Commit graph

15723 commits

Author SHA1 Message Date
Floatingghost
4d6fb43cbd No need to spawn() any more 2024-06-12 02:09:24 +01:00
Floatingghost
ad52135bf5 Convert rich media backfill to oban task 2024-06-11 18:06:51 +01:00
Floatingghost
28d357f52c add diagnostic script 2024-06-10 15:10:47 +01:00
Floatingghost
9c5feb81aa fix tests 2024-06-09 21:26:29 +01:00
Floatingghost
a360836ce3 fix oembed test 2024-06-09 21:17:12 +01:00
Floatingghost
840c70c4fa remove prints 2024-06-09 18:52:09 +01:00
Floatingghost
c65379afea attempt to fix some tests 2024-06-09 18:45:38 +01:00
Floatingghost
16bed0562d Fix tests 2024-06-09 18:28:00 +01:00
Mark Felder
a801dd7b07 Fix module struct matching 2024-06-09 17:38:28 +01:00
Mark Felder
1e86da43f5 Credo 2024-06-09 17:38:24 +01:00
Mark Felder
411831458c Credo 2024-06-09 17:38:18 +01:00
Mark Felder
56463b2121 Fix compile warning
warning: "else" clauses will never match because all patterns in "with" will always match
  lib/pleroma/web/rich_media/parser/ttl/opengraph.ex:10
2024-06-09 17:38:12 +01:00
Mark Felder
2f5eb79473 Mastodon API: Remove deprecated GET /api/v1/statuses/:id/card endpoint
Removed back in 2019

https://github.com/mastodon/mastodon/pull/11213
2024-06-09 17:38:06 +01:00
Mark Felder
f4daa90bd8 Remove test validating missing descriptions are returned as an empty string 2024-06-09 17:37:59 +01:00
Mark Felder
688748b531 Improve test description 2024-06-09 17:37:32 +01:00
Mark Felder
2e5aa71176 Rich Media Cards are fetched asynchonously and not guaranteed to be available on first post render 2024-06-09 17:37:22 +01:00
Mark Felder
7ca655a999 Rich Media Cards are cached by URL not per status 2024-06-09 17:36:57 +01:00
Mark Felder
4746f98851 Fix broken Rich Media parsing when the image URL is a relative path 2024-06-09 17:36:28 +01:00
Mark Felder
765c7e98d2 Respect the TTL returned in OpenGraph tags 2024-06-09 17:36:15 +01:00
Mark Felder
ddbe989461 Fix broken tests 2024-06-09 17:35:47 +01:00
Floatingghost
4a3dd5f65e lost in cherry-pick 2024-06-09 17:34:41 +01:00
Mark Felder
bfe4152385 Increase the :max_body for Rich Media to 5MB
Websites are increasingly getting more bloated with tricks like inlining content (e.g., CNN.com) which puts pages at or above 5MB. This value may still be too low.
2024-06-09 17:34:29 +01:00
Mark Felder
5da9cbd8a5 RichMedia refactor
Rich Media parsing was previously handled on-demand with a 2 second HTTP request timeout and retained only in Cachex. Every time a Pleroma instance is restarted it will have to request and parse the data for each status with a URL detected. When fetching a batch of statuses they were processed in parallel to attempt to keep the maximum latency at 2 seconds, but often resulted in a timeline appearing to hang during loading due to a URL that could not be successfully reached. URLs which had images links that expire (Amazon AWS) were parsed and inserted with a TTL to ensure the image link would not break.

Rich Media data is now cached in the database and fetched asynchronously. Cachex is used as a read-through cache. When the data becomes available we stream an update to the clients. If the result is returned quickly the experience is almost seamless. Activities were already processed for their Rich Media data during ingestion to warm the cache, so users should not normally encounter the asynchronous loading of the Rich Media data.

Implementation notes:

- The async worker is a Task with a globally unique process name to prevent duplicate processing of the same URL
- The Task will attempt to fetch the data 3 times with increasing sleep time between attempts
- The HTTP request obeys the default HTTP request timeout value instead of 2 seconds
- URLs that cannot be successfully parsed due to an unexpected error receives a negative cache entry for 15 minutes
- URLs that fail with an expected error will receive a negative cache with no TTL
- Activities that have no detected URLs insert a nil value in the Cachex :scrubber_cache so we do not repeat parsing the object content with Floki every time the activity is rendered
- Expiring image URLs are handled with an Oban job
- There is no automatic cleanup of the Rich Media data in the database, but it is safe to delete at any time
- The post draft/preview feature makes the URL processing synchronous so the rendered post preview will have an accurate rendering

Overall performance of timelines and creating new posts which contain URLs is greatly improved.
2024-06-09 17:33:48 +01:00
Floatingghost
a924e117fd Add pool timeouts 2024-06-09 17:20:29 +01:00
floatingghost
d1c4b97613 Merge pull request 'Raise minimum PostgreSQL version to 12' (#786) from Oneric/akkoma:psql-min-ver into develop
Reviewed-on: https://akkoma.dev/AkkomaGang/akkoma/pulls/786
2024-06-07 16:53:22 +00:00
Oneric
2180d068ae Raise log level for start failures 2024-06-07 16:21:21 +02:00
Oneric
a3840e7d1f Raise minimum PostgreSQL version to 12
This lets us:
 - avoid issues with broken hash indices for PostgreSQL <10
 - drop runtime checks and legacy codepaths for <11 in db search
 - always enable custom query plans for performance optimisation

PostgreSQL 11 is already EOL since 2023-11-09, so
in theory everyone should already have moved on to 12 anyway.
2024-06-07 16:21:09 +02:00
Oneric
b17d3dc6d8 Fix changelog
Apparently got jumbled during some rebase(s)
2024-06-07 16:20:34 +02:00
floatingghost
f8f364d36d Merge pull request 'Handle errors from HTTP requests gracefully' (#791) from wp-embeds into develop
Reviewed-on: https://akkoma.dev/AkkomaGang/akkoma/pulls/791
2024-06-07 12:58:58 +00:00
floatingghost
329d8fcba8 Merge pull request 'Update PGTune recommendations' (#795) from norm/akkoma:pgtune into develop
Reviewed-on: https://akkoma.dev/AkkomaGang/akkoma/pulls/795
2024-06-07 12:57:00 +00:00
Norm
e2860e5292 Update PGTune recommendations
From experience, setting DB type to "Online transaction processing
system" seems to give the most optimal configuration in terms of
performance.

I also increased the recomended max connections to 25-30 as that leaves
some room for maintenance tasks to run without running out of
connections.

Finally, I removed the example configs since they're probably out of
date and I think it's better to direct people to use PGTune instead.
2024-06-06 12:18:51 -04:00
Floatingghost
0f65dd3ebe remove pointless logger 2024-06-04 14:34:59 +01:00
Floatingghost
38d09cb0ce remove now-pointless clause 2024-06-04 14:34:18 +01:00
Floatingghost
c9a03af7c1 Move rescue to the HTTP request itself 2024-06-04 14:30:16 +01:00
Floatingghost
0f7ae0fa21 am i baka 2024-06-04 14:26:33 +01:00
Floatingghost
30e13a8785 Don't error on rich media fail 2024-06-04 14:21:40 +01:00
Floatingghost
778b213945 enqueue pin fetches after changeset validation 2024-06-01 08:25:35 +01:00
floatingghost
8f97c15b07 Merge pull request 'Preserve Meilisearch’s result ranking' (#772) from Oneric/akkoma:search-meili-order into develop
Reviewed-on: https://akkoma.dev/AkkomaGang/akkoma/pulls/772
2024-05-31 14:12:05 +00:00
Floatingghost
3af0c53a86 use proper workers for fetching pins instead of an ad-hoc task (#788)
Reviewed-on: https://akkoma.dev/AkkomaGang/akkoma/pulls/788
Co-authored-by: Floatingghost <hannah@coffee-and-dreams.uk>
Co-committed-by: Floatingghost <hannah@coffee-and-dreams.uk>
2024-05-31 08:58:52 +00:00
Oneric
fc7e07f424 meilisearch: enable using search_key
Using only the admin key works as well currently
and Akkoma needs to know the admin key to be able
to add new entries etc. However the Meilisearch
key descriptions suggest the admin key is not
supposed to be used for searches, so let’s not.

For compatibility with existings configs, search_key remains optional.
2024-05-29 23:17:27 +00:00
Oneric
59685e25d2 meilisearch: show keys by name not description
This makes show-key’s output match our documentation as of Meilisearch
1.8.0-8-g4d5971f343c00d45c11ef0cfb6f61e83a8508208. Since I’m not sure
if older versions maybe only provided description, it will fallback to
the latter if no name parameter exists.
2024-05-29 23:17:27 +00:00
Oneric
65aeaefa41 meilisearch: respect meili’s result ranking
Meilisearch is already configured to return results sorted by a
particular ranking configured in the meilisearch CLI task.
Resorting the returned top results by date partially negates this and
runs counter to what someone with tweaked settings expects.

Issue and fix identified by AdamK2003 in
https://akkoma.dev/AkkomaGang/akkoma/pulls/579
But instead of using a O(n^2) resorting, this commit directly
retrieves results in the correct order from the database.

Closes: https://akkoma.dev/AkkomaGang/akkoma/pulls/579
2024-05-29 23:17:27 +00:00
Oneric
5d6cb6a459 meilisearch: remove duplicate preload 2024-05-29 23:17:27 +00:00
floatingghost
8afc3bee7a Merge pull request 'Use /var/tmp for media cache path' (#776) from norm/akkoma:nginx-var-tmp into develop
Reviewed-on: https://akkoma.dev/AkkomaGang/akkoma/pulls/776
Reviewed-by: floatingghost <hannah@coffee-and-dreams.uk>
2024-05-28 02:05:17 +00:00
floatingghost
72871d4514 Merge pull request 'Drop unused indices' (#767) from Oneric/akkoma:purge-unused-indices into develop
Reviewed-on: https://akkoma.dev/AkkomaGang/akkoma/pulls/767
2024-05-28 01:35:18 +00:00
floatingghost
72af38c0e9 Merge pull request 'migrate CI config to v2' (#785) from woodpecker-v2 into develop
Reviewed-on: https://akkoma.dev/AkkomaGang/akkoma/pulls/785
2024-05-27 03:32:40 +00:00
Floatingghost
ae19fd90c9 use elixir 1.16 for format checks 2024-05-27 04:07:44 +01:00
Floatingghost
66b3248dd3 mix tests probably shouldn't be async 2024-05-27 04:03:13 +01:00
Floatingghost
73ead8656a don't allow emoji formatter to be async 2024-05-27 03:25:18 +01:00
Floatingghost
f32a7fd76a arch is aarch64 now 2024-05-27 03:02:02 +01:00