be fully-populated.

There are a few commits here, which it might be easier to review individually.

This should pass against the tests in matrix-org/complement#359, though they haven't yet landed.

Part of #11249.

@richvdh richvdh requested a review from a team as a code owner April 6, 2022 15:47
@richvdh richvdh marked this pull request as draft April 6, 2022 15:47
@richvdh richvdh removed the request for review from a team April 12, 2022 09:29
Base automatically changed from rav/faster_joins/06_resync_state to develop April 12, 2022 13:23
@richvdh richvdh force-pushed the rav/faster_joins/07_partial_state_events_tracker branch from f2b8b25 to a6c3810 Compare April 13, 2022 08:36
@richvdh richvdh force-pushed the rav/faster_joins/07_partial_state_events_tracker branch from a6c3810 to 51c8fd7 Compare April 13, 2022 16:18
@richvdh richvdh changed the title Add a mechanism to await un-partial-stating Await un-partial-stating after a partial-state join Apr 13, 2022
richvdh added 3 commits April 13, 2022 17:22
When we join a room via the faster-joins mechanism, we end up with "partial
state" at some points on the event DAG. Many parts of the codebase need to
wait for the full state to load. So, we implement a mechanism to keep track of
which events have partial state, and wait for them to be fully-populated.
@richvdh richvdh force-pushed the rav/faster_joins/07_partial_state_events_tracker branch from 51c8fd7 to 13c8691 Compare April 13, 2022 16:22
@richvdh richvdh marked this pull request as ready for review April 13, 2022 16:25
@richvdh richvdh requested a review from a team April 13, 2022 16:26
async def _get_state_group_for_events(
self,
event_ids: Collection[str],
await_full_state: bool = True,
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

the intention is to follow up with further PRs which relax this by setting await_full_state=False where it is known to be safe to do so.

@squahtx squahtx self-assigned this Apr 20, 2022
Copy link
Contributor

@reivilibre reivilibre left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

squahtx seems to have beaten me to this review, but I did already write...

Comment on lines +805 to +806
if await_full_state:
await self._partial_state_events_tracker.await_full_state(event_ids)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I appreciate the need for this, but this seems like it could block for a long time, potentially exceeding the timeout of a request.
I suppose the plan is to make enough endpoints not need to wait for full state that this won't matter much?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

That's part of the plan. Another part of the plan is to get cancellation working (#3528) so that we don't stack up requests. A final part of the plan is probably to change the api to do something different, but we're a way off that yet.

@@ -515,6 +515,7 @@ async def update_state_for_partial_state_event(
)
return
await self._store.update_state_for_partial_state_event(event, context)
self._state_store.notify_event_un_partial_stated(event.event_id)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Not particularly fussy, but 'un-partial-stated' made me grin; I would have called this completed? 😀

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

maybe? it would have to be notify_event_state_completed or something, to distinguish from other types of completion. Having a single concept which we refer to throughout the codebase as "partial state" seems clearer to me.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You might have a point there, even if 'un-partial' is a little non-obvious, once you notice what it means it makes sense.

Copy link
Contributor

@squahtx squahtx left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

lgtm!

@richvdh richvdh merged commit f5668f0 into develop Apr 21, 2022
@richvdh richvdh deleted the rav/faster_joins/07_partial_state_events_tracker branch April 21, 2022 06:42
babolivier added a commit to matrix-org/synapse-dinsic that referenced this pull request Apr 28, 2022
Synapse 1.58.0rc1 (2022-04-26)
==============================

As of this release, the groups/communities feature in Synapse is now disabled by default. See [\#11584](matrix-org/synapse#11584) for details. As mentioned in [the upgrade notes](https://github.com/matrix-org/synapse/blob/develop/docs/upgrade.md#upgrading-to-v1580), this feature will be removed in Synapse 1.61.

Features
--------

- Implement [MSC3383](matrix-org/matrix-spec-proposals#3383) for including the destination in server-to-server authentication headers. Contributed by @Bubu and @jcgruenhage for Famedly. ([\#11398](matrix-org/synapse#11398))
- Docker images and Debian packages from matrix.org now contain a locked set of Python dependencies, greatly improving build reproducibility. ([Board](https://github.com/orgs/matrix-org/projects/54), [\#11537](matrix-org/synapse#11537))
- Enable processing of device list updates asynchronously. ([\#12365](matrix-org/synapse#12365), [\#12465](matrix-org/synapse#12465))
- Implement [MSC2815](matrix-org/matrix-spec-proposals#2815) to allow room moderators to view redacted event content. Contributed by @tulir. ([\#12427](matrix-org/synapse#12427))
- Build Debian packages for Ubuntu 22.04 "Jammy Jellyfish". ([\#12543](matrix-org/synapse#12543))

Bugfixes
--------

- Prevent a sync request from removing a user's busy presence status. ([\#12213](matrix-org/synapse#12213))
- Fix bug with incremental sync missing events when rejoining/backfilling. Contributed by Nick @ Beeper. ([\#12319](matrix-org/synapse#12319))
- Fix a long-standing bug which incorrectly caused `GET /_matrix/client/v3/rooms/{roomId}/event/{eventId}` to return edited events rather than the original. ([\#12476](matrix-org/synapse#12476))
- Fix a bug introduced in Synapse 1.27.0 where the admin API for [deleting forward extremities](https://github.com/matrix-org/synapse/blob/erikj/fix_delete_event_response_count/docs/admin_api/rooms.md#deleting-forward-extremities) would always return a count of 1, no matter how many extremities were deleted. ([\#12496](matrix-org/synapse#12496))
- Fix a long-standing bug where the image thumbnails embedded into email notifications were broken. ([\#12510](matrix-org/synapse#12510))
- Fix a bug in the implementation of [MSC3202](matrix-org/matrix-spec-proposals#3202) where Synapse would use the field name `device_unused_fallback_keys`, rather than `device_unused_fallback_key_types`. ([\#12520](matrix-org/synapse#12520))
- Fix a bug introduced in Synapse 0.99.3 which could cause Synapse to consume large amounts of RAM when back-paginating in a large room. ([\#12522](matrix-org/synapse#12522))

Improved Documentation
----------------------

- Fix rendering of the documentation site when using the 'print' feature. ([\#12340](matrix-org/synapse#12340))
- Add a manual documenting config file options. ([\#12368](matrix-org/synapse#12368), [\#12527](matrix-org/synapse#12527))
- Update documentation to reflect that both the `run_background_tasks_on` option and the options for moving stream writers off of the main process are no longer experimental. ([\#12451](matrix-org/synapse#12451))
- Update worker documentation and replace old `federation_reader` with `generic_worker`. ([\#12457](matrix-org/synapse#12457))
- Strongly recommend [Poetry](https://python-poetry.org/) for development. ([\#12475](matrix-org/synapse#12475))
- Add some example configurations for workers and update architectural diagram. ([\#12492](matrix-org/synapse#12492))
- Fix a broken link in `README.rst`. ([\#12495](matrix-org/synapse#12495))
- Add HAProxy delegation example with CORS headers to docs. ([\#12501](matrix-org/synapse#12501))
- Remove extraneous comma in User Admin API's device deletion section so that the example JSON is actually valid and works. Contributed by @olmari. ([\#12533](matrix-org/synapse#12533))

Deprecations and Removals
-------------------------

- The groups/communities feature in Synapse is now disabled by default. ([\#12344](matrix-org/synapse#12344))
- Remove unstable identifiers from [MSC3440](matrix-org/matrix-spec-proposals#3440). ([\#12382](matrix-org/synapse#12382))

Internal Changes
----------------

- Preparation for faster-room-join work: start a background process to resynchronise the room state after a room join. ([\#12394](matrix-org/synapse#12394))
- Preparation for faster-room-join work: Implement a tracking mechanism to allow functions to wait for full room state to arrive. ([\#12399](matrix-org/synapse#12399))
- Remove an unstable identifier from [MSC3083](matrix-org/matrix-spec-proposals#3083). ([\#12395](matrix-org/synapse#12395))
- Run CI in the locked [Poetry](https://python-poetry.org/) environment, and remove corresponding `tox` jobs. ([\#12425](matrix-org/synapse#12425), [\#12434](matrix-org/synapse#12434), [\#12438](matrix-org/synapse#12438), [\#12441](matrix-org/synapse#12441), [\#12449](matrix-org/synapse#12449), [\#12478](matrix-org/synapse#12478), [\#12514](matrix-org/synapse#12514), [\#12472](matrix-org/synapse#12472))
- Change Mutual Rooms' `unstable_features` flag to `uk.half-shot.msc2666.mutual_rooms` which matches the current iteration of [MSC2666](matrix-org/matrix-spec-proposals#2666). ([\#12445](matrix-org/synapse#12445))
- Fix typo in the release script help string. ([\#12450](matrix-org/synapse#12450))
- Fix a minor typo in the Debian changelogs generated by the release script. ([\#12497](matrix-org/synapse#12497))
- Reintroduce the list of targets to the linter script, to avoid linting unwanted local-only directories during development. ([\#12455](matrix-org/synapse#12455))
- Limit length of `device_id` to less than 512 characters. ([\#12454](matrix-org/synapse#12454))
- Dockerfile-workers: reduce the amount we install in the image. ([\#12464](matrix-org/synapse#12464))
- Dockerfile-workers: give the master its own log config. ([\#12466](matrix-org/synapse#12466))
- complement-synapse-workers: factor out separate entry point script. ([\#12467](matrix-org/synapse#12467))
- Back out experimental implementation of [MSC2314](matrix-org/matrix-spec-proposals#2314). ([\#12474](matrix-org/synapse#12474))
- Fix grammatical error in federation error response when the room version of a room is unknown. ([\#12483](matrix-org/synapse#12483))
- Remove unnecessary configuration overrides in tests. ([\#12511](matrix-org/synapse#12511))
- Refactor the relations code for clarity. ([\#12519](matrix-org/synapse#12519))
- Add type hints so `docker` and `stubs` directories pass `mypy --disallow-untyped-defs`. ([\#12528](matrix-org/synapse#12528))
- Update `delay_cancellation` to accept any awaitable, rather than just `Deferred`s. ([\#12468](matrix-org/synapse#12468))
- Handle cancellation in `EventsWorkerStore._get_events_from_cache_or_db`. ([\#12529](matrix-org/synapse#12529))
babolivier added a commit to matrix-org/synapse-dinsic that referenced this pull request May 3, 2022
Synapse 1.58.0 (2022-05-03)
===========================

As of this release, the groups/communities feature in Synapse is now disabled by default. See [\#11584](matrix-org/synapse#11584) for details. As mentioned in [the upgrade notes](https://github.com/matrix-org/synapse/blob/develop/docs/upgrade.md#upgrading-to-v1580), this feature will be removed in Synapse 1.61.

No significant changes since 1.58.0rc2.

Synapse 1.58.0rc2 (2022-04-26)
==============================

This release candidate fixes bugs related to Synapse 1.58.0rc1's logic for handling device list updates.

Bugfixes
--------

- Fix a bug introduced in Synapse 1.58.0rc1 where the main process could consume excessive amounts of CPU and memory while handling sentry logging failures. ([\#12554](matrix-org/synapse#12554))
- Fix a bug introduced in Synapse 1.58.0rc1 where opentracing contexts were not correctly sent to whitelisted remote servers with device lists updates. ([\#12555](matrix-org/synapse#12555))

Internal Changes
----------------

- Reduce unnecessary work when handling remote device list updates. ([\#12557](matrix-org/synapse#12557))

Synapse 1.58.0rc1 (2022-04-26)
==============================

Features
--------

- Implement [MSC3383](matrix-org/matrix-spec-proposals#3383) for including the destination in server-to-server authentication headers. Contributed by @Bubu and @jcgruenhage for Famedly. ([\#11398](matrix-org/synapse#11398))
- Docker images and Debian packages from matrix.org now contain a locked set of Python dependencies, greatly improving build reproducibility. ([Board](https://github.com/orgs/matrix-org/projects/54), [\#11537](matrix-org/synapse#11537))
- Enable processing of device list updates asynchronously. ([\#12365](matrix-org/synapse#12365), [\#12465](matrix-org/synapse#12465))
- Implement [MSC2815](matrix-org/matrix-spec-proposals#2815) to allow room moderators to view redacted event content. Contributed by @tulir @ Beeper. ([\#12427](matrix-org/synapse#12427))
- Build Debian packages for Ubuntu 22.04 "Jammy Jellyfish". ([\#12543](matrix-org/synapse#12543))

Bugfixes
--------

- Prevent a sync request from removing a user's busy presence status. ([\#12213](matrix-org/synapse#12213))
- Fix bug with incremental sync missing events when rejoining/backfilling. Contributed by Nick @ Beeper. ([\#12319](matrix-org/synapse#12319))
- Fix a long-standing bug which incorrectly caused `GET /_matrix/client/v3/rooms/{roomId}/event/{eventId}` to return edited events rather than the original. ([\#12476](matrix-org/synapse#12476))
- Fix a bug introduced in Synapse 1.27.0 where the admin API for [deleting forward extremities](https://github.com/matrix-org/synapse/blob/erikj/fix_delete_event_response_count/docs/admin_api/rooms.md#deleting-forward-extremities) would always return a count of 1, no matter how many extremities were deleted. ([\#12496](matrix-org/synapse#12496))
- Fix a long-standing bug where the image thumbnails embedded into email notifications were broken. ([\#12510](matrix-org/synapse#12510))
- Fix a bug in the implementation of [MSC3202](matrix-org/matrix-spec-proposals#3202) where Synapse would use the field name `device_unused_fallback_keys`, rather than `device_unused_fallback_key_types`. ([\#12520](matrix-org/synapse#12520))
- Fix a bug introduced in Synapse 0.99.3 which could cause Synapse to consume large amounts of RAM when back-paginating in a large room. ([\#12522](matrix-org/synapse#12522))

Improved Documentation
----------------------

- Fix rendering of the documentation site when using the 'print' feature. ([\#12340](matrix-org/synapse#12340))
- Add a manual documenting config file options. ([\#12368](matrix-org/synapse#12368), [\#12527](matrix-org/synapse#12527))
- Update documentation to reflect that both the `run_background_tasks_on` option and the options for moving stream writers off of the main process are no longer experimental. ([\#12451](matrix-org/synapse#12451))
- Update worker documentation and replace old `federation_reader` with `generic_worker`. ([\#12457](matrix-org/synapse#12457))
- Strongly recommend [Poetry](https://python-poetry.org/) for development. ([\#12475](matrix-org/synapse#12475))
- Add some example configurations for workers and update architectural diagram. ([\#12492](matrix-org/synapse#12492))
- Fix a broken link in `README.rst`. ([\#12495](matrix-org/synapse#12495))
- Add HAProxy delegation example with CORS headers to docs. ([\#12501](matrix-org/synapse#12501))
- Remove extraneous comma in User Admin API's device deletion section so that the example JSON is actually valid and works. Contributed by @olmari. ([\#12533](matrix-org/synapse#12533))

Deprecations and Removals
-------------------------

- The groups/communities feature in Synapse is now disabled by default. ([\#12344](matrix-org/synapse#12344))
- Remove unstable identifiers from [MSC3440](matrix-org/matrix-spec-proposals#3440). ([\#12382](matrix-org/synapse#12382))

Internal Changes
----------------

- Preparation for faster-room-join work: start a background process to resynchronise the room state after a room join. ([\#12394](matrix-org/synapse#12394))
- Preparation for faster-room-join work: Implement a tracking mechanism to allow functions to wait for full room state to arrive. ([\#12399](matrix-org/synapse#12399))
- Remove an unstable identifier from [MSC3083](matrix-org/matrix-spec-proposals#3083). ([\#12395](matrix-org/synapse#12395))
- Run CI in the locked [Poetry](https://python-poetry.org/) environment, and remove corresponding `tox` jobs. ([\#12425](matrix-org/synapse#12425), [\#12434](matrix-org/synapse#12434), [\#12438](matrix-org/synapse#12438), [\#12441](matrix-org/synapse#12441), [\#12449](matrix-org/synapse#12449), [\#12478](matrix-org/synapse#12478), [\#12514](matrix-org/synapse#12514), [\#12472](matrix-org/synapse#12472))
- Change Mutual Rooms' `unstable_features` flag to `uk.half-shot.msc2666.mutual_rooms` which matches the current iteration of [MSC2666](matrix-org/matrix-spec-proposals#2666). ([\#12445](matrix-org/synapse#12445))
- Fix typo in the release script help string. ([\#12450](matrix-org/synapse#12450))
- Fix a minor typo in the Debian changelogs generated by the release script. ([\#12497](matrix-org/synapse#12497))
- Reintroduce the list of targets to the linter script, to avoid linting unwanted local-only directories during development. ([\#12455](matrix-org/synapse#12455))
- Limit length of `device_id` to less than 512 characters. ([\#12454](matrix-org/synapse#12454))
- Dockerfile-workers: reduce the amount we install in the image. ([\#12464](matrix-org/synapse#12464))
- Dockerfile-workers: give the master its own log config. ([\#12466](matrix-org/synapse#12466))
- complement-synapse-workers: factor out separate entry point script. ([\#12467](matrix-org/synapse#12467))
- Back out experimental implementation of [MSC2314](matrix-org/matrix-spec-proposals#2314). ([\#12474](matrix-org/synapse#12474))
- Fix grammatical error in federation error response when the room version of a room is unknown. ([\#12483](matrix-org/synapse#12483))
- Remove unnecessary configuration overrides in tests. ([\#12511](matrix-org/synapse#12511))
- Refactor the relations code for clarity. ([\#12519](matrix-org/synapse#12519))
- Add type hints so `docker` and `stubs` directories pass `mypy --disallow-untyped-defs`. ([\#12528](matrix-org/synapse#12528))
- Update `delay_cancellation` to accept any awaitable, rather than just `Deferred`s. ([\#12468](matrix-org/synapse#12468))
- Handle cancellation in `EventsWorkerStore._get_events_from_cache_or_db`. ([\#12529](matrix-org/synapse#12529))
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants