Conflicts:
- `app/controllers/statuses_controller.rb`:
Upstream disabled the embed controller for reblogs.
Not a real conflict, but glitch-soc has an extra line to deal
with its theming system.
Ported upstream changes.
- `app/javascript/packs/public.js`:
Upstream made changes to get rid of most inline CSS, this changes
javascript for public pages, which in glitch are split between
different files. Ported those changes.
- `app/models/status.rb`:
Upstream changed the block check in `Status#permitted_for` to
include domain-block checks. Not a real conflict with glitch-soc,
but our scope is slightly different, as our scope for
unauthenticated access do not include instance-local toots.
Ported upstream changes.
- `app/serializers/rest/instance_serializer.rb`:
Not a real conflict, upstream added a new field to the instance
serializer, the conflict is one line above since we added more of
that.
Ported upstream changes.
- `app/views/settings/profiles/show.html.haml`:
Upstream got rid of most inline CSS and moved hidden elements
to data attributes in the process, in fields were we have
different values.
Ported upstream changes while keeping our glitch-specific
values.
- `app/views/statuses/_simple_status.html.haml`:
Upstream got rid of inline CSS on an HAML line we treat
differently, stripping empty text nodes.
Ported upstream changes to the style attribute, keeping
the empty text node stripping behavior.
* 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
Conflicts:
- `app/javascript/packs/public.js`:
Upstream modified code that we split between multiple files due
to glitch-soc's theming system.
Ported those changes.
Conflicts:
- `app/javascript/mastodon/features/compose/components/poll_form.js`:
Upstream bumped poll option character limit, but we already had
a higher one, kept ours.
- `app/validators/poll_validator.rb`:
Upstream bumped poll option character limit, but we already had
a higher one, kept ours.
- `config/initializers/content_security_policy.rb`:
Upstream added a rule, the way we compute ours is different, but
that added rule has been ported.
- `package.json`:
No real conflict, dependency update. Performed the same update.
- `yarn.lock`:
No real conflict, dependency update. Performed the same update.
Conflicts:
- `app/controllers/api/v1/statuses_controller.rb`:
Conflict due to upstream adding a new parameter (with_rate_limit),
too close to glitch-soc's own additional parameter (content_type).
Added upstream's parameter.
- `app/services/post_status_service.rb`:
Conflict due to upstream adding a new parameter (rate_limit),
too close to glitch-soc's own additional parameter (content_type).
Added upstream's parameter.
- `app/views/settings/preferences/appearance/show.html.haml`:
Conflict due to us not exposing theme settings here (as we have
a different flavour/skin menu).
Took upstream change, while still not exposing theme settings.
- `config/webpack/shared.js`:
Coding style fixes for a part we have rewritten.
Discarded upstream changes.
Conflicts:
- `Gemfile`:
We updated httplog in a separate commit.
Took upstream's change which updated it further.
- `Gemfile.lock`:
We updated httplog in a separate commit.
Took upstream's change which updated it further.
- `app/lib/sanitize_config.rb`:
Upstream added better unsupported link stripping,
while we had different sanitizing configs.
Took only upstream's link stripping code.
- `config/locales/simple_form.pl.yml`:
Strings unused in glitch-soc had been removed from
glitch-soc, reintroduced them even if they are not
useful, to reduce the risk of later merge conflicts.
Conflicts:
- `Gemfile.lock`:
No real conflict, glitch-soc-only dependency (redcarpet) too close to an
upstream one (rdf-normalize)
- `README.md`:
we have different READMEs, discarded upstream's changes
- `app/views/admin/custom_emojis/index.html.haml`:
No real conflict, different context because of glitch-soc theming
- `lib/mastodon/statuses_cli.rb`:
Upstream added code to keep bookmarked statuses, we were already doing so
with slightly different code. Discarded upstream's changes.
- `package.json`:
No real conflict, glitch-soc-only dependency (favico.js) too close to
an upstream one
Conflicts:
- `config/locales/en.yml`
No real conflict, upstream added a translatable string “too close” to
one specific to glitch-soc
- `lib/mastodon/statuses_cli.rb`
Fixes made upstream, while changed in glitch-soc to keep bookmarked statuses
- `package.json`
No real conflict, additional dependency in glitch-soc
* 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
Conflicts:
- app/models/media_attachment.rb
Upstream raised max image size from 8MB to 10MB while our limit is
configurable. Raised the default to 10MB.
Conflicts:
- Gemfile
- app/controllers/api/v1/search_controller.rb
Conflict because we changed the number of default results to be
configurable
- app/lib/settings/scoped_settings.rb
Addition of a new “noindex” site-wide setting,
conflict due to our change of the two other site-wide settings
(default flavour and skin instead of theme)
- spec/controllers/application_controller_spec.rb
Addition of a new “noindex” site-wide setting,
conflict due to our change of the two other site-wide settings
(default flavour and skin instead of theme)
* 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
Conflicts:
- app/models/media_attachment.rb
Upstream added audio attachment support
- app/serializers/initial_state_serializer.rb
Upstream added audio attachment support and how mimetypes are returned
- app/serializers/rest/instance_serializer.rb
Upstream added a few fields
- config/application.rb
Upstream added a different paperclip transcoder
Conflicts:
- app/controllers/settings/preferences_controller.rb
- app/lib/user_settings_decorator.rb
- app/models/user.rb
Conflicts due to the addition of a new preference upstream,
“advanced layout”.
* 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
Conflicts:
- app/controllers/statuses_controller.rb
minor conflict because of glitch-soc's theming system
- app/controllers/stream_entries_controller.rb
minor conflict because of glitch-soc's theming system
Conflicts:
- app/javascript/mastodon/features/compose/components/compose_form.js
Upstream cleaned up a bit, including on lines in which
we replaced the hardcoded 500 character limit with a maxChar
constant. Applied the changes while keeping maxChar instead of 500.
- app/javascript/packs/public.js
Moved upstream's new animated avatar hover handling in
app/javascript/core/public.js
- app/javascript/styles/fonts/montserrat.scss
Upstream fixed local font name, applied those changes.
- app/javascript/styles/fonts/roboto.scss
Upstream fixed local font name, applied those changes.
- lib/mastodon/version.rb
Upstream made repo URL configurable, did the same, but
default to glitch-soc
* config: Add GITHUB_REPOSITORY for repository name
* config: Add SOURCE_BASE_URL for repository url
* Show source_url and repository name on getting started
Conflicts:
- app/views/admin/pending_accounts/index.html.haml
No real conflict, upstream changes too close to glitch-specific
theming system changes.
- config/navigation.rb
Upstream redesigned the settings navigation, took those changes
and re-inserted the flavours menu.
* 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.
Conflicts:
- Dockerfile
Took upstream version and reverted f13ebd02c9.
Hopefuly it's not needed anymore.
- app/controllers/api/v1/search_controller.rb
The conflict was due to us raising the number of results returned.
Upstream raised it further, so took it.
- config/locales/de.yml
Took upstream changes to theme translation strings.
- config/locales/gl.yml
Took upstream changes to theme translation strings.
- config/locales/nl.yml
Took upstream changes to theme translation strings.
- config/locales/sk.yml
Took upstream changes to theme translation strings.
Conflicts:
- .eslintrc.yml
Removed, as upstream removed it.
- app/controllers/admin/statuses_controller.rb
Minor code cleanup when porting one of our features.
- app/models/account.rb
Note length validation has changed upstream.
We now use upstream's validation (dropped legacy glitch-soc
account metadata stuff) but with configurable limit.
- app/services/post_status_service.rb
Upstream has added support for scheduled toots, refactoring
the code a bit. Adapted our changes to this refactoring.
- app/views/stream_entries/_detailed_status.html.haml
Not a real conflict, changes too close.
- app/views/stream_entries/_simple_status.html.haml
Not a real conflict, changes too close.
* 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
Conflicts:
- app/controllers/admin/base_controller.rb
Some refactoring made upstream, no real conflict.
- app/javascript/mastodon/features/compose/components/compose_form.js
Updated using upstream's code but using maxChars instead of the
hardcoded length of 500 characters per toot.
- app/javascript/styles/mastodon/components.scss
Upstream redesigned the onboarding modal. Not sure why we had a
conflict there.
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
Conflicts:
- app/controllers/auth/sessions_controller.rb
Upstream reverted something we partially reverted already.
Reverted the rest to match upstream.
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
Conflicts:
- app/controllers/admin/base_controller.rb
- app/controllers/filters_controller.rb
- app/controllers/invites_controller.rb
- app/controllers/settings/deletes_controller.rb
- app/controllers/settings/exports_controller.rb
- app/controllers/settings/follower_domains_controller.rb
- app/controllers/settings/migrations_controller.rb
- app/controllers/settings/notifications_controller.rb
- app/controllers/settings/preferences_controller.rb
- app/controllers/settings/two_factor_authentication/recovery_codes_controller.rb
- app/javascript/packs/public.js
- app/views/settings/profiles/show.html.haml
Conflicts were mostly due to the addition of body classes to the settings page,
this was caused by rejecting upstream changes for most of those files and
modifying Settings::BaseController instead.
Another cause of conflicts was the deletion of client-side checking of
display name / bio length, this was modified in app/javascript/core/settings.js
instead.
Conflicts:
- .github/ISSUE_TEMPLATE/bug_report.md
Took our version.
- CONTRIBUTING.md
Updated the embedded copy of upstream's version.
- README.md
Took our version.
- app/policies/status_policy.rb
Not a real conflict, took code from both.
- app/views/layouts/embedded.html.haml
Added upstream's changes (dns-prefetch) and fixed
`%body.embed`
- app/views/settings/preferences/show.html.haml
Reverted some of upstream changes, as we have a
page dedicated for flavours and skins.
- config/initializers/content_security_policy.rb
Kept our version of the CSP.
- config/initializers/doorkeeper.rb
Not a real conflict, took code from both.
Conflicts:
db/migrate/20170716191202_add_hide_notifications_to_mute.rb
spec/controllers/application_controller_spec.rb
Took our version, upstream changes were only minor style linting.
* 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
Conflicts:
app/views/layouts/application.html.haml
Edited:
app/helpers/application_helper.rb
app/views/admin/domain_blocks/new.html.haml
Conflict wasn't really one, just two changes too close to one another.
Edition was to adapt the class names for themes to class names for
skins and flavours.
Also edited app/views/admin/domain_blocks/new.html.haml to strip the
duplicate admin pack inclusion thing.
Manually-resolved conflicts:
.circleci/config.yml
app/controllers/accounts_controller.rb
app/controllers/auth/passwords_controller.rb
app/controllers/statuses_controller.rb
app/javascript/packs/public.js
app/models/media_attachment.rb
app/views/stream_entries/_content_spoiler.html.haml
app/views/stream_entries/_media.html.haml
config/locales/en.yml
config/locales/ja.yml
config/locales/pl.yml
lib/mastodon/version.rb
Some content from app/javascript/packs/public.js has been split to
app/javascript/core/settings.js.
Translation strings for glitch-soc's keyword mutes were dropped.
Everything else was mostly “take both”.
Conflicts:
README.md
app/controllers/statuses_controller.rb
app/lib/feed_manager.rb
config/navigation.rb
spec/lib/feed_manager_spec.rb
Conflicts were resolved by taking both versions for each change.
This means the two filter systems (glitch-soc's keyword mutes and tootsuite's
custom filters) are in place, which will be changed in a follow-up commit.
Conflicts:
app/javascript/styles/mastodon-light.scss
config/locales/en.yml
config/locales/fr.yml
config/locales/simple_form.pl.yml
config/themes.yml
Conflicts resolved by deleting config/themes.yml,
marking app/javascript/styles/mastodon-light.scss as added,
and taking all new translation strings, not removing anything from
them.
* 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
* 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
* 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
- 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
* 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.
