Under certain conditions, files fetched from remotes trigger an error when
being uploaded using OpenStack Swift. This is because in some cases, the
remote server will not return a content-length, so our ResponseWithLimitAdapter
will hold a `nil` value for `#size`, which will lead to an invalid value
for the Content-Length header of the Swift API call.
This commit fixes that by taking the size from the actually-downloaded file
size rather than the upstream-provided Content-Length header value.
In order to work around https://github.com/mastodon/mastodon/issues/16895,
add a warning to .env.production.sample, and change the mastodon:setup rake
task to:
- output a warning if a variable will be interpreted differently by dotenv
and docker-compose
- ensure the printed config is compatible with docker-compose
* Do not block existing users' emails on self-destruct
That is wasteful and unintuitive
* Do not close registrations when running tootctl self-destruct with --dry-run
* Close registrations on self-destruct regardless of known remote accounts
* Fix tootctl self-destruct not sending Deletes for recently-suspended accounts
* Suspend local users even if no remote account is known
* Do not show scary confirmation text if ran with --dry-run
* Switch from unmaintained paperclip to kt-paperclip
* Drop some compatibility monkey-patches not required by kt-paperclip
* Drop media spoof check monkey-patching
It's broken with kt-paperclip and hopefully it won't be needed anymore
* Fix regression introduced by paperclip 6.1.0
* Do not rely on pathname to call FastImage
* Add test for ogg vorbis file with cover art
* Add audio/vorbis to the accepted content-types
This seems erroneous as this would be the content-type for a vorbis stream
without an ogg container, but that's what the `marcel` gem outputs, so…
* Restore missing for_as_default method
* Refactor Attachmentable concern and delay Paperclip's content-type spoof check
Check for content-type spoofing *after* setting the extension ourselves, this
fixes a regression with kt-paperclip's validations being more strict than
paperclip 6.0.0 and rejecting some Pleroma uploads because of unknown
extensions.
* Please CodeClimate
* Add audio/vorbis to the unreliable set
It doesn't correspond to a file format and thus has no extension associated.
* Add account statuses cleanup policy model
* Record last inspected toot to delete to speed up successive calls to statuses_to_delete
* Add service to cleanup a given account's statuses within a budget
* Add worker to go through account policies and delete old toots
* Fix last inspected status id logic
All existing statuses older or equal to last inspected status id must be
kept by the current policy. This is an invariant that must be kept so that
resuming deletion from the last inspected status remains sound.
* Add tests
* Refactor scheduler and add tests
* Add user interface
* Add support for discriminating based on boosts/favs
* Add UI support for min_reblogs and min_favs, rework UI
* Address first round of review comments
* Replace Snowflake#id_at_start with with_random parameter
* Add tests
* Add tests for StatusesCleanupController
* Rework settings page
* Adjust load-avoiding mechanisms
* Please CodeClimate
* Change references to tootsuite/mastodon to mastodon/mastodon
* Remove obsolete test fixture
* Replace occurrences of tootsuite/mastodon with mastodon/mastodon in CHANGELOG
And a few other places
* Fix account deletion sometimes failing because of optimistic locks
In some rare occasions[1], deleting accounts would fail with a
`StaleObjectError` exception.
Indeed, account deletion manually sets the `AccountStat` values without
handling cases where the optimistic locking on `AccountStat` would fail.
To my knowledge, with the rewrite of account counters in #15913, the
`DeleteAccountService` is now the only place that changes the counters in
a way that is not atomic.
Since in this specific case, we do not care about the previous values of the
account counters, it appears we don't need locking at all for this table
anymore.
[1]: https://discourse.joinmastodon.org/t/account-cant-be-deleted/3602
* Bump MAX_SUPPORTED_VERSION in maintenance script
* Fix issues with POSIX::Spawn, Terrapin and Ruby 3.0
Also improve the Terrapin monkey-patch for the stderr/stdout issue.
* Fix keyword argument handling throughout the codebase
* Monkey-patch Paperclip to fix keyword arguments handling in validators
* Change validation_extensions to please CodeClimate
* Bump microformats from 4.2.1 to 4.3.1
* Allow Ruby 3.0
* Add Ruby 3.0 test target to CircleCI
* Add test for admin dashboard warnings
* Fix admin dashboard warnings on Ruby 3.0
* Fix media processing getting stuck on too much stdin/stderr
See thoughtbot/terrapin#5
* Remove dependency on paperclip-av-transcoder gem
* Remove dependency on streamio-ffmpeg gem
* Disable stdin on ffmpeg process
* Update devise-two-factor to unreleased fork for Rails 6 support
Update tests to match new `rotp` version.
* Update nsa gem to unreleased fork for Rails 6 support
* Update rails to 6.1.3 and rails-i18n to 6.0
* Update to unreleased fork of pluck_each for Ruby 6 support
* Run "rails app:update"
* Add missing ActiveStorage config file
* Use config.ssl_options instead of removed ApplicationController#force_ssl
Disabled force_ssl-related tests as they do not seem to be easily testable
anymore.
* Fix nonce directives by removing Rails 5 specific monkey-patching
* Fix fixture_file_upload deprecation warning
* Fix yield-based test failing with Rails 6
* Use Rails 6's index_with when possible
* Use ActiveRecord::Cache::Store#delete_multi from Rails 6
This will yield better performances when deleting an account
* Disable Rails 6.1's automatic preload link headers
Since Rails 6.1, ActionView adds preload links for javascript files
in the Links header per default.
In our case, that will bloat headers too much and potentially cause
issues with reverse proxies. Furhermore, we don't need those links,
as we already output them as HTML link tags.
* Switch to Rails 6.0 default config
* Switch to Rails 6.1 default config
* Do not include autoload paths in the load path
* Improve account counters handling
* Use ActiveRecord::Base::sanitize_sql to pass values instead of interpolating them
Keep using string interpolation for `key` as it is safe and using
“ActiveRecord::Base::sanitize_sql_hash_for_assignment” would require stitching
bits of SQL in a way that is not more easily checked for safety.
* Add migration hook to catch PostgreSQL versions earlier than 9.5
Mastodon::MigrationHelpers has been forked from Gitlab a long time ago, but
Mastodon has never supported using a MySQL database.
Removing MySQL support from Mastodon::MigrationHelpers makes it a little easier
to maintain. In particular, it removes code that would need updating with
Rails 6.
* Use ActiveRecord::Result#to_ary instead of deprecated to_hash
They do the same thing, and to_hash has been removed from Rails 6.1
* Explicitly name polymorphic indexes to workaround a bug in Rails 6.1
cf. https://github.com/rails/rails/issues/41693
* Fix incorrect usage of “foreign_key” in migration script
* Use `ActiveModel::Errors#delete` instead of deprecated clear method
* Fix link headers tests on Rails 6.1
Rails 6.1 adds values to the Link header by default, thus it is not a
LinkHeader object anymore. Fix the test to parse the Link header instead
of assuming it is a LinkHeader.
* Prepare Mastodon for zeitwerk autoloader (Rails 6)
Add inflections and rename/move a few classes.
In particular, app/lib/exceptions.rb and app/lib/sanitize_config.rb
were manually loaded while still in autoload paths.
* Add inflection for Url → URL
* Fix misuse of foreign_type
* Fix use of removed "add_template_helper"
* Use response.media_type instead of response.content_type in tests
* Fix CSV export controller test on Rails 6
Rails 6 sets a "filename*" field in the Content-Disposition header to
explicitly encode the filename as UTF-8.
This changes checks the first part of the Content-Disposition header so
it matches in both Rails 5 and Rails 6.
* Fix emoji formatting with Rails 6
* Make emoji output more idiomatic and robust
* Switch from redis-rails gem to built-in Rails redis cache storage
* Drop dependency on secure_headers, use always_write_cookie instead
* Fix cookies in Tor Hidden Services by moving configuration to application.rb
* Instead of setting always_write_cookie at boot, monkey-patch ActionDispatch
* Added .deepsource.toml
* Removed bad use of `alias`
* Fixed operand order in the binary expression
* Prefixed unused method arguments with an underscore
* Replaced the old OpenSSL algorithmic constants with the newer strings initializers.
* Removed unnecessary UTF-8 encoding comment
* Fix maintenance script not re-indexing some indexes on textual values
Fixes#15475
* Refresh instance view at the end of the maintenance script run
Co-authored-by: Claire <claire.github-309c@sitedethib.com>
* disable NewCops
* update TargetRubyVersion
* Fix Lint/MissingSuper for ActiveModelSerializers::Model
* Fix Lint/MissingSuper for feed
* Fix Lint/FloatComparison
* Do not use instance variables
* Improve Emoji import
Skip macOS '._' shadow files in tar archive to speed up import
* Fix codeclimate format issue with whitespace
* Update lib/mastodon/emoji_cli.rb
suggestions from Gargron to improve comment
Co-authored-by: Eugen Rochko <eugen@zeonfederated.com>
* Update emoji_cli.rb
Remove extraneous comment (macOS-specific comment now with correct line)
Co-authored-by: Eugen Rochko <eugen@zeonfederated.com>
* Fix being able to import more than allowed number of follows
Without this commit, if someone tries importing a second list of accounts to
follow before the first one has been processed, this will queue imports for
the two whole lists, even if they exceed the account's allowed number of
outgoing follows.
This commit changes it so the individual queued imports aren't exempt from
the follow limit check (they remain exempt from the rate-limiting check
though).
* Catch validation errors to not re-queue failed follows
Co-authored-by: Claire <claire.github-309c@sitedethib.com>
* Fix external user creation failing when invite request text is required
Also fixes tootctl-based user creation.
* Add test about invites when invite request text is otherwise required
Co-authored-by: Claire <claire.github-309c@sitedethib.com>
* Fix ResolveAccountService accepting mismatching acct: URI
* Set attributes that should be updated regardless of suspension
* Fix key fetching
* Automatically merge remote accounts with duplicate `uri`
* Add tests
* Add "tootctl accounts fix-duplicates"
Finds duplicate accounts sharing a same ActivityPub `id`, re-fetch them and
merge them under the canonical `acct:` URI.
Co-authored-by: Claire <claire.github-309c@sitedethib.com>
* Add tootctl maintenance fix-duplicates
This tool goes through the database to detect and fix duplicates.
This operation is very slow and may cause data loss (of data that would be
inaccessible without intervention because of the existing index corruptions).
It tries its best to make sensible decisions, and asks the user in some cases.
* Add warning message in db:migrate hook
* Clear Rails cache after being done with database deduplication
Avoids followers hash cache being incorrect, among other things
* Increase DNS timeout from 1 second to 5 seconds for MX check
1 seconds is rather short when using a recursive DNS resolver which
hasn't got a cached result already available. Use 5 seconds instead,
which is the timeout value we use for outgoing HTTP queries.
* Add more precise error messages for invalid e-mail addresses
* Fix contrast calculation for thumbnail color extraction
Luminance calculation was using 0-255 RGB values instead of 0-1 sRGB values,
leading to incorrectly-computed contrast values.
Since we use ColorDiff already, just use its XYZ colorspace conversion code
to get the value.
* Require at least 3:1 contrast for both accent and foreground colors
* Lower required contrast for the accent color
* Change content-type to be always computed from file data
Restore previous behavior, detecting the content-type isn't very
expensive, and some instances may serve files as application/octet-stream
regardless of their true type, making fetching media from them fail, while
it used to work pre-3.2.0.
* Add test
- Change audio files to not be stripped of metadata
- Automatically extract cover art from audio if it exists
- Add `thumbnail` parameter to `POST /api/v1/media`, `POST /api/v2/media` and `PUT /api/v1/media/:id`
- Add `icon` to represent it in attachments in ActivityPub
- Fix `preview_url` containing URL of missing missing image when there is no thumbnail instead of null
- Fix duration of audio not being displayed on public pages until the file is loaded
* Add tootctl email_domains (block|unblock)
* fix codeclimate issues.
* fix codeclimate issues.
* fix codeclimate issues.
* add list subcommand, remove log_action.
* fix codeclimate issues.
* filter duplicate hostnames,ips before block
* rebase from currnet master branch.
rename email_domains_cli.rb to email_domain_blocks_cli.rb .
rename Mastodon::EmailDomainsCLI to Mastodon::EmailDomainBlocksCLI .
rename command email_domains to email-domain-blocks . (Thor recognizes both of - and _ )
rename subcommand block to add .
rename subcommand unblock to remove .
change the color in list subcommand to while for domain or cyan for childlen.
don't use include() in list subcommand.
suppress console output about succeeded entry.
add console output about count of processed/skipped.
remove capitalization in subcommand description.
remove long_desc in subcommand 'remove'.
remove duplicate where in subcommand 'remove'.
* fix codeclimate issue.
* Add emojis:generate_borders Rake task
* Address review
* Border all dark emoji
* Combine stroke with filter to reduce artifacting
* Cleanup Camera with Flash
* Add stroke-linejoin="round"
The previous filter and tweaks were effectively a poor imitation of it.
There are no artifacts for any dark emoji now!
* Set stroke-width using property
This fixes old versions of Firefox.
* Store emoji in string instead of array
* Use separate arguments for each path segment
* Remove "background: black;"
Fix a regression introduced in #13928, caused by TTY::Command building
shell commands by chaining string substitutions.
Ditch TTY::Command and use system instead (both do shell out).
When using one of the docker-compose containers, mastodon:setup will use the
existing .env.production rather than the generated one during the setup steps.
This is because dotenv does not overwrite env variables that are alreayd
defined, and the docker-compose.yml file loads the environment variables
from .env.production.
- Fix not moving original files of custom emojis
- Fix command failing to move any files with S3 storage
- Fix command marking records as upgraded when move failed
Fix#13594
* add emoji export command to cli
* fix codeclimate issues
* add error when no matching category was found
* add other suggestions
* exit 1 when no matching category is found
* changes according to suggestions
* 👀
* RubyNein
Y u always autoformat :c
* Add announcements
Fix#11006
* Add reactions to announcements
* Add admin UI for announcements
* Add unit tests
* Fix issues
- Add `with_dismissed` param to announcements API
- Fix end date not being formatted when time range is given
- Fix announcement delete causing reactions to send streaming updates
- Fix announcements container growing too wide and mascot too small
- Fix `all_day` being settable when no time range is given
- Change text "Update" to "Announcement"
* Fix scheduler unpublishing announcements before they are due
* Fix filter params not being passed to announcements filter
* Fix comments in statuses_cli
The comment for the part of query keeping statuses which have replies was
incorrect, explaining the part of the query kept favourited statuses instead
* Keep statuses favourited by local users in tootctl statuses remove
* Do delete old replies that are not direct interactions with local users
* Skip reblogs of local statuses instead of all reblogs
* Optimize some queries
* Keep old statuses with recent reblogs
* Add option to clear toots from followed accounts too
* Change animated GIF detection to not shell out to ImageMagick
Signed-off-by: Eugen Rochko <eugen@zeonfederated.com>
* Change video encoding parameters to limit to 10800 video frames
Signed-off-by: Eugen Rochko <eugen@zeonfederated.com>
* Limit GIF image size further
Signed-off-by: Eugen Rochko <eugen@zeonfederated.com>
* Always strip metadata from video files
* Fix code style issues
It's possible that after commit callbacks were not firing when
exceptions occurred in the process. Also, the default Sidekiq
strategy does not push indexing jobs immediately, which is not
necessary and could be part of the issue too.
* Add missing locale file for ga and add rake task to check for it
* Update lib/tasks/repo.rake
Co-Authored-By: Yamagishi Kazutoshi <ykzts@desire.sh>
* Fix check-i18n build
* Add "--exclude-suspended" to tootctl domains crawl
This new option ignores any instances suspended server-wide as
well as their associated subdomains. This queries all domain
blocks up front, then runs a regexp on each domain. This improves
performance over what may be the obvious implementation, which is
to ask `DomainBlocks.blocked?(domain)` for each domain -- this
hits the DB many times, slowing things down considerably.
* cleaning up code style
* Compiling regex
* Removing ternary operator
* Add `tootctl preview_cards remove`
* fix code style
* Remove `Scheduler::PreviewCardsCleanupScheduler` file
* fix code style again
Add exclude case where image_file_name is blank
* Added a function to output confirmation if the specified number of days is less than 2 weeks
* Record account suspend/silence time and keep track of domain blocks
* Also unblock users who were suspended/silenced before dates were recorded
* Add tests
* Keep track of suspending date for users suspended through the CLI
* Show accurate number of accounts that would be affected by unsuspending an instance
* Change migration to set silenced_at and suspended_at
* Revert "Also unblock users who were suspended/silenced before dates were recorded"
This reverts commit a015c65d2d1e28c7b7cfab8b3f8cd5fb48b8b71c.
* Switch from using suspended and silenced to suspended_at and silenced_at
* Add post-deployment migration script to remove `suspended` and `silenced` columns
* Use Account#silence! and Account#suspend! instead of updating the underlying property
* Add silenced_at and suspended_at migration to post-migration
* Change account fabricator to translate suspended and silenced attributes
* Minor fixes
* Make unblocking domains always retroactive
* Add blurhash
* Use fallback color for spoiler when blurhash missing
* Federate the blurhash and accept it as long as it's at most 5x5
* Display unknown media attachments as blurhash placeholders
* Improve style of embed actions and spoiler button
* Change blurhash resolution from 3x3 to 4x4
* Improve dependency definitions
* Fix code style issues
* config: Add GITHUB_REPOSITORY for repository name
* config: Add SOURCE_BASE_URL for repository url
* Show source_url and repository name on getting started
* List the actual accounts that would have been culled during a dry run.
Otherwise, the dry run mode is basically useless.
* Prevent unreachable domains from inheriting the previous status code.
* Update CHANGELOG.md for #10460.
* Add REST API for creating an account
The method is available to apps with a token obtained via the client
credentials grant. It creates a user and account records, as well as
an access token for the app that initiated the request. The user is
unconfirmed, and an e-mail is sent as usual.
The method returns the access token, which the app should save for
later. The REST API is not available to users with unconfirmed
accounts, so the app must be smart to wait for the user to click a
link in their e-mail inbox.
The method is rate-limited by IP to 5 requests per 30 minutes.
* Redirect users back to app from confirmation if they were created with an app
* Add tests
* Return 403 on the method if registrations are not open
* Require agreement param to be true in the API when creating an account
This allowed you to brick your system when running that command, because the accounts would continue to advertise the old public key, but sign things with the new one
- Some associations were missing from the clean-up
- Some attributes were not reset on suspension
- Skip federation and streaming deletes when purging a dead domain
- Move account association definitions to concern
Just the color is not enough change since not everyone uses colored
terminals.
Touching the account makes it so that the account is not in the
threshold window in case of running again
Leave `tootctl accounts cull` to simply check removed accounts from
live domains, and skip temporarily unavailable domains, while listing
them in the final output for further action.
Add `tootctl domains purge DOMAIN` to be able to purge a domain from
that list manually
* Move more tasks to tootctl
- tootctl feeds build
- tootctl feeds clear
- tootctl accounts refresh
Clean up exit codes and help messages
* Move user modifying to tootctl
* Improve user modification through CLI, rename commands
add -> create
mod -> modify
del -> delete
To remove ambiguity
* Fix code style issues
* Fix not being able to unset admin/mod role
* Fix that can't delete media files even if "tootctl media remove" execute when "--background" not attached.
* Revert
This reverts commit 5aa7e09645b27bae38a26030148b23e553ee2662.
* Change to obtain and pass all columns when "--background" option is false.
* If an Update is signed with known key, skip re-following procedure
Because it means the remote actor did *not* lose their database
* Add CLI method for rotating keys
bin/tootctl accounts rotate [USERNAME]
Generates a new RSA key per account and sends out an Update activity
signed with the old key.
* Key rotation: Space out Update fan-outs every 5 minutes per 1000 accounts
* Skip suspended accounts in key rotation
Adopted from GitLab CE. Generate new migration with:
rails g post_deployment_migration name_of_migration_here
By default they are run together with db:migrate. To not run them,
the env variable SKIP_POST_DEPLOYMENT_MIGRATIONS must be set
Code by Yorick Peterse <yorickpeterse@gmail.com>, see also:
83c8241160
To improve the way super tall or super ride images are treated, the
numbers remain the same, 1280x1280 and 400x400, but if an image
is less in one dimension than the other, the other can become larger
Thanks to @WAHa_06x36@mastodon.social for the tip
* Speed up some rake tasks by moving execution to Sidekiq
mastodon:media:remove_silenced
mastodon:media:remove_remote
mastodon:media:redownload_avatars
mastodon:feeds:build
* Fix code style issue
* Revert "Fixes/do not override timestamps (#7331)"
This reverts commit 581a5c9d29.
* Document Snowflake ID corner-case a bit more
Snowflake IDs are used for two purposes: making object identifiers harder to
guess and ensuring they are in chronological order. For this reason, they
are based on the `created_at` attribute of the object.
Unfortunately, inserting items with older snowflakes IDs will break the
assumption of consumers of the paging APIs that new items will always have
a greater identifier than the last seen one.
* Add `override_timestamps` virtual attribute to not correlate snowflake ID with created_at
* No need to re-require sidekiq plugins, they are required via Gemfile
* Add derailed_benchmarks tool, no need to require TTY gems in Gemfile
* Replace ruby-oembed with FetchOEmbedService
Reduce startup by 45382 allocated objects
* Remove preloaded JSON-LD in favour of caching HTTP responses
Reduce boot RAM by about 6 MiB
* Fix tests
* Fix test suite by stubbing out JSON-LD contexts
* Add support for separate Redis for cache
CACHE_REDIS_URL to allow using a different Redis server for cache
purposes, with cache-specific configuration such as key eviction
* Fix code style issues
Comparison was downcasing only one side, therefore if previously
existing account had a non-lowercase spelling, it would be ignored
when checking for duplicates.
New rake task `mastodon:maintenance:find_duplicate_usernames` will
help find constraint violations that might have occured from the
presence of this bug.
Bump version to 2.3.3
HTTP connections must be explicitly closed in many cases, and letting
perform method close connections makes its callers less redundant and
prevent them from forgetting to close connections.
* add detailed SMTP settings setup in mastodon:setup
* add localhost SMTP settings setup in mastodon:setup
* SMTP settings setup should exit after successful delivery of test mail
Media attachments are part of the association cache of statuses,
since they are presumed to be immutable. Unless this cache is
cleared manually, the statuses will continue to look like they
have media embedded.
* Ensure the app does not even start if OTP_SECRET is not set
* Remove PAPERCLIP_SECRET (it's not used by anything, actually)
Imports are for internal consumption and the url option isn't even
used correctly, so we can remove the hash stuff from them
* Fix#201: Account archive download
* Export actor and private key in the archive
* Optimize BackupService
- Add conversation to cached associations of status, because
somehow it was forgotten and is source of N+1 queries
- Explicitly call GC between batches of records being fetched
(Model class allocations are the worst offender)
- Stream media files into the tar in 1MB chunks
(Do not allocate media file (up to 8MB) as string into memory)
- Use #bytesize instead of #size to calculate file size for JSON
(Fix FileOverflow error)
- Segment media into subfolders by status ID because apparently
GIF-to-MP4 media are all named "media.mp4" for some reason
* Keep uniquely generated filename in Paperclip::GifTranscoder
* Ensure dumped files do not overwrite each other by maintaing directory partitions
* Give tar archives a good name
* Add scheduler to remove week-old backups
* Fix code style issue
* Fix avatar and header issues by using custom geometry detector
Revert a part of #6508. The file passed to dynamic styles method
was not actually a file, but an instance of Paperclip::Attachment,
which broke all styles by always returning {} from the method.
One problem with GIF avatars was that Paperclip::GeometryDetector
reported wrong dimensions for them, e.g. 120x120 GIF avatar would
for some reason be detected as 120x53. By writing our own geometry
parser, we can use FastImage, which also happens to be faster than
ImageMagick, to detect image dimensions, which are also correct.
Unfortunately, this PR does not implement skipping a `convert`
entirely if the dimensions are already correct, as I found no easy
way to write that behaviour into Paperclip without rewriting the
Paperclip::Thumbnail class.
* Only invoke convert if dimension or format needs to be changed
* Add better CLI prompt
* Add rake mastodon:setup interactive wizard
* Test db/redis/smtp configurations and add admin user at the end
* Test database connection even when database does not exist yet
* Use PNG images in HTML e-mails
* Make webpack use URLs with host so fonts load inside HTML e-mails
Convert this back to a relative URL in the premailer CSS loader
since local requests are quicker
* Improve responsive design
* Add missing PNG icon
* Add rake task to check and purge accounts that are missing in origin
* Add progress bar and --force options to mastodon:maintenance:purge_removed_accounts
* Add moderator role and add pundit policies for admin actions
* Add rake task for turning user into mod and revoking it again
* Fix handling of unauthorized exception
* Deliver new report e-mails to staff, not just admins
* Add promote/demote to admin UI, hide some actions conditionally
* Fix unused i18n
- Rename Mastodon::TimestampIds into Mastodon::Snowflake for clarity
- Skip for statuses coming from inbox, aka delivered in real-time
- Skip for statuses that claim to be from the future
* Use non-serial IDs
This change makes a number of nontrivial tweaks to the data model in
Mastodon:
* All IDs are now 8 byte integers (rather than mixed 4- and 8-byte)
* IDs are now assigned as:
* Top 6 bytes: millisecond-resolution time from epoch
* Bottom 2 bytes: serial (within the millisecond) sequence number
* See /lib/tasks/db.rake's `define_timestamp_id` for details, but
note that the purpose of these changes is to make it difficult to
determine the number of objects in a table from the ID of any
object.
* The Redis sorted set used for the feed will have values used to look
up toots, rather than scores. This is almost always the same as the
existing behavior, except in the case of boosted toots. This change
was made because Redis stores scores as double-precision floats,
which cannot store the new ID format exactly. Note that this doesn't
cause problems with sorting/pagination, because ZREVRANGEBYSCORE
sorts lexicographically when scores are tied. (This will still cause
sorting issues when the ID gains a new significant digit, but that's
extraordinarily uncommon.)
Note a couple of tradeoffs have been made in this commit:
* lib/tasks/db.rake is used to enforce many/most column constraints,
because this commit seems likely to take a while to bring upstream.
Enforcing a post-migrate hook is an easier way to maintain the code
in the interim.
* Boosted toots will appear in the timeline as many times as they have
been boosted. This is a tradeoff due to the way the feed is saved in
Redis at the moment, but will be handled by a future commit.
This would effectively close Mastodon's #1059, as it is a
snowflake-like system of generating IDs. However, given how involved
the changes were simply within Mastodon, it may have unexpected
interactions with some clients, if they store IDs as doubles
(or as 4-byte integers). This was a problem that Twitter ran into with
their "snowflake" transition, particularly in JavaScript clients that
treated IDs as JS integers, rather than strings. It therefore would be
useful to test these changes at least in the web interface and popular
clients before pushing them to all users.
* Fix JavaScript interface with long IDs
Somewhat predictably, the JS interface handled IDs as numbers, which in
JS are IEEE double-precision floats. This loses some precision when
working with numbers as large as those generated by the new ID scheme,
so we instead handle them here as strings. This is relatively simple,
and doesn't appear to have caused any problems, but should definitely
be tested more thoroughly than the built-in tests. Several days of use
appear to support this working properly.
BREAKING CHANGE:
The major(!) change here is that IDs are now returned as strings by the
REST endpoints, rather than as integers. In practice, relatively few
changes were required to make the existing JS UI work with this change,
but it will likely hit API clients pretty hard: it's an entirely
different type to consume. (The one API client I tested, Tusky, handles
this with no problems, however.)
Twitter ran into this issue when introducing Snowflake IDs, and decided
to instead introduce an `id_str` field in JSON responses. I have opted
to *not* do that, and instead force all IDs to 64-bit integers
represented by strings in one go. (I believe Twitter exacerbated their
problem by rolling out the changes three times: once for statuses, once
for DMs, and once for user IDs, as well as by leaving an integer ID
value in JSON. As they said, "If you’re using the `id` field with JSON
in a Javascript-related language, there is a very high likelihood that
the integers will be silently munged by Javascript interpreters. In most
cases, this will result in behavior such as being unable to load or
delete a specific direct message, because the ID you're sending to the
API is different than the actual identifier associated with the
message." [1]) However, given that this is a significant change for API
users, alternatives or a transition time may be appropriate.
1: https://blog.twitter.com/developer/en_us/a/2011/direct-messages-going-snowflake-on-sep-30-2011.html
* Restructure feed pushes/unpushes
This was necessary because the previous behavior used Redis zset scores
to identify statuses, but those are IEEE double-precision floats, so we
can't actually use them to identify all 64-bit IDs. However, it leaves
the code in a much better state for refactoring reblog handling /
coalescing.
Feed-management code has been consolidated in FeedManager, including:
* BatchedRemoveStatusService no longer directly manipulates feed zsets
* RemoveStatusService no longer directly manipulates feed zsets
* PrecomputeFeedService has moved its logic to FeedManager#populate_feed
(PrecomputeFeedService largely made lots of calls to FeedManager, but
didn't follow the normal adding-to-feed process.)
This has the effect of unifying all of the feed push/unpush logic in
FeedManager, making it much more tractable to update it in the future.
Due to some additional checks that must be made during, for example,
batch status removals, some Redis pipelining has been removed. It does
not appear that this should cause significantly increased load, but if
necessary, some optimizations are possible in batch cases. These were
omitted in the pursuit of simplicity, but a batch_push and batch_unpush
would be possible in the future.
Tests were added to verify that pushes happen under expected conditions,
and to verify reblog behavior (both on pushing and unpushing). In the
case of unpushing, this includes testing behavior that currently leads
to confusion such as Mastodon's #2817, but this codifies that the
behavior is currently expected.
* Rubocop fixes
I could swear I made these changes already, but I must have lost them
somewhere along the line.
* Address review comments
This addresses the first two comments from review of this feature:
https://github.com/tootsuite/mastodon/pull/4801#discussion_r139336735https://github.com/tootsuite/mastodon/pull/4801#discussion_r139336931
This adds an optional argument to FeedManager#key, the subtype of feed
key to generate. It also tests to ensure that FeedManager's settings are
such that reblogs won't be tracked forever.
* Hardcode IdToBigints migration columns
This addresses a comment during review:
https://github.com/tootsuite/mastodon/pull/4801#discussion_r139337452
This means we'll need to make sure that all _id columns going forward
are bigints, but that should happen automatically in most cases.
* Additional fixes for stringified IDs in JSON
These should be the last two. These were identified using eslint to try
to identify any plain casts to JavaScript numbers. (Some such casts are
legitimate, but these were not.)
Adding the following to .eslintrc.yml will identify casts to numbers:
~~~
no-restricted-syntax:
- warn
- selector: UnaryExpression[operator='+'] > :not(Literal)
message: Avoid the use of unary +
- selector: CallExpression[callee.name='Number']
message: Casting with Number() may coerce string IDs to numbers
~~~
The remaining three casts appear legitimate: two casts to array indices,
one in a server to turn an environment variable into a number.
* Only implement timestamp IDs for Status IDs
Per discussion in #4801, this is only being merged in for Status IDs at
this point. We do this in a migration, as there is no longer use for
a post-migration hook. We keep the initialization of the timestamp_id
function as a Rake task, as it is also needed after db:schema:load (as
db/schema.rb doesn't store Postgres functions).
* Change internal streaming payloads to stringified IDs as well
This is equivalent to 591a9af356faf2d5c7e66e3ec715502796c875cd from
#5019, with an extra change for the addition to FeedManager#unpush.
* Ensure we have a status_id_seq sequence
Apparently this is not a given when specifying a custom ID function,
so now we ensure it gets created. This uses the generic version of this
function to more easily support adding additional tables with timestamp
IDs in the future, although it would be possible to cut this down to a
less generic version if necessary. It is only run during db:schema:load
or the relevant migration, so the overhead is extraordinarily minimal.
* Transition reblogs to new Redis format
This provides a one-way migration to transition old Redis reblog entries
into the new format, with a separate tracking entry for reblogs.
It is not invertible because doing so could (if timestamp IDs are used)
require a database query for each status in each users' feed, which is
likely to be a significant toll on major instances.
* Address review comments from @akihikodaki
No functional changes.
* Additional review changes
* Heredoc cleanup
* Run db:schema:load hooks for test in development
This matches the behavior in Rails'
ActiveRecord::Tasks::DatabaseTasks.each_current_configuration, which
would otherwise break `rake db:setup` in development.
It also moves some functionality out to a library, which will be a good
place to put additional related functionality in the near future.
* Make IdsToBigints (mostly!) non-blocking
This pulls in GitLab's MigrationHelpers, which include code to make
column changes in ways that Postgres can do without locking. In general,
this involves creating a new column, adding an index and any foreign
keys as appropriate, adding a trigger to keep it populated alongside
the old column, and then progressively copying data over to the new
column, before removing the old column and replacing it with the new
one.
A few changes to GitLab's MigrationHelpers were necessary:
* Some changes were made to remove dependencies on other GitLab code.
* We explicitly wait for index creation before forging ahead on column
replacements.
* We use different temporary column names, to avoid running into index
name length limits.
* We rename the generated indices back to what they "should" be after
replacing columns.
* We rename the generated foreign keys to use the new column names when
we had to create them. (This allows the migration to be rolled back
without incident.)
# Big Scary Warning
There are two things here that may trip up large instances:
1. The change for tables' "id" columns is not concurrent. In
particular, the stream_entries table may be big, and does not
concurrently migrate its id column. (On the other hand, x_id type
columns are all concurrent.)
2. This migration will take a long time to run, *but it should not
lock tables during that time* (with the exception of the "id"
columns as described above). That means this should probably be run
in `screen` or some other session that can be run for a long time.
Notably, the migration will take *longer* than it would without
these changes, but the website will still be responsive during that
time.
These changes were tested on a relatively large statuses table (256k
entries), and the service remained responsive during the migration.
Migrations both forward and backward were tested.
* Rubocop fixes
* MigrationHelpers: Support ID columns in some cases
This doesn't work in cases where the ID column is referred to as a
foreign key by another table.
* MigrationHelpers: support foreign keys for ID cols
Note that this does not yet support foreign keys on non-primary-key
columns, but Mastodon also doesn't yet have any that we've needed to
migrate.
This means we can perform fully "concurrent" migrations to change ID
column types, and the IdsToBigints migration can happen with effectively
no downtime. (A few operations require a transaction, such as renaming
columns or deleting them, but these transactions should not block for
noticeable amounts of time.)
The algorithm for generating foreign key names has changed with this,
and therefore all of those changed in schema.rb.
* Provide status, allow for interruptions
The MigrationHelpers now allow restarting the rename of a column if it
was interrupted, by removing the old "new column" and re-starting the
process.
Along with this, they now provide status updates on the changes which
are happening, as well as indications about when the changes can be
safely interrupted (when there are at least 10 seconds estimated to be
left before copying data is complete).
The IdsToBigints migration now also sorts the columns it migrates by
size, starting with the largest tables. This should provide
administrators a worst-case scenario estimate for the length of
migrations: each successive change will get faster, giving admins a
chance to abort early on if they need to run the migration later. The
idea is that this does not force them to try to time interruptions
between smaller migrations.
* Fix column sorting in IdsToBigints
Not a significant change, but it impacts the order of columns in the
database and db/schema.rb.
* Actually pause before IdsToBigints
- 500.html generated with admin-set default locale if set
- Error page `<title>` includes Mastodon site title
- 500 title changed to "This page is not
correct" (ref: <https://www.youtube.com/watch?v=2VCAP_seh1A>)
- 500 content appended with "on our end" to make clear it's
not user's fault
A new rake task emojis:generate downloads a full list of valid
unicode sequences from unicode.org and checks it against existing
Twemoji files, finally generating a map from each sequence to the
existing file (e.g. when there's multiple ways an emoji can be
expressed). The map is dumped into app/javascript/mastodon/emoji_map.json
That file is loaded by emojione_light.js (now a misnomer) which
decorates it further with shortcodes taken from emoji-mart's index.
* Add emoji autosuggest
Some credit goes to glitch-soc/mastodon#149
* Remove server-side shortcode->unicode conversion
* Insert shortcode when suggestion is custom emoji
* Remove remnant of server-side emojis
* Update style of autosuggestions
* Fix wrong emoji filenames generated in autosuggest item
* Do not lazy load emoji picker, as that no longer works
* Fix custom emoji autosuggest
* Fix multiple "Custom" categories getting added to emoji index, only add once
* Revert "Enable UniqueRetryJobMiddleware even when called from sidekiq worker (#4836)"
This reverts commit 6859d4c028.
* Revert "Do not execute the job with the same arguments as the retry job (#4814)"
This reverts commit be7ffa2d75.
