2016-02-21 08:53:20 +11:00
# This file is auto-generated from the current state of the database. Instead
# of editing this file, please use the migrations feature of Active Record to
# incrementally modify your database, and then regenerate this schema definition.
#
# Note that this schema.rb definition is the authoritative source for your
# database schema. If you need to create the application database on another
# system, you should be using db:schema:load, not running all the migrations
# from scratch. The latter is a flawed and unsustainable approach (the more migrations
# you'll amass, the slower it'll run and the greater likelihood for issues).
#
# It's strongly recommended that you check this file into your version control system.
2020-03-13 08:35:20 +11:00
ActiveRecord :: Schema . define ( version : 2020_03_12_185443 ) do
2018-10-08 08:44:58 +11:00
2016-02-21 08:53:20 +11:00
# These are extensions that must be enabled in order to support this database
enable_extension " plpgsql "
2019-09-20 04:58:19 +10:00
create_table " account_aliases " , force : :cascade do | t |
t . bigint " account_id "
t . string " acct " , default : " " , null : false
t . string " uri " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " ] , name : " index_account_aliases_on_account_id "
end
2018-10-08 08:44:58 +11:00
create_table " account_conversations " , force : :cascade do | t |
t . bigint " account_id "
t . bigint " conversation_id "
t . bigint " participant_account_ids " , default : [ ] , null : false , array : true
t . bigint " status_ids " , default : [ ] , null : false , array : true
t . bigint " last_status_id "
t . integer " lock_version " , default : 0 , null : false
2018-10-19 10:47:29 +11:00
t . boolean " unread " , default : false , null : false
2018-10-08 08:44:58 +11:00
t . index [ " account_id " , " conversation_id " , " participant_account_ids " ] , name : " index_unique_conversations " , unique : true
t . index [ " account_id " ] , name : " index_account_conversations_on_account_id "
t . index [ " conversation_id " ] , name : " index_account_conversations_on_conversation_id "
end
2017-09-22 21:20:04 +10:00
create_table " account_domain_blocks " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . string " domain "
Account domain blocks (#2381)
* Add <ostatus:conversation /> tag to Atom input/output
Only uses ref attribute (not href) because href would be
the alternate link that's always included also.
Creates new conversation for every non-reply status. Carries
over conversation for every reply. Keeps remote URIs verbatim,
generates local URIs on the fly like the rest of them.
* Conversation muting - prevents notifications that reference a conversation
(including replies, favourites, reblogs) from being created. API endpoints
/api/v1/statuses/:id/mute and /api/v1/statuses/:id/unmute
Currently no way to tell when a status/conversation is muted, so the web UI
only has a "disable notifications" button, doesn't work as a toggle
* Display "Dismiss notifications" on all statuses in notifications column, not just own
* Add "muted" as a boolean attribute on statuses JSON
For now always false on contained reblogs, since it's only relevant for
statuses returned from the notifications endpoint, which are not nested
Remove "Disable notifications" from detailed status view, since it's
only relevant in the notifications column
* Up max class length
* Remove pending test for conversation mute
* Add tests, clean up
* Rename to "mute conversation" and "unmute conversation"
* Raise validation error when trying to mute/unmute status without conversation
* Adding account domain blocks that filter notifications and public timelines
* Add tests for domain blocks in notifications, public timelines
Filter reblogs of blocked domains from home
* Add API for listing and creating account domain blocks
* API for creating/deleting domain blocks, tests for Status#ancestors
and Status#descendants, filter domain blocks from them
* Filter domains in streaming API
* Update account_domain_block_spec.rb
2017-05-19 09:14:30 +10:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id "
2017-06-03 00:18:54 +10:00
t . index [ " account_id " , " domain " ] , name : " index_account_domain_blocks_on_account_id_and_domain " , unique : true
Account domain blocks (#2381)
* Add <ostatus:conversation /> tag to Atom input/output
Only uses ref attribute (not href) because href would be
the alternate link that's always included also.
Creates new conversation for every non-reply status. Carries
over conversation for every reply. Keeps remote URIs verbatim,
generates local URIs on the fly like the rest of them.
* Conversation muting - prevents notifications that reference a conversation
(including replies, favourites, reblogs) from being created. API endpoints
/api/v1/statuses/:id/mute and /api/v1/statuses/:id/unmute
Currently no way to tell when a status/conversation is muted, so the web UI
only has a "disable notifications" button, doesn't work as a toggle
* Display "Dismiss notifications" on all statuses in notifications column, not just own
* Add "muted" as a boolean attribute on statuses JSON
For now always false on contained reblogs, since it's only relevant for
statuses returned from the notifications endpoint, which are not nested
Remove "Disable notifications" from detailed status view, since it's
only relevant in the notifications column
* Up max class length
* Remove pending test for conversation mute
* Add tests, clean up
* Rename to "mute conversation" and "unmute conversation"
* Raise validation error when trying to mute/unmute status without conversation
* Adding account domain blocks that filter notifications and public timelines
* Add tests for domain blocks in notifications, public timelines
Filter reblogs of blocked domains from home
* Add API for listing and creating account domain blocks
* API for creating/deleting domain blocks, tests for Status#ancestors
and Status#descendants, filter domain blocks from them
* Filter domains in streaming API
* Update account_domain_block_spec.rb
2017-05-19 09:14:30 +10:00
end
2019-03-19 07:00:55 +11:00
create_table " account_identity_proofs " , force : :cascade do | t |
t . bigint " account_id "
t . string " provider " , default : " " , null : false
t . string " provider_username " , default : " " , null : false
t . text " token " , default : " " , null : false
t . boolean " verified " , default : false , null : false
t . boolean " live " , default : false , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " , " provider " , " provider_username " ] , name : " index_account_proofs_on_account_and_provider_and_username " , unique : true
t . index [ " account_id " ] , name : " index_account_identity_proofs_on_account_id "
end
2019-09-20 04:58:19 +10:00
create_table " account_migrations " , force : :cascade do | t |
t . bigint " account_id "
t . string " acct " , default : " " , null : false
t . bigint " followers_count " , default : 0 , null : false
t . bigint " target_account_id "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " ] , name : " index_account_migrations_on_account_id "
t . index [ " target_account_id " ] , name : " index_account_migrations_on_target_account_id "
end
2017-10-08 05:26:43 +11:00
create_table " account_moderation_notes " , force : :cascade do | t |
t . text " content " , null : false
2017-10-10 22:12:17 +11:00
t . bigint " account_id " , null : false
t . bigint " target_account_id " , null : false
2017-10-08 05:26:43 +11:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " ] , name : " index_account_moderation_notes_on_account_id "
t . index [ " target_account_id " ] , name : " index_account_moderation_notes_on_target_account_id "
end
2018-08-09 17:56:53 +10:00
create_table " account_pins " , force : :cascade do | t |
t . bigint " account_id "
t . bigint " target_account_id "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " , " target_account_id " ] , name : " index_account_pins_on_account_id_and_target_account_id " , unique : true
t . index [ " account_id " ] , name : " index_account_pins_on_account_id "
t . index [ " target_account_id " ] , name : " index_account_pins_on_target_account_id "
end
2018-11-19 10:43:52 +11:00
create_table " account_stats " , force : :cascade do | t |
t . bigint " account_id " , null : false
t . bigint " statuses_count " , default : 0 , null : false
t . bigint " following_count " , default : 0 , null : false
t . bigint " followers_count " , default : 0 , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2018-12-07 03:36:11 +11:00
t . datetime " last_status_at "
2019-10-02 12:59:37 +10:00
t . integer " lock_version " , default : 0 , null : false
2018-11-19 10:43:52 +11:00
t . index [ " account_id " ] , name : " index_account_stats_on_account_id " , unique : true
end
2018-12-07 03:36:11 +11:00
create_table " account_tag_stats " , force : :cascade do | t |
t . bigint " tag_id " , null : false
t . bigint " accounts_count " , default : 0 , null : false
t . boolean " hidden " , default : false , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " tag_id " ] , name : " index_account_tag_stats_on_tag_id " , unique : true
end
Add moderation warnings (#9519)
* Add moderation warnings
Replace individual routes for disabling, silencing, and suspending
a user, as well as the report update route, with a unified account
action controller that allows you to select an action (none,
disable, silence, suspend) as well as whether it should generate an
e-mail notification with optional custom text. That notification,
with the optional custom text, is saved as a warning.
Additionally, there are warning presets you can configure to save
time when performing the above.
* Use Account#local_username_and_domain
2018-12-23 06:02:09 +11:00
create_table " account_warning_presets " , force : :cascade do | t |
t . text " text " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2020-03-13 03:57:59 +11:00
t . string " title " , default : " " , null : false
Add moderation warnings (#9519)
* Add moderation warnings
Replace individual routes for disabling, silencing, and suspending
a user, as well as the report update route, with a unified account
action controller that allows you to select an action (none,
disable, silence, suspend) as well as whether it should generate an
e-mail notification with optional custom text. That notification,
with the optional custom text, is saved as a warning.
Additionally, there are warning presets you can configure to save
time when performing the above.
* Use Account#local_username_and_domain
2018-12-23 06:02:09 +11:00
end
create_table " account_warnings " , force : :cascade do | t |
t . bigint " account_id "
t . bigint " target_account_id "
t . integer " action " , default : 0 , null : false
t . text " text " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " ] , name : " index_account_warnings_on_account_id "
t . index [ " target_account_id " ] , name : " index_account_warnings_on_target_account_id "
end
2017-09-22 21:20:04 +10:00
create_table " accounts " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . string " username " , default : " " , null : false
t . string " domain "
t . string " secret " , default : " " , null : false
t . text " private_key "
t . text " public_key " , default : " " , null : false
t . string " remote_url " , default : " " , null : false
t . string " salmon_url " , default : " " , null : false
t . string " hub_url " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . text " note " , default : " " , null : false
t . string " display_name " , default : " " , null : false
t . string " uri " , default : " " , null : false
t . string " url "
t . string " avatar_file_name "
t . string " avatar_content_type "
t . integer " avatar_file_size "
2016-02-28 10:02:59 +11:00
t . datetime " avatar_updated_at "
2017-06-03 00:18:54 +10:00
t . string " header_file_name "
t . string " header_content_type "
t . integer " header_file_size "
2016-03-13 06:47:22 +11:00
t . datetime " header_updated_at "
2017-06-03 00:18:54 +10:00
t . string " avatar_remote_url "
2016-09-20 08:39:03 +10:00
t . datetime " subscription_expires_at "
2017-06-03 00:18:54 +10:00
t . boolean " locked " , default : false , null : false
t . string " header_remote_url " , default : " " , null : false
2017-04-17 04:32:17 +10:00
t . datetime " last_webfingered_at "
2017-07-20 01:06:46 +10:00
t . string " inbox_url " , default : " " , null : false
t . string " outbox_url " , default : " " , null : false
t . string " shared_inbox_url " , default : " " , null : false
t . string " followers_url " , default : " " , null : false
t . integer " protocol " , default : 0 , null : false
2017-11-08 05:06:44 +11:00
t . boolean " memorial " , default : false , null : false
2017-11-19 05:39:02 +11:00
t . bigint " moved_to_account_id "
2018-03-04 19:19:11 +11:00
t . string " featured_collection_url "
2018-04-14 20:41:08 +10:00
t . jsonb " fields "
2018-05-07 17:31:07 +10:00
t . string " actor_type "
2018-12-07 03:36:11 +11:00
t . boolean " discoverable "
2018-12-29 12:24:36 +11:00
t . string " also_known_as " , array : true
2019-05-15 03:05:02 +10:00
t . datetime " silenced_at "
t . datetime " suspended_at "
2019-07-14 00:45:50 +10:00
t . integer " trust_level "
2020-03-09 10:10:29 +11:00
t . boolean " hide_collections "
2017-03-18 06:47:38 +11:00
t . index " (((setweight(to_tsvector('simple'::regconfig, (display_name)::text), 'A':: \" char \" ) || setweight(to_tsvector('simple'::regconfig, (username)::text), 'B':: \" char \" )) || setweight(to_tsvector('simple'::regconfig, (COALESCE(domain, ''::character varying))::text), 'C':: \" char \" ))) " , name : " search_index " , using : :gin
2018-05-30 10:51:26 +10:00
t . index " lower((username)::text), lower((domain)::text) " , name : " index_accounts_on_username_and_domain_lower " , unique : true
2018-08-22 04:11:34 +10:00
t . index [ " moved_to_account_id " ] , name : " index_accounts_on_moved_to_account_id "
2017-06-03 00:18:54 +10:00
t . index [ " uri " ] , name : " index_accounts_on_uri "
t . index [ " url " ] , name : " index_accounts_on_url "
2016-02-21 08:53:20 +11:00
end
2018-12-07 03:36:11 +11:00
create_table " accounts_tags " , id : false , force : :cascade do | t |
t . bigint " account_id " , null : false
t . bigint " tag_id " , null : false
t . index [ " account_id " , " tag_id " ] , name : " index_accounts_tags_on_account_id_and_tag_id "
t . index [ " tag_id " , " account_id " ] , name : " index_accounts_tags_on_tag_id_and_account_id " , unique : true
end
2017-11-24 12:05:53 +11:00
create_table " admin_action_logs " , force : :cascade do | t |
t . bigint " account_id "
t . string " action " , default : " " , null : false
t . string " target_type "
t . bigint " target_id "
t . text " recorded_changes " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " ] , name : " index_admin_action_logs_on_account_id "
t . index [ " target_type " , " target_id " ] , name : " index_admin_action_logs_on_target_type_and_target_id "
end
2020-01-24 08:00:13 +11:00
create_table " announcement_mutes " , force : :cascade do | t |
t . bigint " account_id "
t . bigint " announcement_id "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " , " announcement_id " ] , name : " index_announcement_mutes_on_account_id_and_announcement_id " , unique : true
t . index [ " account_id " ] , name : " index_announcement_mutes_on_account_id "
t . index [ " announcement_id " ] , name : " index_announcement_mutes_on_announcement_id "
end
create_table " announcement_reactions " , force : :cascade do | t |
t . bigint " account_id "
t . bigint " announcement_id "
t . string " name " , default : " " , null : false
t . bigint " custom_emoji_id "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " , " announcement_id " , " name " ] , name : " index_announcement_reactions_on_account_id_and_announcement_id " , unique : true
t . index [ " account_id " ] , name : " index_announcement_reactions_on_account_id "
t . index [ " announcement_id " ] , name : " index_announcement_reactions_on_announcement_id "
t . index [ " custom_emoji_id " ] , name : " index_announcement_reactions_on_custom_emoji_id "
end
create_table " announcements " , force : :cascade do | t |
t . text " text " , default : " " , null : false
t . boolean " published " , default : false , null : false
t . boolean " all_day " , default : false , null : false
t . datetime " scheduled_at "
t . datetime " starts_at "
t . datetime " ends_at "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2020-01-27 08:43:18 +11:00
t . datetime " published_at "
2020-04-05 20:51:22 +10:00
t . bigint " status_ids " , array : true
2020-01-24 08:00:13 +11:00
end
2018-02-22 09:21:32 +11:00
create_table " backups " , force : :cascade do | t |
t . bigint " user_id "
t . string " dump_file_name "
t . string " dump_content_type "
t . datetime " dump_updated_at "
t . boolean " processed " , default : false , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2020-01-24 08:00:13 +11:00
t . bigint " dump_file_size "
2018-02-22 09:21:32 +11:00
end
2017-09-22 21:20:04 +10:00
create_table " blocks " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id " , null : false
t . bigint " target_account_id " , null : false
2018-05-05 05:14:34 +10:00
t . string " uri "
2017-06-03 00:18:54 +10:00
t . index [ " account_id " , " target_account_id " ] , name : " index_blocks_on_account_id_and_target_account_id " , unique : true
2018-08-22 04:11:34 +10:00
t . index [ " target_account_id " ] , name : " index_blocks_on_target_account_id "
2016-10-04 02:11:54 +11:00
end
2019-11-14 09:02:10 +11:00
create_table " bookmarks " , force : :cascade do | t |
t . bigint " account_id " , null : false
t . bigint " status_id " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " , " status_id " ] , name : " index_bookmarks_on_account_id_and_status_id " , unique : true
t . index [ " account_id " ] , name : " index_bookmarks_on_account_id "
t . index [ " status_id " ] , name : " index_bookmarks_on_status_id "
end
2017-09-22 21:20:04 +10:00
create_table " conversation_mutes " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . bigint " conversation_id " , null : false
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id " , null : false
2017-06-03 00:18:54 +10:00
t . index [ " account_id " , " conversation_id " ] , name : " index_conversation_mutes_on_account_id_and_conversation_id " , unique : true
Feature conversations muting (#3017)
* Add <ostatus:conversation /> tag to Atom input/output
Only uses ref attribute (not href) because href would be
the alternate link that's always included also.
Creates new conversation for every non-reply status. Carries
over conversation for every reply. Keeps remote URIs verbatim,
generates local URIs on the fly like the rest of them.
* Conversation muting - prevents notifications that reference a conversation
(including replies, favourites, reblogs) from being created. API endpoints
/api/v1/statuses/:id/mute and /api/v1/statuses/:id/unmute
Currently no way to tell when a status/conversation is muted, so the web UI
only has a "disable notifications" button, doesn't work as a toggle
* Display "Dismiss notifications" on all statuses in notifications column, not just own
* Add "muted" as a boolean attribute on statuses JSON
For now always false on contained reblogs, since it's only relevant for
statuses returned from the notifications endpoint, which are not nested
Remove "Disable notifications" from detailed status view, since it's
only relevant in the notifications column
* Up max class length
* Remove pending test for conversation mute
* Add tests, clean up
* Rename to "mute conversation" and "unmute conversation"
* Raise validation error when trying to mute/unmute status without conversation
2017-05-15 11:04:13 +10:00
end
2017-06-03 00:18:54 +10:00
create_table " conversations " , force : :cascade do | t |
2019-07-29 01:47:37 +10:00
t . string " uri "
2017-05-13 03:09:21 +10:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2019-07-29 01:47:37 +10:00
t . index [ " uri " ] , name : " index_conversations_on_uri " , unique : true
2017-05-13 03:09:21 +10:00
end
2019-06-28 23:54:10 +10:00
create_table " custom_emoji_categories " , force : :cascade do | t |
t . string " name "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " name " ] , name : " index_custom_emoji_categories_on_name " , unique : true
end
2017-09-19 10:42:40 +10:00
create_table " custom_emojis " , force : :cascade do | t |
t . string " shortcode " , default : " " , null : false
t . string " domain "
t . string " image_file_name "
t . string " image_content_type "
t . integer " image_file_size "
t . datetime " image_updated_at "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2017-10-06 08:42:05 +11:00
t . boolean " disabled " , default : false , null : false
2017-10-08 02:43:42 +11:00
t . string " uri "
t . string " image_remote_url "
2017-10-28 01:11:30 +11:00
t . boolean " visible_in_picker " , default : true , null : false
2019-06-28 23:54:10 +10:00
t . bigint " category_id "
2017-09-19 10:42:40 +10:00
t . index [ " shortcode " , " domain " ] , name : " index_custom_emojis_on_shortcode_and_domain " , unique : true
end
2018-06-29 23:34:36 +10:00
create_table " custom_filters " , force : :cascade do | t |
t . bigint " account_id "
t . datetime " expires_at "
t . text " phrase " , default : " " , null : false
t . string " context " , default : [ ] , null : false , array : true
t . boolean " irreversible " , default : false , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2018-08-09 17:56:53 +10:00
t . boolean " whole_word " , default : true , null : false
2018-06-29 23:34:36 +10:00
t . index [ " account_id " ] , name : " index_custom_filters_on_account_id "
end
2019-07-30 19:10:46 +10:00
create_table " domain_allows " , force : :cascade do | t |
t . string " domain " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " domain " ] , name : " index_domain_allows_on_domain " , unique : true
end
2017-09-22 21:20:04 +10:00
create_table " domain_blocks " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . string " domain " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . integer " severity " , default : 0
2017-07-13 11:12:25 +10:00
t . boolean " reject_media " , default : false , null : false
2018-10-20 17:02:44 +11:00
t . boolean " reject_reports " , default : false , null : false
2019-08-08 04:20:23 +10:00
t . text " private_comment "
t . text " public_comment "
2017-06-03 00:18:54 +10:00
t . index [ " domain " ] , name : " index_domain_blocks_on_domain " , unique : true
2016-10-09 23:48:43 +11:00
end
2017-10-05 00:16:10 +11:00
create_table " email_domain_blocks " , force : :cascade do | t |
2017-11-15 06:37:17 +11:00
t . string " domain " , default : " " , null : false
2017-10-05 00:16:10 +11:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2020-03-13 08:35:20 +11:00
t . bigint " parent_id "
2017-11-15 06:37:17 +11:00
t . index [ " domain " ] , name : " index_email_domain_blocks_on_domain " , unique : true
2017-10-05 00:16:10 +11:00
end
2017-09-22 21:20:04 +10:00
create_table " favourites " , force : :cascade do | t |
2016-02-24 05:17:37 +11:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id " , null : false
t . bigint " status_id " , null : false
2017-07-26 11:35:25 +10:00
t . index [ " account_id " , " id " ] , name : " index_favourites_on_account_id_and_id "
2017-06-03 00:18:54 +10:00
t . index [ " account_id " , " status_id " ] , name : " index_favourites_on_account_id_and_status_id " , unique : true
t . index [ " status_id " ] , name : " index_favourites_on_status_id "
2016-02-24 05:17:37 +11:00
end
2019-02-04 14:25:59 +11:00
create_table " featured_tags " , force : :cascade do | t |
t . bigint " account_id "
t . bigint " tag_id "
t . bigint " statuses_count " , default : 0 , null : false
t . datetime " last_status_at "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " ] , name : " index_featured_tags_on_account_id "
t . index [ " tag_id " ] , name : " index_featured_tags_on_tag_id "
end
2017-09-22 21:20:04 +10:00
create_table " follow_requests " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id " , null : false
t . bigint " target_account_id " , null : false
Allow hiding of reblogs from followed users (#5762)
* Allow hiding of reblogs from followed users
This adds a new entry to the account menu to allow users to hide
future reblogs from a user (and then if they've done that, to show
future reblogs instead).
This does not remove or add historical reblogs from/to the user's
timeline; it only affects new statuses.
The API for this operates by sending a "reblogs" key to the follow
endpoint. If this is sent when starting a new follow, it will be
respected from the beginning of the follow relationship (even if
the follow request must be approved by the followee). If this is
sent when a follow relationship already exists, it will simply
update the existing follow relationship. As with the notification
muting, this will now return an object ({reblogs: [true|false]}) or
false for each follow relationship when requesting relationship
information for an account. This should cause few issues due to an
object being truthy in many languages, but some modifications may
need to be made in pickier languages.
Database changes: adds a show_reblogs column (default true,
non-nullable) to the follows and follow_requests tables. Because
these are non-nullable, we use the existing MigrationHelpers to
perform this change without locking those tables, although the
tables are likely to be small anyway.
Tests included.
See also <https://github.com/glitch-soc/mastodon/pull/212>.
* Rubocop fixes
* Code review changes
* Test fixes
This patchset closes #648 and resolves #3271.
* Rubocop fix
* Revert reblogs defaulting in argument, fix tests
It turns out we needed this for the same reason we needed it in muting:
if nil gets passed in somehow (most usually by an API client not passing
any value), we need to detect and handle it.
We could specify a default in the parameter and then also catch nil, but
there's no great reason to duplicate the default value.
2017-11-29 01:00:35 +11:00
t . boolean " show_reblogs " , default : true , null : false
2018-05-05 05:14:34 +10:00
t . string " uri "
2017-06-03 00:18:54 +10:00
t . index [ " account_id " , " target_account_id " ] , name : " index_follow_requests_on_account_id_and_target_account_id " , unique : true
2016-12-23 09:03:57 +11:00
end
2017-09-22 21:20:04 +10:00
create_table " follows " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id " , null : false
t . bigint " target_account_id " , null : false
Allow hiding of reblogs from followed users (#5762)
* Allow hiding of reblogs from followed users
This adds a new entry to the account menu to allow users to hide
future reblogs from a user (and then if they've done that, to show
future reblogs instead).
This does not remove or add historical reblogs from/to the user's
timeline; it only affects new statuses.
The API for this operates by sending a "reblogs" key to the follow
endpoint. If this is sent when starting a new follow, it will be
respected from the beginning of the follow relationship (even if
the follow request must be approved by the followee). If this is
sent when a follow relationship already exists, it will simply
update the existing follow relationship. As with the notification
muting, this will now return an object ({reblogs: [true|false]}) or
false for each follow relationship when requesting relationship
information for an account. This should cause few issues due to an
object being truthy in many languages, but some modifications may
need to be made in pickier languages.
Database changes: adds a show_reblogs column (default true,
non-nullable) to the follows and follow_requests tables. Because
these are non-nullable, we use the existing MigrationHelpers to
perform this change without locking those tables, although the
tables are likely to be small anyway.
Tests included.
See also <https://github.com/glitch-soc/mastodon/pull/212>.
* Rubocop fixes
* Code review changes
* Test fixes
This patchset closes #648 and resolves #3271.
* Rubocop fix
* Revert reblogs defaulting in argument, fix tests
It turns out we needed this for the same reason we needed it in muting:
if nil gets passed in somehow (most usually by an API client not passing
any value), we need to detect and handle it.
We could specify a default in the parameter and then also catch nil, but
there's no great reason to duplicate the default value.
2017-11-29 01:00:35 +11:00
t . boolean " show_reblogs " , default : true , null : false
2018-05-05 05:14:34 +10:00
t . string " uri "
2017-06-03 00:18:54 +10:00
t . index [ " account_id " , " target_account_id " ] , name : " index_follows_on_account_id_and_target_account_id " , unique : true
2018-08-22 04:11:34 +10:00
t . index [ " target_account_id " ] , name : " index_follows_on_target_account_id "
2016-02-23 02:00:20 +11:00
end
2018-11-27 23:56:25 +11:00
create_table " identities " , force : :cascade do | t |
2018-02-04 15:42:13 +11:00
t . string " provider " , default : " " , null : false
t . string " uid " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2018-11-27 23:56:25 +11:00
t . bigint " user_id "
2018-02-04 15:42:13 +11:00
t . index [ " user_id " ] , name : " index_identities_on_user_id "
end
2017-09-22 21:20:04 +10:00
create_table " imports " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . integer " type " , null : false
2017-07-13 11:12:25 +10:00
t . boolean " approved " , default : false , null : false
2017-06-03 00:18:54 +10:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . string " data_file_name "
t . string " data_content_type "
t . integer " data_file_size "
2017-03-31 04:42:33 +11:00
t . datetime " data_updated_at "
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id " , null : false
2019-02-03 13:59:51 +11:00
t . boolean " overwrite " , default : false , null : false
2017-03-31 04:42:33 +11:00
end
2017-11-28 02:07:59 +11:00
create_table " invites " , force : :cascade do | t |
2018-02-08 02:35:44 +11:00
t . bigint " user_id " , null : false
2017-11-28 02:07:59 +11:00
t . string " code " , default : " " , null : false
t . datetime " expires_at "
t . integer " max_uses "
t . integer " uses " , default : 0 , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2018-06-16 02:00:23 +10:00
t . boolean " autofollow " , default : false , null : false
2019-08-19 19:40:42 +10:00
t . text " comment "
2017-11-28 02:07:59 +11:00
t . index [ " code " ] , name : " index_invites_on_code " , unique : true
t . index [ " user_id " ] , name : " index_invites_on_user_id "
end
2017-11-18 10:16:48 +11:00
create_table " list_accounts " , force : :cascade do | t |
t . bigint " list_id " , null : false
t . bigint " account_id " , null : false
2019-11-04 23:02:01 +11:00
t . bigint " follow_id "
2017-11-18 10:16:48 +11:00
t . index [ " account_id " , " list_id " ] , name : " index_list_accounts_on_account_id_and_list_id " , unique : true
t . index [ " follow_id " ] , name : " index_list_accounts_on_follow_id "
t . index [ " list_id " , " account_id " ] , name : " index_list_accounts_on_list_id_and_account_id "
end
create_table " lists " , force : :cascade do | t |
2017-12-12 14:11:17 +11:00
t . bigint " account_id " , null : false
2017-11-18 10:16:48 +11:00
t . string " title " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " ] , name : " index_lists_on_account_id "
end
2019-09-06 21:55:51 +10:00
create_table " markers " , force : :cascade do | t |
t . bigint " user_id "
t . string " timeline " , default : " " , null : false
t . bigint " last_read_id " , default : 0 , null : false
t . integer " lock_version " , default : 0 , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " user_id " , " timeline " ] , name : " index_markers_on_user_id_and_timeline " , unique : true
t . index [ " user_id " ] , name : " index_markers_on_user_id "
end
2017-09-22 21:20:04 +10:00
create_table " media_attachments " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . bigint " status_id "
t . string " file_file_name "
t . string " file_content_type "
t . integer " file_file_size "
2016-09-06 01:46:36 +10:00
t . datetime " file_updated_at "
2017-06-03 00:18:54 +10:00
t . string " remote_url " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . string " shortcode "
t . integer " type " , default : 0 , null : false
t . json " file_meta "
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id "
2017-09-28 23:31:31 +10:00
t . text " description "
2019-01-05 22:43:28 +11:00
t . bigint " scheduled_status_id "
2019-04-27 11:24:09 +10:00
t . string " blurhash "
2020-03-09 09:56:18 +11:00
t . integer " processing "
2017-06-03 00:18:54 +10:00
t . index [ " account_id " ] , name : " index_media_attachments_on_account_id "
2019-01-05 22:43:28 +11:00
t . index [ " scheduled_status_id " ] , name : " index_media_attachments_on_scheduled_status_id "
2017-06-03 00:18:54 +10:00
t . index [ " shortcode " ] , name : " index_media_attachments_on_shortcode " , unique : true
t . index [ " status_id " ] , name : " index_media_attachments_on_status_id "
2016-09-06 01:46:36 +10:00
end
2017-09-22 21:20:04 +10:00
create_table " mentions " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . bigint " status_id "
2016-02-25 10:17:01 +11:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id "
2018-10-18 02:13:04 +11:00
t . boolean " silent " , default : false , null : false
2017-06-03 00:18:54 +10:00
t . index [ " account_id " , " status_id " ] , name : " index_mentions_on_account_id_and_status_id " , unique : true
t . index [ " status_id " ] , name : " index_mentions_on_status_id "
2016-02-25 10:17:01 +11:00
end
2017-09-22 21:20:04 +10:00
create_table " mutes " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2019-05-10 06:03:03 +10:00
t . boolean " hide_notifications " , default : true , null : false
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id " , null : false
t . bigint " target_account_id " , null : false
2017-06-03 00:18:54 +10:00
t . index [ " account_id " , " target_account_id " ] , name : " index_mutes_on_account_id_and_target_account_id " , unique : true
2018-08-22 04:11:34 +10:00
t . index [ " target_account_id " ] , name : " index_mutes_on_target_account_id "
2017-02-06 12:51:56 +11:00
end
2017-09-22 21:20:04 +10:00
create_table " notifications " , force : :cascade do | t |
2018-03-24 22:51:28 +11:00
t . bigint " activity_id " , null : false
t . string " activity_type " , null : false
2017-06-03 00:18:54 +10:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2018-03-24 22:51:28 +11:00
t . bigint " account_id " , null : false
t . bigint " from_account_id " , null : false
2017-06-03 00:18:54 +10:00
t . index [ " account_id " , " activity_id " , " activity_type " ] , name : " account_activity " , unique : true
2017-12-27 03:56:31 +11:00
t . index [ " account_id " , " id " ] , name : " index_notifications_on_account_id_and_id " , order : { id : :desc }
2017-06-03 00:18:54 +10:00
t . index [ " activity_id " , " activity_type " ] , name : " index_notifications_on_activity_id_and_activity_type "
2018-08-22 04:11:34 +10:00
t . index [ " from_account_id " ] , name : " index_notifications_on_from_account_id "
2016-11-20 10:33:02 +11:00
end
2017-09-22 21:20:04 +10:00
create_table " oauth_access_grants " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . string " token " , null : false
t . integer " expires_in " , null : false
t . text " redirect_uri " , null : false
t . datetime " created_at " , null : false
2016-03-07 22:42:33 +11:00
t . datetime " revoked_at "
2017-06-03 00:18:54 +10:00
t . string " scopes "
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " application_id " , null : false
t . bigint " resource_owner_id " , null : false
2018-08-22 04:11:34 +10:00
t . index [ " resource_owner_id " ] , name : " index_oauth_access_grants_on_resource_owner_id "
2017-06-03 00:18:54 +10:00
t . index [ " token " ] , name : " index_oauth_access_grants_on_token " , unique : true
2016-03-07 22:42:33 +11:00
end
2017-09-22 21:20:04 +10:00
create_table " oauth_access_tokens " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . string " token " , null : false
t . string " refresh_token "
t . integer " expires_in "
2016-03-07 22:42:33 +11:00
t . datetime " revoked_at "
2017-06-03 00:18:54 +10:00
t . datetime " created_at " , null : false
t . string " scopes "
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " application_id "
t . bigint " resource_owner_id "
2017-06-03 00:18:54 +10:00
t . index [ " refresh_token " ] , name : " index_oauth_access_tokens_on_refresh_token " , unique : true
t . index [ " resource_owner_id " ] , name : " index_oauth_access_tokens_on_resource_owner_id "
t . index [ " token " ] , name : " index_oauth_access_tokens_on_token " , unique : true
2016-03-07 22:42:33 +11:00
end
2017-09-22 21:20:04 +10:00
create_table " oauth_applications " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . string " name " , null : false
t . string " uid " , null : false
t . string " secret " , null : false
t . text " redirect_uri " , null : false
t . string " scopes " , default : " " , null : false
2016-03-07 22:42:33 +11:00
t . datetime " created_at "
t . datetime " updated_at "
2017-08-23 08:59:35 +10:00
t . boolean " superapp " , default : false , null : false
t . string " website "
t . string " owner_type "
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " owner_id "
2018-08-15 04:24:47 +10:00
t . boolean " confidential " , default : true , null : false
2017-08-23 08:59:35 +10:00
t . index [ " owner_id " , " owner_type " ] , name : " index_oauth_applications_on_owner_id_and_owner_type "
2017-06-03 00:18:54 +10:00
t . index [ " uid " ] , name : " index_oauth_applications_on_uid " , unique : true
2016-03-07 22:42:33 +11:00
end
2018-10-10 04:35:14 +11:00
create_table " pghero_space_stats " , force : :cascade do | t |
t . text " database "
t . text " schema "
t . text " relation "
t . bigint " size "
t . datetime " captured_at "
t . index [ " database " , " captured_at " ] , name : " index_pghero_space_stats_on_database_and_captured_at "
end
2019-03-04 08:18:23 +11:00
create_table " poll_votes " , force : :cascade do | t |
t . bigint " account_id "
t . bigint " poll_id "
t . integer " choice " , default : 0 , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2019-03-05 08:51:23 +11:00
t . string " uri "
2019-03-04 08:18:23 +11:00
t . index [ " account_id " ] , name : " index_poll_votes_on_account_id "
t . index [ " poll_id " ] , name : " index_poll_votes_on_poll_id "
end
create_table " polls " , force : :cascade do | t |
t . bigint " account_id "
t . bigint " status_id "
t . datetime " expires_at "
t . string " options " , default : [ ] , null : false , array : true
t . bigint " cached_tallies " , default : [ ] , null : false , array : true
t . boolean " multiple " , default : false , null : false
t . boolean " hide_totals " , default : false , null : false
t . bigint " votes_count " , default : 0 , null : false
t . datetime " last_fetched_at "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2019-03-07 05:53:57 +11:00
t . integer " lock_version " , default : 0 , null : false
2019-09-30 06:58:01 +10:00
t . bigint " voters_count "
2019-03-04 08:18:23 +11:00
t . index [ " account_id " ] , name : " index_polls_on_account_id "
t . index [ " status_id " ] , name : " index_polls_on_status_id "
end
2017-09-02 00:20:16 +10:00
create_table " preview_cards " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . string " url " , default : " " , null : false
2017-09-02 00:20:16 +10:00
t . string " title " , default : " " , null : false
t . string " description " , default : " " , null : false
2017-06-03 00:18:54 +10:00
t . string " image_file_name "
t . string " image_content_type "
t . integer " image_file_size "
2017-01-20 11:00:14 +11:00
t . datetime " image_updated_at "
2017-06-03 00:18:54 +10:00
t . integer " type " , default : 0 , null : false
t . text " html " , default : " " , null : false
t . string " author_name " , default : " " , null : false
t . string " author_url " , default : " " , null : false
t . string " provider_name " , default : " " , null : false
t . string " provider_url " , default : " " , null : false
t . integer " width " , default : 0 , null : false
t . integer " height " , default : 0 , null : false
2017-09-02 00:20:16 +10:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2017-12-07 13:37:43 +11:00
t . string " embed_url " , default : " " , null : false
2017-09-02 00:20:16 +10:00
t . index [ " url " ] , name : " index_preview_cards_on_url " , unique : true
end
create_table " preview_cards_statuses " , id : false , force : :cascade do | t |
t . bigint " preview_card_id " , null : false
t . bigint " status_id " , null : false
t . index [ " status_id " , " preview_card_id " ] , name : " index_preview_cards_statuses_on_status_id_and_preview_card_id "
2017-01-20 11:00:14 +11:00
end
2018-07-13 10:16:06 +10:00
create_table " relays " , force : :cascade do | t |
t . string " inbox_url " , default : " " , null : false
t . string " follow_activity_id "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2018-08-13 02:16:26 +10:00
t . integer " state " , default : 0 , null : false
2018-07-13 10:16:06 +10:00
end
2018-04-03 06:04:14 +10:00
create_table " report_notes " , force : :cascade do | t |
t . text " content " , null : false
t . bigint " report_id " , null : false
t . bigint " account_id " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " account_id " ] , name : " index_report_notes_on_account_id "
t . index [ " report_id " ] , name : " index_report_notes_on_report_id "
end
2017-09-22 21:20:04 +10:00
create_table " reports " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . bigint " status_ids " , default : [ ] , null : false , array : true
t . text " comment " , default : " " , null : false
t . boolean " action_taken " , default : false , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id " , null : false
2017-09-22 21:20:04 +10:00
t . bigint " action_taken_by_account_id "
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " target_account_id " , null : false
2018-04-03 06:04:14 +10:00
t . bigint " assigned_account_id "
2019-03-18 01:34:56 +11:00
t . string " uri "
2017-06-03 00:18:54 +10:00
t . index [ " account_id " ] , name : " index_reports_on_account_id "
t . index [ " target_account_id " ] , name : " index_reports_on_target_account_id "
2017-02-15 06:59:26 +11:00
end
2019-01-05 22:43:28 +11:00
create_table " scheduled_statuses " , force : :cascade do | t |
t . bigint " account_id "
t . datetime " scheduled_at "
t . jsonb " params "
t . index [ " account_id " ] , name : " index_scheduled_statuses_on_account_id "
t . index [ " scheduled_at " ] , name : " index_scheduled_statuses_on_scheduled_at "
end
2017-06-24 02:50:53 +10:00
create_table " session_activations " , force : :cascade do | t |
t . string " session_id " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2017-06-26 00:54:30 +10:00
t . string " user_agent " , default : " " , null : false
t . inet " ip "
2017-09-22 21:20:04 +10:00
t . bigint " access_token_id "
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " user_id " , null : false
2017-09-22 21:20:04 +10:00
t . bigint " web_push_subscription_id "
2018-08-22 04:11:34 +10:00
t . index [ " access_token_id " ] , name : " index_session_activations_on_access_token_id "
2017-06-24 02:50:53 +10:00
t . index [ " session_id " ] , name : " index_session_activations_on_session_id " , unique : true
t . index [ " user_id " ] , name : " index_session_activations_on_user_id "
end
2017-09-22 21:20:04 +10:00
create_table " settings " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . string " var " , null : false
t . text " value "
t . string " thing_type "
2016-10-07 22:17:56 +11:00
t . datetime " created_at "
t . datetime " updated_at "
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " thing_id "
2017-06-03 00:18:54 +10:00
t . index [ " thing_type " , " thing_id " , " var " ] , name : " index_settings_on_thing_type_and_thing_id_and_var " , unique : true
2016-10-07 22:17:56 +11:00
end
2017-09-14 08:04:30 +10:00
create_table " site_uploads " , force : :cascade do | t |
t . string " var " , default : " " , null : false
t . string " file_file_name "
t . string " file_content_type "
t . integer " file_file_size "
t . datetime " file_updated_at "
t . json " meta "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " var " ] , name : " index_site_uploads_on_var " , unique : true
end
2017-08-25 09:41:18 +10:00
create_table " status_pins " , force : :cascade do | t |
t . bigint " account_id " , null : false
t . bigint " status_id " , null : false
2017-08-26 02:50:52 +10:00
t . datetime " created_at " , default : - > { " now() " } , null : false
t . datetime " updated_at " , default : - > { " now() " } , null : false
2017-08-25 09:41:18 +10:00
t . index [ " account_id " , " status_id " ] , name : " index_status_pins_on_account_id_and_status_id " , unique : true
end
2018-08-15 03:19:32 +10:00
create_table " status_stats " , force : :cascade do | t |
t . bigint " status_id " , null : false
t . bigint " replies_count " , default : 0 , null : false
t . bigint " reblogs_count " , default : 0 , null : false
t . bigint " favourites_count " , default : 0 , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " status_id " ] , name : " index_status_stats_on_status_id " , unique : true
end
Non-Serial ("Snowflake") IDs (#4801)
* 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_r139336735
https://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.
2017-10-04 18:56:37 +11:00
create_table " statuses " , id : :bigint , default : - > { " timestamp_id('statuses'::text) " } , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . string " uri "
t . text " text " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . bigint " in_reply_to_id "
t . bigint " reblog_of_id "
t . string " url "
2017-07-13 11:12:25 +10:00
t . boolean " sensitive " , default : false , null : false
2017-06-03 00:18:54 +10:00
t . integer " visibility " , default : 0 , null : false
t . text " spoiler_text " , default : " " , null : false
2017-07-13 11:12:25 +10:00
t . boolean " reply " , default : false , null : false
2017-06-10 02:09:37 +10:00
t . string " language "
2017-06-03 00:18:54 +10:00
t . bigint " conversation_id "
2017-09-07 04:57:52 +10:00
t . boolean " local "
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id " , null : false
t . bigint " application_id "
t . bigint " in_reply_to_account_id "
2019-03-04 08:18:23 +11:00
t . bigint " poll_id "
2019-08-23 05:55:56 +10:00
t . datetime " deleted_at "
t . index [ " account_id " , " id " , " visibility " , " updated_at " ] , name : " index_statuses_20190820 " , order : { id : :desc } , where : " (deleted_at IS NULL) "
2019-08-24 12:12:27 +10:00
t . index [ " id " , " account_id " ] , name : " index_statuses_local_20190824 " , order : { id : :desc } , where : " ((local OR (uri IS NULL)) AND (deleted_at IS NULL) AND (visibility = 0) AND (reblog_of_id IS NULL) AND ((NOT reply) OR (in_reply_to_account_id = account_id))) "
2020-01-22 04:56:34 +11:00
t . index [ " id " , " account_id " ] , name : " index_statuses_public_20200119 " , order : { id : :desc } , where : " ((deleted_at IS NULL) AND (visibility = 0) AND (reblog_of_id IS NULL) AND ((NOT reply) OR (in_reply_to_account_id = account_id))) "
2018-08-22 04:11:34 +10:00
t . index [ " in_reply_to_account_id " ] , name : " index_statuses_on_in_reply_to_account_id "
2017-06-03 00:18:54 +10:00
t . index [ " in_reply_to_id " ] , name : " index_statuses_on_in_reply_to_id "
2017-11-28 06:22:27 +11:00
t . index [ " reblog_of_id " , " account_id " ] , name : " index_statuses_on_reblog_of_id_and_account_id "
2017-06-03 00:18:54 +10:00
t . index [ " uri " ] , name : " index_statuses_on_uri " , unique : true
2016-11-06 01:20:05 +11:00
end
create_table " statuses_tags " , id : false , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . bigint " status_id " , null : false
2017-09-22 21:20:04 +10:00
t . bigint " tag_id " , null : false
2017-06-03 00:18:54 +10:00
t . index [ " status_id " ] , name : " index_statuses_tags_on_status_id "
t . index [ " tag_id " , " status_id " ] , name : " index_statuses_tags_on_tag_id_and_status_id " , unique : true
2016-02-21 08:53:20 +11:00
end
2017-09-22 21:20:04 +10:00
create_table " tags " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . string " name " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2019-08-06 03:54:29 +10:00
t . boolean " usable "
t . boolean " trendable "
t . boolean " listable "
t . datetime " reviewed_at "
t . datetime " requested_review_at "
2019-08-18 11:45:51 +10:00
t . datetime " last_status_at "
2019-09-03 02:11:13 +10:00
t . float " max_score "
t . datetime " max_score_at "
2019-07-28 13:59:51 +10:00
t . index " lower((name)::text) " , name : " index_tags_on_name_lower " , unique : true
2016-11-05 05:12:59 +11:00
end
2019-01-19 01:56:55 +11:00
create_table " tombstones " , force : :cascade do | t |
t . bigint " account_id "
t . string " uri " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2019-05-10 06:03:03 +10:00
t . boolean " by_moderator "
2019-01-19 01:56:55 +11:00
t . index [ " account_id " ] , name : " index_tombstones_on_account_id "
t . index [ " uri " ] , name : " index_tombstones_on_uri "
end
2019-04-10 00:06:30 +10:00
create_table " user_invite_requests " , force : :cascade do | t |
t . bigint " user_id "
t . text " text "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . index [ " user_id " ] , name : " index_user_invite_requests_on_user_id "
end
2017-09-22 21:20:04 +10:00
create_table " users " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . string " email " , default : " " , null : false
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
t . string " encrypted_password " , default : " " , null : false
t . string " reset_password_token "
2016-03-05 23:12:24 +11:00
t . datetime " reset_password_sent_at "
t . datetime " remember_created_at "
2017-06-03 00:18:54 +10:00
t . integer " sign_in_count " , default : 0 , null : false
2016-03-05 23:12:24 +11:00
t . datetime " current_sign_in_at "
t . datetime " last_sign_in_at "
2017-06-03 00:18:54 +10:00
t . inet " current_sign_in_ip "
t . inet " last_sign_in_ip "
2017-07-13 11:12:25 +10:00
t . boolean " admin " , default : false , null : false
2017-06-03 00:18:54 +10:00
t . string " confirmation_token "
2016-10-04 01:38:22 +11:00
t . datetime " confirmed_at "
t . datetime " confirmation_sent_at "
2017-06-03 00:18:54 +10:00
t . string " unconfirmed_email "
t . string " locale "
t . string " encrypted_otp_secret "
t . string " encrypted_otp_secret_iv "
t . string " encrypted_otp_secret_salt "
t . integer " consumed_timestep "
2017-07-13 11:12:25 +10:00
t . boolean " otp_required_for_login " , default : false , null : false
2017-03-04 09:45:48 +11:00
t . datetime " last_emailed_at "
2017-06-03 00:18:54 +10:00
t . string " otp_backup_codes " , array : true
t . string " filtered_languages " , default : [ ] , null : false , array : true
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
t . bigint " account_id " , null : false
2017-11-08 05:06:44 +11:00
t . boolean " disabled " , default : false , null : false
2017-11-12 06:23:33 +11:00
t . boolean " moderator " , default : false , null : false
2017-11-28 02:07:59 +11:00
t . bigint " invite_id "
2018-02-02 20:18:55 +11:00
t . string " remember_token "
2018-06-17 21:54:02 +10:00
t . string " chosen_languages " , array : true
2018-12-25 05:12:38 +11:00
t . bigint " created_by_application_id "
2019-03-14 15:28:30 +11:00
t . boolean " approved " , default : true , null : false
2017-06-03 00:18:54 +10:00
t . index [ " account_id " ] , name : " index_users_on_account_id "
t . index [ " confirmation_token " ] , name : " index_users_on_confirmation_token " , unique : true
2018-12-25 05:12:38 +11:00
t . index [ " created_by_application_id " ] , name : " index_users_on_created_by_application_id "
2017-06-03 00:18:54 +10:00
t . index [ " email " ] , name : " index_users_on_email " , unique : true
2019-09-18 18:58:08 +10:00
t . index [ " remember_token " ] , name : " index_users_on_remember_token " , unique : true
2017-06-03 00:18:54 +10:00
t . index [ " reset_password_token " ] , name : " index_users_on_reset_password_token " , unique : true
2016-02-23 02:00:20 +11:00
end
2017-07-14 06:15:32 +10:00
create_table " web_push_subscriptions " , force : :cascade do | t |
t . string " endpoint " , null : false
t . string " key_p256dh " , null : false
t . string " key_auth " , null : false
t . json " data "
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2018-05-11 19:49:12 +10:00
t . bigint " access_token_id "
t . bigint " user_id "
t . index [ " access_token_id " ] , name : " index_web_push_subscriptions_on_access_token_id "
t . index [ " user_id " ] , name : " index_web_push_subscriptions_on_user_id "
2017-07-14 06:15:32 +10:00
end
2017-09-22 21:20:04 +10:00
create_table " web_settings " , force : :cascade do | t |
2017-06-03 00:18:54 +10:00
t . json " data "
2017-01-10 00:00:55 +11:00
t . datetime " created_at " , null : false
t . datetime " updated_at " , null : false
2018-02-08 02:35:44 +11:00
t . bigint " user_id " , null : false
2017-06-03 00:18:54 +10:00
t . index [ " user_id " ] , name : " index_web_settings_on_user_id " , unique : true
2017-01-10 00:00:55 +11:00
end
2019-09-20 04:58:19 +10:00
add_foreign_key " account_aliases " , " accounts " , on_delete : :cascade
2018-10-08 08:44:58 +11:00
add_foreign_key " account_conversations " , " accounts " , on_delete : :cascade
add_foreign_key " account_conversations " , " conversations " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " account_domain_blocks " , " accounts " , name : " fk_206c6029bd " , on_delete : :cascade
2019-03-19 07:00:55 +11:00
add_foreign_key " account_identity_proofs " , " accounts " , on_delete : :cascade
2019-09-20 04:58:19 +10:00
add_foreign_key " account_migrations " , " accounts " , column : " target_account_id " , on_delete : :nullify
add_foreign_key " account_migrations " , " accounts " , on_delete : :cascade
2017-10-10 22:12:17 +11:00
add_foreign_key " account_moderation_notes " , " accounts "
2017-10-08 05:26:43 +11:00
add_foreign_key " account_moderation_notes " , " accounts " , column : " target_account_id "
2018-08-09 17:56:53 +10:00
add_foreign_key " account_pins " , " accounts " , column : " target_account_id " , on_delete : :cascade
add_foreign_key " account_pins " , " accounts " , on_delete : :cascade
2018-11-19 10:43:52 +11:00
add_foreign_key " account_stats " , " accounts " , on_delete : :cascade
2018-12-07 03:36:11 +11:00
add_foreign_key " account_tag_stats " , " tags " , on_delete : :cascade
Add moderation warnings (#9519)
* Add moderation warnings
Replace individual routes for disabling, silencing, and suspending
a user, as well as the report update route, with a unified account
action controller that allows you to select an action (none,
disable, silence, suspend) as well as whether it should generate an
e-mail notification with optional custom text. That notification,
with the optional custom text, is saved as a warning.
Additionally, there are warning presets you can configure to save
time when performing the above.
* Use Account#local_username_and_domain
2018-12-23 06:02:09 +11:00
add_foreign_key " account_warnings " , " accounts " , column : " target_account_id " , on_delete : :cascade
add_foreign_key " account_warnings " , " accounts " , on_delete : :nullify
2017-11-19 05:39:02 +11:00
add_foreign_key " accounts " , " accounts " , column : " moved_to_account_id " , on_delete : :nullify
2017-11-24 12:05:53 +11:00
add_foreign_key " admin_action_logs " , " accounts " , on_delete : :cascade
2020-01-24 08:00:13 +11:00
add_foreign_key " announcement_mutes " , " accounts " , on_delete : :cascade
add_foreign_key " announcement_mutes " , " announcements " , on_delete : :cascade
add_foreign_key " announcement_reactions " , " accounts " , on_delete : :cascade
add_foreign_key " announcement_reactions " , " announcements " , on_delete : :cascade
add_foreign_key " announcement_reactions " , " custom_emojis " , on_delete : :cascade
2018-02-28 16:54:55 +11:00
add_foreign_key " backups " , " users " , on_delete : :nullify
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " blocks " , " accounts " , column : " target_account_id " , name : " fk_9571bfabc1 " , on_delete : :cascade
add_foreign_key " blocks " , " accounts " , name : " fk_4269e03e65 " , on_delete : :cascade
2019-11-14 09:02:10 +11:00
add_foreign_key " bookmarks " , " accounts " , on_delete : :cascade
add_foreign_key " bookmarks " , " statuses " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " conversation_mutes " , " accounts " , name : " fk_225b4212bb " , on_delete : :cascade
2017-06-05 21:24:00 +10:00
add_foreign_key " conversation_mutes " , " conversations " , on_delete : :cascade
2018-06-29 23:34:36 +10:00
add_foreign_key " custom_filters " , " accounts " , on_delete : :cascade
2020-03-13 08:35:20 +11:00
add_foreign_key " email_domain_blocks " , " email_domain_blocks " , column : " parent_id " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " favourites " , " accounts " , name : " fk_5eb6c2b873 " , on_delete : :cascade
add_foreign_key " favourites " , " statuses " , name : " fk_b0e856845e " , on_delete : :cascade
2019-02-04 14:25:59 +11:00
add_foreign_key " featured_tags " , " accounts " , on_delete : :cascade
add_foreign_key " featured_tags " , " tags " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " follow_requests " , " accounts " , column : " target_account_id " , name : " fk_9291ec025d " , on_delete : :cascade
add_foreign_key " follow_requests " , " accounts " , name : " fk_76d644b0e7 " , on_delete : :cascade
add_foreign_key " follows " , " accounts " , column : " target_account_id " , name : " fk_745ca29eac " , on_delete : :cascade
add_foreign_key " follows " , " accounts " , name : " fk_32ed1b5560 " , on_delete : :cascade
2018-11-27 23:56:25 +11:00
add_foreign_key " identities " , " users " , name : " fk_bea040f377 " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " imports " , " accounts " , name : " fk_6db1b6e408 " , on_delete : :cascade
2017-11-28 02:07:59 +11:00
add_foreign_key " invites " , " users " , on_delete : :cascade
2017-11-18 10:16:48 +11:00
add_foreign_key " list_accounts " , " accounts " , on_delete : :cascade
add_foreign_key " list_accounts " , " follows " , on_delete : :cascade
add_foreign_key " list_accounts " , " lists " , on_delete : :cascade
add_foreign_key " lists " , " accounts " , on_delete : :cascade
2019-09-06 21:55:51 +10:00
add_foreign_key " markers " , " users " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " media_attachments " , " accounts " , name : " fk_96dd81e81b " , on_delete : :nullify
2019-01-05 22:43:28 +11:00
add_foreign_key " media_attachments " , " scheduled_statuses " , on_delete : :nullify
2017-06-05 21:24:00 +10:00
add_foreign_key " media_attachments " , " statuses " , on_delete : :nullify
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " mentions " , " accounts " , name : " fk_970d43f9d1 " , on_delete : :cascade
2017-06-05 21:24:00 +10:00
add_foreign_key " mentions " , " statuses " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " mutes " , " accounts " , column : " target_account_id " , name : " fk_eecff219ea " , on_delete : :cascade
add_foreign_key " mutes " , " accounts " , name : " fk_b8d8daf315 " , on_delete : :cascade
add_foreign_key " notifications " , " accounts " , column : " from_account_id " , name : " fk_fbd6b0bf9e " , on_delete : :cascade
add_foreign_key " notifications " , " accounts " , name : " fk_c141c8ee55 " , on_delete : :cascade
add_foreign_key " oauth_access_grants " , " oauth_applications " , column : " application_id " , name : " fk_34d54b0a33 " , on_delete : :cascade
add_foreign_key " oauth_access_grants " , " users " , column : " resource_owner_id " , name : " fk_63b044929b " , on_delete : :cascade
add_foreign_key " oauth_access_tokens " , " oauth_applications " , column : " application_id " , name : " fk_f5fc4c1ee3 " , on_delete : :cascade
add_foreign_key " oauth_access_tokens " , " users " , column : " resource_owner_id " , name : " fk_e84df68546 " , on_delete : :cascade
add_foreign_key " oauth_applications " , " users " , column : " owner_id " , name : " fk_b0988c7c0a " , on_delete : :cascade
2019-03-04 08:18:23 +11:00
add_foreign_key " poll_votes " , " accounts " , on_delete : :cascade
add_foreign_key " poll_votes " , " polls " , on_delete : :cascade
add_foreign_key " polls " , " accounts " , on_delete : :cascade
add_foreign_key " polls " , " statuses " , on_delete : :cascade
2018-04-03 06:04:14 +10:00
add_foreign_key " report_notes " , " accounts " , on_delete : :cascade
add_foreign_key " report_notes " , " reports " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " reports " , " accounts " , column : " action_taken_by_account_id " , name : " fk_bca45b75fd " , on_delete : :nullify
2018-04-03 06:04:14 +10:00
add_foreign_key " reports " , " accounts " , column : " assigned_account_id " , on_delete : :nullify
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " reports " , " accounts " , column : " target_account_id " , name : " fk_eb37af34f0 " , on_delete : :cascade
add_foreign_key " reports " , " accounts " , name : " fk_4b81f7522c " , on_delete : :cascade
2019-01-05 22:43:28 +11:00
add_foreign_key " scheduled_statuses " , " accounts " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " session_activations " , " oauth_access_tokens " , column : " access_token_id " , name : " fk_957e5bda89 " , on_delete : :cascade
add_foreign_key " session_activations " , " users " , name : " fk_e5fda67334 " , on_delete : :cascade
add_foreign_key " status_pins " , " accounts " , name : " fk_d4cb435b62 " , on_delete : :cascade
2017-08-25 09:41:18 +10:00
add_foreign_key " status_pins " , " statuses " , on_delete : :cascade
2018-08-15 03:19:32 +10:00
add_foreign_key " status_stats " , " statuses " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " statuses " , " accounts " , column : " in_reply_to_account_id " , name : " fk_c7fa917661 " , on_delete : :nullify
add_foreign_key " statuses " , " accounts " , name : " fk_9bda1543f7 " , on_delete : :cascade
2017-06-05 21:24:00 +10:00
add_foreign_key " statuses " , " statuses " , column : " in_reply_to_id " , on_delete : :nullify
2017-02-17 12:33:10 +11:00
add_foreign_key " statuses " , " statuses " , column : " reblog_of_id " , on_delete : :cascade
2017-06-05 21:24:00 +10:00
add_foreign_key " statuses_tags " , " statuses " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " statuses_tags " , " tags " , name : " fk_3081861e21 " , on_delete : :cascade
2019-01-19 01:56:55 +11:00
add_foreign_key " tombstones " , " accounts " , on_delete : :cascade
2019-04-10 00:06:30 +10:00
add_foreign_key " user_invite_requests " , " users " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " users " , " accounts " , name : " fk_50500f500d " , on_delete : :cascade
2017-11-28 02:07:59 +11:00
add_foreign_key " users " , " invites " , on_delete : :nullify
2018-12-25 05:12:38 +11:00
add_foreign_key " users " , " oauth_applications " , column : " created_by_application_id " , on_delete : :nullify
2018-05-11 19:49:12 +10:00
add_foreign_key " web_push_subscriptions " , " oauth_access_tokens " , column : " access_token_id " , on_delete : :cascade
add_foreign_key " web_push_subscriptions " , " users " , on_delete : :cascade
Make IdsToBigints (mostly!) non-blocking (#5088)
* 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
2017-10-03 06:28:59 +11:00
add_foreign_key " web_settings " , " users " , name : " fk_11910667b2 " , on_delete : :cascade
2017-04-27 22:42:22 +10:00
end