* Try to optimize SelectOneTimeKeys
Signed-off-by: Till Faelligen <tfaelligen@gmail.com>
* Use pg.Array when using ANY...
Co-authored-by: Kegsay <kegan@matrix.org>
* Use LimitReader to prevent DoS risk
Signed-off-by: Till Faelligen <tfaelligen@gmail.com>
* Check if bytesWritten is equal to the maxFileSize
Add tests
Signed-off-by: Till Faelligen <tfaelligen@gmail.com>
* Use oldschool defer to cleanup after the tests
* Let LimitReader read MaxFileSizeBytes + 1
Co-authored-by: Kegsay <kegan@matrix.org>
* Avoid crash on non-compiled room regex
Roughly fixes#1845 (actual compiling still needed)
Signed-off-by: Bohdan Horbeshko <bodqhrohro@gmail.com>
* Compile regexes for all namespaces
Deadheres the regex compiling from building larger regexes for possibly
exclusive namespaces only. A complete fix for #1845, so regexes for
rooms namespaces and other non-whitelisted namespaces can be used
more safely.
Signed-off-by: Bohdan Horbeshko <bodqhrohro@gmail.com>
* Appservice config: handle regexp parsing errors
Signed-off-by: diamondburned <datutbrus@gmail.com>
Signed-off-by: Bohdan Horbeshko <bodqhrohro@gmail.com>
Co-authored-by: Kegsay <kegan@matrix.org>
The update part wasn't executed actually for SQLite, so it is moved to
a separate statement.
Fixes#1852.
Signed-off-by: Bohdan Horbeshko <bodqhrohro@gmail.com>
Now that MSC1772 passed FCP its identifiers have stabilised
This outright drops support for experimental spaces but that's what you get for being on the bleeding edge
* Hash-deduplicated state storage (and migrations) for PostgreSQL and SQLite
* Refactor droomserver database setup for migrations
* Fix conflict statements
* Update migration names
* Set a boundary for old to new block/snapshot IDs so we don't rewrite them more than once accidentally
* Create sequence if not exists
* Fix boundary queries
* Fix boundary queries
* Use Query
* Break out queries a bit
* More sequence tweaks
* Query parameters are not playing the game
* Injection escaping may not work for CREATE SEQUENCE after all
* Fix snapshot sequence name
* Use boundaried IDs in SQLite too
* Use IFNULL for SQLite
* Use COALESCE in PostgreSQL
* Review comments @Kegsay
* fedsender: tolerate dupe membership events
Previously if the fedsender got a duplicate membership event it would cause
the entire process to crash. Now it doesn't. This masks an issue with the
roomserver where it can emit duplicate membership events.
* Update joined_hosts_table.go
* Remove the Content-Length requirement
* Make sure that the file size does not exceed max content size
* Address review comment - universally check if temp file size exceeds max file size
Just changing the Media API's `max_file_size_bytes` isn't enough if
Dendrite is running behind a proxy; document the need for a proxy config
change in the place the admin is most likely to notice it.
Signed-off-by: Tim McCormack <cortex@brainonfire.net>
* Implement OpenID module (#599)
- Unrelated: change Riot references to Element in client API routing
Signed-off-by: Bruce MacDonald <contact@bruce-macdonald.com>
* OpenID module tweaks (#599)
- specify expiry is ms rather than vague ts
- add OpenID token lifetime to configuration
- use Go naming conventions for the path params
- store plaintext token rather than hash
- remove openid table sqllite mutex
* Add default OpenID token lifetime (#599)
* Update dendrite-config.yaml
Co-authored-by: Kegsay <kegsay@gmail.com>
Co-authored-by: Kegsay <kegan@matrix.org>
* Add a per-room mutex to federationapi when processing transactions
This has numerous benefits:
- Prevents us doing lots of state resolutions in busy rooms. Previously, room forks would always result
in a state resolution being performed immediately, without checking if we were already doing this in
a different transaction. Now they will queue up, resulting in fewer calls to `/state_ids`, `/g_m_e`, etc.
- Prevents memory usage from growing too large as a result and potentially OOMing.
And costs:
- High traffic rooms will be slightly slower due to head-of-line blocking from other servers,
though this has always been an issue as roomserver has a per-room mutex already.
* Fix unit tests
* Correct mutex lock ordering
