Commit Graph

123 Commits

Author SHA1 Message Date
Erik Johnston
4c67f0391b
Split up deleting devices into batches (#16766)
Otherwise for users with large numbers of devices this can cause a lot
of woe.
2024-01-10 13:55:16 +00:00
Patrick Cloke
8e1e62c9e0 Update license headers 2023-11-21 15:29:58 -05:00
Patrick Cloke
ab3f1b3b53
Convert simple_select_one_txn and simple_select_one to return tuples. (#16612) 2023-11-09 11:13:31 -05:00
Patrick Cloke
9738b1c497
Avoid executing no-op queries. (#16583)
If simple_{insert,upsert,update}_many_txn is called without any data
to modify then return instead of executing the query.

This matches the behavior of simple_{select,delete}_many_txn.
2023-11-07 14:00:25 -05:00
Patrick Cloke
cfb6d38c47
Remove remaining usage of cursor_to_dict. (#16564) 2023-10-31 13:13:28 -04:00
Patrick Cloke
679c691f6f
Remove more usages of cursor_to_dict. (#16551)
Mostly to improve type safety.
2023-10-26 15:12:28 -04:00
Patrick Cloke
9407d5ba78
Convert simple_select_list and simple_select_list_txn to return lists of tuples (#16505)
This should use fewer allocations and improves type hints.
2023-10-26 13:01:36 -04:00
Patrick Cloke
a4904dcb04
Convert simple_select_many_batch, simple_select_many_txn to tuples. (#16444) 2023-10-11 13:24:56 -04:00
Patrick Cloke
fa907025f4
Remove manys calls to cursor_to_dict (#16431)
This avoids calling cursor_to_dict and then immediately
unpacking the values in the dict for other users. By not
creating the intermediate dictionary we can avoid allocating
the dictionary and strings for the keys, which should generally
be more performant.

Additionally this improves type hints by avoid Dict[str, Any]
dictionaries coming out of the database layer.
2023-10-05 11:07:38 -04:00
Patrick Cloke
d7c89c5908
Return immutable objects for cachedList decorators (#16350) 2023-09-19 15:26:44 -04:00
Erik Johnston
1cd410a783
Recheck if remote device is cached before requesting it (#16252)
This fixes a bug where we could get stuck re-requesting the device over
replication again and again.
2023-09-07 12:45:43 +00:00
Patrick Cloke
32fb264120 Merge remote-tracking branch 'origin/release-v1.92' into develop 2023-09-06 13:08:22 -04:00
Quentin Gliech
1940d990a3
Revert MSC3861 introspection cache, admin impersonation and account lock (#16258) 2023-09-06 15:19:51 +01:00
Mathieu Velten
4f1840a88a
Delete device messages asynchronously and in staged batches (#16240) 2023-09-06 09:30:53 +02:00
Shay
69048f7b48
Add an admin endpoint to allow authorizing server to signal token revocations (#16125) 2023-08-22 14:15:34 +00:00
Shay
0328b56468
Support MSC3814: Dehydrated Devices Part 2 (#16010) 2023-08-08 12:04:46 -07:00
pacien
07d7cbfe69
devices: use combined ANY clause for faster cleanup (#15861)
Old device entries for the same user were being removed in individual
SQL commands, making the batch take way longer than necessary.

This combines the commands into a single one with a IN/ANY clause.

Example of log entry before the change, regularly observed with
"log_min_duration_statement = 10000" in PostgreSQL's config:

    LOG:  duration: 42538.282 ms  statement:
    DELETE FROM device_lists_stream
    WHERE user_id = '@someone' AND device_id = 'someid1'
    AND stream_id < 123456789
    ;
    DELETE FROM device_lists_stream
    WHERE user_id = '@someone' AND device_id = 'someid2'
    AND stream_id < 123456789
    ;
    [repeated for each device ID of that user, potentially a lot...]

With the patch applied on my instance for the past couple of days, I
no longer notice overly long statements of that particular kind.

Signed-off-by: pacien <pacien.trangirard@pacien.net>
2023-07-03 16:39:38 +02:00
Erik Johnston
5ed0e8c61f
Cache requests for user's devices from federation (#15675)
This should mitigate the issue where lots of different servers requests
the same user's devices all at once.
2023-06-01 13:25:20 +00:00
Patrick Cloke
375b0a8a11
Update code to refer to "workers". (#15606)
A bunch of comments and variables are out of date and use
obsolete terms.
2023-05-16 15:56:38 -04:00
Erik Johnston
6204c3663e
Revert pruning of old devices (#15360)
* Revert "Fix registering a device on an account with lots of devices (#15348)"

This reverts commit f0d8f66eaa.

* Revert "Delete stale non-e2e devices for users, take 3 (#15183)"

This reverts commit 78cdb72cd6.
2023-03-31 13:51:51 +01:00
Erik Johnston
f0d8f66eaa
Fix registering a device on an account with lots of devices (#15348)
Fixes up #15183
2023-03-29 13:37:06 +00:00
Erik Johnston
78cdb72cd6
Delete stale non-e2e devices for users, take 3 (#15183)
This should help reduce the number of devices e.g. simple bots the repeatedly login rack up.

We only delete non-e2e devices as they should be safe to delete, whereas if we delete e2e devices for a user we may accidentally break their ability to receive e2e keys for a message.
2023-03-29 12:07:14 +01:00
Patrick Cloke
02f74f3a99
Combine AbstractStreamIdTracker and AbstractStreamIdGenerator. (#15192)
AbstractStreamIdTracker (now) has only a single sub-class: AbstractStreamIdGenerator,
combine them to simplify some code and remove any direct references to
AbstractStreamIdTracker.
2023-03-03 08:13:37 -05:00
dependabot[bot]
9bb2eac719
Bump black from 22.12.0 to 23.1.0 (#15103) 2023-02-22 15:29:09 -05:00
Sean Quah
d0c713cc85
Return read-only collections from @cached methods (#13755)
It's important that collections returned from `@cached` methods are not
modified, otherwise future retrievals from the cache will return the
modified collection.

This applies to the return values from `@cached` methods and the values
inside the dictionaries returned by `@cachedList` methods. It's not
necessary for the dictionaries returned by `@cachedList` methods
themselves to be read-only.

Signed-off-by: Sean Quah <seanq@matrix.org>
Co-authored-by: David Robertson <davidr@element.io>
2023-02-10 23:29:00 +00:00
Patrick Cloke
a481fb9f98
Refactor get_user_devices_from_cache to avoid mutating cached values. (#15040)
The previous version of the code could mutate a cached value,
but only if the input requested all devices of a user *and* a specific
device.

To avoid this nonsensical situation we no longer fetch a specific
device ID if all of a user's devices are returned.
2023-02-10 08:09:47 -05:00
Erik Johnston
fd296b7343
Fix exception on start up about device lists (#15041)
Fixes #15010.
2023-02-10 09:52:35 +00:00
Sean Quah
cf66d712c6
Fix initialization of _device_list_id_gen (#14914)
On startup, the `_device_list_id_gen` stream id generator is initialized
using the maximum stream id seen in a list of tables. When we started
populating the `device_list_remote_pending` table in #13913, we forgot
to add it to the aforementioned list of tables, so the stream id
generator can hand out old stream ids after a restart. The end result is
that Synapse can fail to handle device list update EDUs after a restart
when a partial state join is in progress.

Add the `device_list_remote_pending` table to the list of tables to
consider when initializing the `_device_list_id_gen` stream id generator.

Signed-off-by: Sean Quah <seanq@matrix.org>
2023-01-26 10:38:49 +00:00
Erik Johnston
65d0386693
Always notify replication when a stream advances (#14877)
This ensures that all other workers are told about stream updates in a timely manner, without having to remember to manually poke replication.
2023-01-20 18:02:18 +00:00
Erik Johnston
2b084c5b71
Merge device list replication streams (#14833) 2023-01-17 09:29:58 +00:00
reivilibre
ba4ea7d13f
Batch up replication requests to request the resyncing of remote users's devices. (#14716) 2023-01-10 11:17:59 +00:00
Nick Mills-Barrett
db1cfe9c80
Update all stream IDs after processing replication rows (#14723)
This creates a new store method, `process_replication_position` that
is called after `process_replication_rows`. By moving stream ID advances
here this guarantees any relevant cache invalidations will have been
applied before the stream is advanced.

This avoids race conditions where Python switches between threads mid
way through processing the `process_replication_rows` method where stream
IDs may be advanced before caches are invalidated due to class resolution
ordering.

See this comment/issue for further discussion:
	https://github.com/matrix-org/synapse/issues/14158#issuecomment-1344048703
2023-01-04 11:49:26 +00:00
reivilibre
74b89c2761
Revert the deletion of stale devices due to performance issues. (#14662) 2022-12-12 13:55:23 +00:00
Erik Johnston
94bc21e69f
Limit the number of devices we delete at once (#14649) 2022-12-09 13:31:32 +00:00
Erik Johnston
c2de2ca630
Delete stale non-e2e devices for users, take 2 (#14595)
This should help reduce the number of devices e.g. simple bots the repeatedly login rack up.

We only delete non-e2e devices as they should be safe to delete, whereas if we delete e2e devices for a user we may accidentally break their ability to receive e2e keys for a message.
2022-12-09 09:37:07 +00:00
Erik Johnston
cee9445884
Better return type for get_all_entities_changed (#14604)
Help callers from using the return value incorrectly by ensuring
that callers explicitly check if there was a cache hit or not.
2022-12-05 15:19:14 -05:00
Patrick Cloke
fac8a38525
Properly handle unknown results for the stream change cache. (#14592)
StreamChangeCache.get_all_changed_entities can return None to signify
it does not have information at the given stream position. Two callers (related
to device lists and presence) were treating this response the same as an empty
list (i.e. there being no updates).
2022-12-02 10:28:41 -05:00
David Robertson
c29e2c6306
Revert "POC delete stale non-e2e devices for users (#14038)" (#14582) 2022-11-29 17:48:48 +00:00
David Robertson
e860316818
Fix UndefinedColumn: column "key_json" does not exist errors when handling users with more than 50 non-E2E devices (#14580) 2022-11-29 13:05:07 +00:00
Erik Johnston
c7e29ca277
POC delete stale non-e2e devices for users (#14038)
This should help reduce the number of devices e.g. simple bots the repeatedly login rack up.

We only delete non-e2e devices as they should be safe to delete, whereas if we delete e2e devices for a user we may accidentally break their ability to receive e2e keys for a message.

Co-authored-by: Patrick Cloke <clokep@users.noreply.github.com>
Co-authored-by: Sean Quah <8349537+squahtx@users.noreply.github.com>
2022-11-29 10:36:41 +00:00
Sean Quah
f792dd74e1
Remove option to skip locking of tables during emulated upserts (#14469)
To perform an emulated upsert into a table safely, we must either:
 * lock the table,
 * be the only writer upserting into the table
 * or rely on another unique index being present.

When the 2nd or 3rd cases were applicable, we previously avoided locking
the table as an optimization. However, as seen in #14406, it is easy to
slip up when adding new schema deltas and corrupt the database.

The only time we lock when performing emulated upserts is while waiting
for background updates on postgres. On sqlite, we do no locking at all.

Let's remove the option to skip locking tables, so that we don't shoot
ourselves in the foot again.

Signed-off-by: Sean Quah <seanq@matrix.org>
2022-11-28 13:42:06 +00:00
Erik Johnston
f38d7d79c8
Add another index to device_lists_changes_in_room (#14534)
This helps avoid reading unnecessarily large amounts of data from the
table when querying with a set of room IDs.
2022-11-23 14:09:00 +00:00
Sean Quah
9cae44f49e
Track unconverted device list outbound pokes using a position instead (#14516)
When a local device list change is added to
`device_lists_changes_in_room`, the `converted_to_destinations` flag is
set to `FALSE` and the `_handle_new_device_update_async` background
process is started. This background process looks for unconverted rows
in `device_lists_changes_in_room`, copies them to
`device_lists_outbound_pokes` and updates the flag.

To update the `converted_to_destinations` flag, the database performs a
`DELETE` and `INSERT` internally, which fragments the table. To avoid
this, track unconverted rows using a `(stream ID, room ID)` position
instead of the flag.

From now on, the `converted_to_destinations` column indicates rows that
need converting to outbound pokes, but does not indicate whether the
conversion has already taken place.

Closes #14037.

Signed-off-by: Sean Quah <seanq@matrix.org>
2022-11-22 16:46:52 +00:00
David Robertson
115f0eb233
Reintroduce #14376, with bugfix for monoliths (#14468)
* Add tests for StreamIdGenerator

* Drive-by: annotate all defs

* Revert "Revert "Remove slaved id tracker (#14376)" (#14463)"

This reverts commit d63814fd73, which in
turn reverted 36097e88c4. This restores
the latter.

* Fix StreamIdGenerator not handling unpersisted IDs

Spotted by @erikjohnston.

Closes #14456.

* Changelog

Co-authored-by: Nick Mills-Barrett <nick@fizzadar.com>
Co-authored-by: Erik Johnston <erik@matrix.org>
2022-11-16 22:16:46 +00:00
Patrick Cloke
d8cc86eff4
Remove redundant types from comments. (#14412)
Remove type hints from comments which have been added
as Python type hints. This helps avoid drift between comments
and reality, as well as removing redundant information.

Also adds some missing type hints which were simple to fill in.
2022-11-16 15:25:24 +00:00
Erik Johnston
d63814fd73
Revert "Remove slaved id tracker (#14376)" (#14463)
This reverts commit 36097e88c4.
2022-11-16 13:50:07 +00:00
Nick Mills-Barrett
36097e88c4
Remove slaved id tracker (#14376)
This matches the multi instance writer ID generator class which can
both handle advancing the current token over replication and by calling
the database.
2022-11-14 17:31:36 +00:00
Nick Mills-Barrett
3a4f80f8c6
Merge/remove Slaved* stores into WorkerStores (#14375) 2022-11-11 10:51:49 +00:00
Richard van der Hoff
1469fed0e3
Add debugging to help diagnose lost device-list-update (#14268) 2022-10-24 10:45:10 +01:00
Aaron Raimist
2a76a7369f
Fix hiding devices names over federation (#10015)
And don't include blank opentracing stuff in device list updates.

Signed-off-by: Aaron Raimist <aaron@raim.ist>
2022-10-18 20:54:27 +00:00