Change tootctl to use inline parallelization instead of Sidekiq (#11776)

- Remove --background option
- Add --concurrency(=5) option
- Add progress bars
This commit is contained in:
Eugen Rochko 2019-09-10 13:48:48 +02:00 committed by GitHub
parent 9045f5e3f8
commit 8674814825
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
15 changed files with 200 additions and 337 deletions

View file

@ -77,6 +77,7 @@ gem 'rails-settings-cached', '~> 0.6'
gem 'redis', '~> 4.1', require: ['redis', 'redis/connection/hiredis'] gem 'redis', '~> 4.1', require: ['redis', 'redis/connection/hiredis']
gem 'mario-redis-lock', '~> 1.2', require: 'redis_lock' gem 'mario-redis-lock', '~> 1.2', require: 'redis_lock'
gem 'rqrcode', '~> 0.10' gem 'rqrcode', '~> 0.10'
gem 'ruby-progressbar', '~> 1.10'
gem 'sanitize', '~> 5.1' gem 'sanitize', '~> 5.1'
gem 'sidekiq', '~> 5.2' gem 'sidekiq', '~> 5.2'
gem 'sidekiq-scheduler', '~> 3.0' gem 'sidekiq-scheduler', '~> 3.0'

View file

@ -769,6 +769,7 @@ DEPENDENCIES
rspec-sidekiq (~> 3.0) rspec-sidekiq (~> 3.0)
rubocop (~> 0.74) rubocop (~> 0.74)
rubocop-rails (~> 2.3) rubocop-rails (~> 2.3)
ruby-progressbar (~> 1.10)
sanitize (~> 5.1) sanitize (~> 5.1)
sidekiq (~> 5.2) sidekiq (~> 5.2)
sidekiq-bulk (~> 0.2.0) sidekiq-bulk (~> 0.2.0)

View file

@ -129,6 +129,7 @@ class MediaAttachment < ApplicationRecord
scope :unattached, -> { where(status_id: nil, scheduled_status_id: nil) } scope :unattached, -> { where(status_id: nil, scheduled_status_id: nil) }
scope :local, -> { where(remote_url: '') } scope :local, -> { where(remote_url: '') }
scope :remote, -> { where.not(remote_url: '') } scope :remote, -> { where.not(remote_url: '') }
scope :cached, -> { remote.where.not(file_file_name: nil) }
default_scope { order(id: :asc) } default_scope { order(id: :asc) }

View file

@ -43,6 +43,8 @@ class PreviewCard < ApplicationRecord
validates_attachment_size :image, less_than: LIMIT validates_attachment_size :image, less_than: LIMIT
remotable_attachment :image, LIMIT remotable_attachment :image, LIMIT
scope :cached, -> { where.not(image_file_name: [nil, '']) }
before_save :extract_dimensions, if: :link? before_save :extract_dimensions, if: :link?
def save_with_optional_image! def save_with_optional_image!

View file

@ -1,14 +0,0 @@
# frozen_string_literal: true
class Maintenance::DestroyMediaWorker
include Sidekiq::Worker
sidekiq_options queue: 'pull'
def perform(media_attachment_id)
media = media_attachment_id.is_a?(MediaAttachment) ? media_attachment_id : MediaAttachment.find(media_attachment_id)
media.destroy
rescue ActiveRecord::RecordNotFound
true
end
end

View file

@ -1,16 +0,0 @@
# frozen_string_literal: true
class Maintenance::RedownloadAccountMediaWorker
include Sidekiq::Worker
sidekiq_options queue: 'pull', retry: false
def perform(account_id)
account = account_id.is_a?(Account) ? account_id : Account.find(account_id)
account.reset_avatar!
account.reset_header!
account.save
rescue ActiveRecord::RecordNotFound
true
end
end

View file

@ -1,18 +0,0 @@
# frozen_string_literal: true
class Maintenance::UncacheMediaWorker
include Sidekiq::Worker
sidekiq_options queue: 'pull'
def perform(media_attachment_id)
media = media_attachment_id.is_a?(MediaAttachment) ? media_attachment_id : MediaAttachment.find(media_attachment_id)
return if media.file.blank?
media.file.destroy
media.save
rescue ActiveRecord::RecordNotFound
true
end
end

View file

@ -1,18 +0,0 @@
# frozen_string_literal: true
class Maintenance::UncachePreviewWorker
include Sidekiq::Worker
sidekiq_options queue: 'pull'
def perform(preview_card_id)
preview_card = PreviewCard.find(preview_card_id)
return if preview_card.image.blank?
preview_card.image.destroy
preview_card.save
rescue ActiveRecord::RecordNotFound
true
end
end

View file

@ -7,6 +7,8 @@ require_relative 'cli_helper'
module Mastodon module Mastodon
class AccountsCLI < Thor class AccountsCLI < Thor
include CLIHelper
def self.exit_on_failure? def self.exit_on_failure?
true true
end end
@ -26,18 +28,20 @@ module Mastodon
if options[:all] if options[:all]
processed = 0 processed = 0
delay = 0 delay = 0
scope = Account.local.without_suspended
progress = create_progress_bar(scope.count)
Account.local.without_suspended.find_in_batches do |accounts| scope.find_in_batches do |accounts|
accounts.each do |account| accounts.each do |account|
rotate_keys_for_account(account, delay) rotate_keys_for_account(account, delay)
progress.increment
processed += 1 processed += 1
say('.', :green, false)
end end
delay += 5.minutes delay += 5.minutes
end end
say progress.finish
say("OK, rotated keys for #{processed} accounts", :green) say("OK, rotated keys for #{processed} accounts", :green)
elsif username.present? elsif username.present?
rotate_keys_for_account(Account.find_local(username)) rotate_keys_for_account(Account.find_local(username))
@ -206,6 +210,8 @@ module Mastodon
say('OK', :green) say('OK', :green)
end end
option :concurrency, type: :numeric, default: 5, aliases: [:c]
option :verbose, type: :boolean, aliases: [:v]
option :dry_run, type: :boolean option :dry_run, type: :boolean
desc 'cull', 'Remove remote accounts that no longer exist' desc 'cull', 'Remove remote accounts that no longer exist'
long_desc <<-LONG_DESC long_desc <<-LONG_DESC
@ -215,63 +221,45 @@ module Mastodon
Accounts that have had confirmed activity within the last week Accounts that have had confirmed activity within the last week
are excluded from the checks. are excluded from the checks.
Domains that are unreachable are not checked.
With the --dry-run option, no deletes will actually be carried
out.
LONG_DESC LONG_DESC
def cull def cull
skip_threshold = 7.days.ago skip_threshold = 7.days.ago
culled = 0
dry_run_culled = []
skip_domains = Set.new
dry_run = options[:dry_run] ? ' (DRY RUN)' : '' dry_run = options[:dry_run] ? ' (DRY RUN)' : ''
skip_domains = Concurrent::Set.new
Account.remote.where(protocol: :activitypub).partitioned.find_each do |account| processed, culled = parallelize_with_progress(Account.remote.where(protocol: :activitypub).partitioned) do |account|
next if account.updated_at >= skip_threshold || (account.last_webfingered_at.present? && account.last_webfingered_at >= skip_threshold) next if account.updated_at >= skip_threshold || (account.last_webfingered_at.present? && account.last_webfingered_at >= skip_threshold) || skip_domains.include?(account.domain)
code = 0 code = 0
unless skip_domains.include?(account.domain)
begin begin
code = Request.new(:head, account.uri).perform(&:code) code = Request.new(:head, account.uri).perform(&:code)
rescue HTTP::ConnectionError rescue HTTP::ConnectionError
skip_domains << account.domain skip_domains << account.domain
rescue StandardError
next
end
end end
if [404, 410].include?(code) if [404, 410].include?(code)
if options[:dry_run] SuspendAccountService.new.call(account, destroy: true) unless options[:dry_run]
dry_run_culled << account.acct 1
else
SuspendAccountService.new.call(account, destroy: true)
end
culled += 1
say('+', :green, false)
else else
account.touch # Touch account even during dry run to avoid getting the account into the window again # Touch account even during dry run to avoid getting the account into the window again
say('.', nil, false) account.touch
end end
end end
say say("Visited #{processed} accounts, removed #{culled}#{dry_run}", :green)
say("Removed #{culled} accounts. #{skip_domains.size} servers skipped#{dry_run}", skip_domains.empty? ? :green : :yellow)
unless skip_domains.empty? unless skip_domains.empty?
say('The following servers were not available during the check:', :yellow) say('The following domains were not available during the check:', :yellow)
skip_domains.each { |domain| say(' ' + domain) } skip_domains.each { |domain| say(' ' + domain) }
end end
unless dry_run_culled.empty?
say('The following accounts would have been deleted:', :green)
dry_run_culled.each { |account| say(' ' + account) }
end
end end
option :all, type: :boolean option :all, type: :boolean
option :domain option :domain
option :concurrency, type: :numeric, default: 5, aliases: [:c]
option :verbose, type: :boolean, aliases: [:v]
option :dry_run, type: :boolean
desc 'refresh [USERNAME]', 'Fetch remote user data and files' desc 'refresh [USERNAME]', 'Fetch remote user data and files'
long_desc <<-LONG_DESC long_desc <<-LONG_DESC
Fetch remote user data and files for one or multiple accounts. Fetch remote user data and files for one or multiple accounts.
@ -280,21 +268,23 @@ module Mastodon
Through the --domain option, this can be narrowed down to a Through the --domain option, this can be narrowed down to a
specific domain only. Otherwise, a single remote account must specific domain only. Otherwise, a single remote account must
be specified with USERNAME. be specified with USERNAME.
All processing is done in the background through Sidekiq.
LONG_DESC LONG_DESC
def refresh(username = nil) def refresh(username = nil)
dry_run = options[:dry_run] ? ' (DRY RUN)' : ''
if options[:domain] || options[:all] if options[:domain] || options[:all]
queued = 0
scope = Account.remote scope = Account.remote
scope = scope.where(domain: options[:domain]) if options[:domain] scope = scope.where(domain: options[:domain]) if options[:domain]
scope.select(:id).reorder(nil).find_in_batches do |accounts| processed, = parallelize_with_progress(scope) do |account|
Maintenance::RedownloadAccountMediaWorker.push_bulk(accounts.map(&:id)) next if options[:dry_run]
queued += accounts.size
account.reset_avatar!
account.reset_header!
account.save
end end
say("Scheduled refreshment of #{queued} accounts", :green, true) say("Refreshed #{processed} accounts#{dry_run}", :green, true)
elsif username.present? elsif username.present?
username, domain = username.split('@') username, domain = username.split('@')
account = Account.find_remote(username, domain) account = Account.find_remote(username, domain)
@ -304,76 +294,53 @@ module Mastodon
exit(1) exit(1)
end end
Maintenance::RedownloadAccountMediaWorker.perform_async(account.id) unless options[:dry_run]
say('OK', :green) account.reset_avatar!
account.reset_header!
account.save
end
say("OK#{dry_run}", :green)
else else
say('No account(s) given', :red) say('No account(s) given', :red)
exit(1) exit(1)
end end
end end
desc 'follow ACCT', 'Make all local accounts follow account specified by ACCT' option :concurrency, type: :numeric, default: 5, aliases: [:c]
long_desc <<-LONG_DESC option :verbose, type: :boolean, aliases: [:v]
Make all local accounts follow another local account specified by ACCT. desc 'follow USERNAME', 'Make all local accounts follow account specified by USERNAME'
ACCT should be the username only. def follow(username)
LONG_DESC target_account = Account.find_local(username)
def follow(acct)
if acct.include? '@'
say('Target account name should not contain a target instance, since it has to be a local account.', :red)
exit(1)
end
target_account = ResolveAccountService.new.call(acct)
processed = 0
failed = 0
if target_account.nil? if target_account.nil?
say("Target account (#{acct}) could not be resolved", :red) say('No such account', :red)
exit(1) exit(1)
end end
Account.local.without_suspended.find_each do |account| processed, = parallelize_with_progress(Account.local.without_suspended) do |account|
begin FollowService.new.call(account, target_account)
FollowService.new.call(account, target_account)
processed += 1
say('.', :green, false)
rescue StandardError
failed += 1
say('.', :red, false)
end
end end
say("OK, followed target from #{processed} accounts, skipped #{failed}", :green) say("OK, followed target from #{processed} accounts", :green)
end end
option :concurrency, type: :numeric, default: 5, aliases: [:c]
option :verbose, type: :boolean, aliases: [:v]
desc 'unfollow ACCT', 'Make all local accounts unfollow account specified by ACCT' desc 'unfollow ACCT', 'Make all local accounts unfollow account specified by ACCT'
long_desc <<-LONG_DESC
Make all local accounts unfollow an account specified by ACCT. ACCT can be
a simple username, in case of a local user. It can also be in the format
username@domain, in case of a remote user.
LONG_DESC
def unfollow(acct) def unfollow(acct)
target_account = Account.find_remote(*acct.split('@')) target_account = Account.find_remote(*acct.split('@'))
processed = 0
failed = 0
if target_account.nil? if target_account.nil?
say("Target account (#{acct}) was not found", :red) say('No such account', :red)
exit(1) exit(1)
end end
target_account.followers.local.find_each do |account| parallelize_with_progress(target_account.followers.local) do |account|
begin UnfollowService.new.call(account, target_account)
UnfollowService.new.call(account, target_account)
processed += 1
say('.', :green, false)
rescue StandardError
failed += 1
say('.', :red, false)
end
end end
say("OK, unfollowed target from #{processed} accounts, skipped #{failed}", :green) say("OK, unfollowed target from #{processed} accounts", :green)
end end
option :follows, type: :boolean, default: false option :follows, type: :boolean, default: false
@ -396,51 +363,50 @@ module Mastodon
account = Account.find_local(username) account = Account.find_local(username)
if account.nil? if account.nil?
say('No user with such username', :red) say('No such account', :red)
exit(1) exit(1)
end end
total = 0
total += Account.where(id: ::Follow.where(account: account).select(:target_account_id)).count if options[:follows]
total += Account.where(id: ::Follow.where(target_account: account).select(:account_id)).count if options[:followers]
progress = create_progress_bar(total)
processed = 0
if options[:follows] if options[:follows]
processed = 0 scope = Account.where(id: ::Follow.where(account: account).select(:target_account_id))
failed = 0
say("Unfollowing #{account.username}'s followees, this might take a while...") scope.find_each do |target_account|
Account.where(id: ::Follow.where(account: account).select(:target_account_id)).find_each do |target_account|
begin begin
UnfollowService.new.call(account, target_account) UnfollowService.new.call(account, target_account)
rescue => e
progress.log pastel.red("Error processing #{target_account.id}: #{e}")
ensure
progress.increment
processed += 1 processed += 1
say('.', :green, false)
rescue StandardError
failed += 1
say('.', :red, false)
end end
end end
BootstrapTimelineWorker.perform_async(account.id) BootstrapTimelineWorker.perform_async(account.id)
say("OK, unfollowed #{processed} followees, skipped #{failed}", :green)
end end
if options[:followers] if options[:followers]
processed = 0 scope = Account.where(id: ::Follow.where(target_account: account).select(:account_id))
failed = 0
say("Removing #{account.username}'s followers, this might take a while...") scope.find_each do |target_account|
Account.where(id: ::Follow.where(target_account: account).select(:account_id)).find_each do |target_account|
begin begin
UnfollowService.new.call(target_account, account) UnfollowService.new.call(target_account, account)
rescue => e
progress.log pastel.red("Error processing #{target_account.id}: #{e}")
ensure
progress.increment
processed += 1 processed += 1
say('.', :green, false)
rescue StandardError
failed += 1
say('.', :red, false)
end end
end end
say("OK, removed #{processed} followers, skipped #{failed}", :green)
end end
progress.finish
say("Processed #{processed} relationships", :green, true)
end end
option :number, type: :numeric, aliases: [:n] option :number, type: :numeric, aliases: [:n]

View file

@ -6,6 +6,8 @@ require_relative 'cli_helper'
module Mastodon module Mastodon
class CacheCLI < Thor class CacheCLI < Thor
include CLIHelper
def self.exit_on_failure? def self.exit_on_failure?
true true
end end
@ -16,6 +18,8 @@ module Mastodon
say('OK', :green) say('OK', :green)
end end
option :concurrency, type: :numeric, default: 5, aliases: [:c]
option :verbose, type: :boolean, aliases: [:v]
desc 'recount TYPE', 'Update hard-cached counters' desc 'recount TYPE', 'Update hard-cached counters'
long_desc <<~LONG_DESC long_desc <<~LONG_DESC
Update hard-cached counters of TYPE by counting referenced Update hard-cached counters of TYPE by counting referenced
@ -25,32 +29,24 @@ module Mastodon
size of the database. size of the database.
LONG_DESC LONG_DESC
def recount(type) def recount(type)
processed = 0
case type case type
when 'accounts' when 'accounts'
Account.local.includes(:account_stat).find_each do |account| processed, = parallelize_with_progress(Account.local.includes(:account_stat)) do |account|
account_stat = account.account_stat account_stat = account.account_stat
account_stat.following_count = account.active_relationships.count account_stat.following_count = account.active_relationships.count
account_stat.followers_count = account.passive_relationships.count account_stat.followers_count = account.passive_relationships.count
account_stat.statuses_count = account.statuses.where.not(visibility: :direct).count account_stat.statuses_count = account.statuses.where.not(visibility: :direct).count
account_stat.save if account_stat.changed? account_stat.save if account_stat.changed?
processed += 1
say('.', :green, false)
end end
when 'statuses' when 'statuses'
Status.includes(:status_stat).find_each do |status| processed, = parallelize_with_progress(Status.includes(:status_stat)) do |status|
status_stat = status.status_stat status_stat = status.status_stat
status_stat.replies_count = status.replies.where.not(visibility: :direct).count status_stat.replies_count = status.replies.where.not(visibility: :direct).count
status_stat.reblogs_count = status.reblogs.count status_stat.reblogs_count = status.reblogs.count
status_stat.favourites_count = status.favourites.count status_stat.favourites_count = status.favourites.count
status_stat.save if status_stat.changed? status_stat.save if status_stat.changed?
processed += 1
say('.', :green, false)
end end
else else
say("Unknown type: #{type}", :red) say("Unknown type: #{type}", :red)

View file

@ -7,3 +7,52 @@ ActiveRecord::Base.logger = dev_null
ActiveJob::Base.logger = dev_null ActiveJob::Base.logger = dev_null
HttpLog.configuration.logger = dev_null HttpLog.configuration.logger = dev_null
Paperclip.options[:log] = false Paperclip.options[:log] = false
module Mastodon
module CLIHelper
def create_progress_bar(total = nil)
ProgressBar.create(total: total, format: '%c/%u |%b%i| %e')
end
def parallelize_with_progress(scope)
ActiveRecord::Base.configurations[Rails.env]['pool'] = options[:concurrency]
progress = create_progress_bar(scope.count)
pool = Concurrent::FixedThreadPool.new(options[:concurrency])
total = Concurrent::AtomicFixnum.new(0)
aggregate = Concurrent::AtomicFixnum.new(0)
scope.reorder(nil).find_in_batches do |items|
futures = []
items.each do |item|
futures << Concurrent::Future.execute(executor: pool) do
ActiveRecord::Base.connection_pool.with_connection do
begin
progress.log("Processing #{item.id}") if options[:verbose]
result = yield(item)
aggregate.increment(result) if result.is_a?(Integer)
rescue => e
progress.log pastel.red("Error processing #{item.id}: #{e}")
ensure
progress.increment
end
end
end
end
total.increment(items.size)
futures.map(&:value)
end
progress.finish
[total.value, aggregate.value]
end
def pastel
@pastel ||= Pastel.new
end
end
end

View file

@ -7,10 +7,14 @@ require_relative 'cli_helper'
module Mastodon module Mastodon
class DomainsCLI < Thor class DomainsCLI < Thor
include CLIHelper
def self.exit_on_failure? def self.exit_on_failure?
true true
end end
option :concurrency, type: :numeric, default: 5, aliases: [:c]
option :verbose, type: :boolean, aliases: [:v]
option :dry_run, type: :boolean option :dry_run, type: :boolean
option :whitelist_mode, type: :boolean option :whitelist_mode, type: :boolean
desc 'purge [DOMAIN]', 'Remove accounts from a DOMAIN without a trace' desc 'purge [DOMAIN]', 'Remove accounts from a DOMAIN without a trace'
@ -24,7 +28,6 @@ module Mastodon
are removed from the database. are removed from the database.
LONG_DESC LONG_DESC
def purge(domain = nil) def purge(domain = nil)
removed = 0
dry_run = options[:dry_run] ? ' (DRY RUN)' : '' dry_run = options[:dry_run] ? ' (DRY RUN)' : ''
scope = begin scope = begin
@ -38,25 +41,22 @@ module Mastodon
end end
end end
scope.find_each do |account| processed, = parallelize_with_progress(scope) do |account|
SuspendAccountService.new.call(account, destroy: true) unless options[:dry_run] SuspendAccountService.new.call(account, destroy: true) unless options[:dry_run]
removed += 1
say('.', :green, false)
end end
DomainBlock.where(domain: domain).destroy_all unless options[:dry_run] DomainBlock.where(domain: domain).destroy_all unless options[:dry_run]
say say("Removed #{processed} accounts#{dry_run}", :green)
say("Removed #{removed} accounts#{dry_run}", :green)
custom_emojis = CustomEmoji.where(domain: domain) custom_emojis = CustomEmoji.where(domain: domain)
custom_emojis_count = custom_emojis.count custom_emojis_count = custom_emojis.count
custom_emojis.destroy_all unless options[:dry_run] custom_emojis.destroy_all unless options[:dry_run]
say("Removed #{custom_emojis_count} custom emojis", :green) say("Removed #{custom_emojis_count} custom emojis", :green)
end end
option :concurrency, type: :numeric, default: 50, aliases: [:c] option :concurrency, type: :numeric, default: 50, aliases: [:c]
option :silent, type: :boolean, default: false, aliases: [:s]
option :format, type: :string, default: 'summary', aliases: [:f] option :format, type: :string, default: 'summary', aliases: [:f]
option :exclude_suspended, type: :boolean, default: false, aliases: [:x] option :exclude_suspended, type: :boolean, default: false, aliases: [:x]
desc 'crawl [START]', 'Crawl all known peers, optionally beginning at START' desc 'crawl [START]', 'Crawl all known peers, optionally beginning at START'
@ -69,8 +69,6 @@ module Mastodon
The --concurrency (-c) option controls the number of threads performing HTTP The --concurrency (-c) option controls the number of threads performing HTTP
requests at the same time. More threads means the crawl may complete faster. requests at the same time. More threads means the crawl may complete faster.
The --silent (-s) option controls progress output.
The --format (-f) option controls how the data is displayed at the end. By The --format (-f) option controls how the data is displayed at the end. By
default (`summary`), a summary of the statistics is returned. The other options default (`summary`), a summary of the statistics is returned. The other options
are `domains`, which returns a newline-delimited list of all discovered peers, are `domains`, which returns a newline-delimited list of all discovered peers,
@ -87,6 +85,7 @@ module Mastodon
start_at = Time.now.to_f start_at = Time.now.to_f
seed = start ? [start] : Account.remote.domains seed = start ? [start] : Account.remote.domains
blocked_domains = Regexp.new('\\.?' + DomainBlock.where(severity: 1).pluck(:domain).join('|') + '$') blocked_domains = Regexp.new('\\.?' + DomainBlock.where(severity: 1).pluck(:domain).join('|') + '$')
progress = create_progress_bar
pool = Concurrent::ThreadPoolExecutor.new(min_threads: 0, max_threads: options[:concurrency], idletime: 10, auto_terminate: true, max_queue: 0) pool = Concurrent::ThreadPoolExecutor.new(min_threads: 0, max_threads: options[:concurrency], idletime: 10, auto_terminate: true, max_queue: 0)
@ -95,7 +94,6 @@ module Mastodon
next if options[:exclude_suspended] && domain.match(blocked_domains) next if options[:exclude_suspended] && domain.match(blocked_domains)
stats[domain] = nil stats[domain] = nil
processed.increment
begin begin
Request.new(:get, "https://#{domain}/api/v1/instance").perform do |res| Request.new(:get, "https://#{domain}/api/v1/instance").perform do |res|
@ -115,11 +113,11 @@ module Mastodon
next unless res.code == 200 next unless res.code == 200
stats[domain]['activity'] = Oj.load(res.to_s) stats[domain]['activity'] = Oj.load(res.to_s)
end end
say('.', :green, false) unless options[:silent]
rescue StandardError rescue StandardError
failed.increment failed.increment
say('.', :red, false) unless options[:silent] ensure
processed.increment
progress.increment unless progress.finished?
end end
end end
@ -133,10 +131,9 @@ module Mastodon
pool.shutdown pool.shutdown
pool.wait_for_termination(20) pool.wait_for_termination(20)
ensure ensure
progress.finish
pool.shutdown pool.shutdown
say unless options[:silent]
case options[:format] case options[:format]
when 'summary' when 'summary'
stats_to_summary(stats, processed, failed, start_at) stats_to_summary(stats, processed, failed, start_at)

View file

@ -6,55 +6,33 @@ require_relative 'cli_helper'
module Mastodon module Mastodon
class FeedsCLI < Thor class FeedsCLI < Thor
include CLIHelper
def self.exit_on_failure? def self.exit_on_failure?
true true
end end
option :all, type: :boolean, default: false option :all, type: :boolean, default: false
option :background, type: :boolean, default: false option :concurrency, type: :numeric, default: 5, aliases: [:c]
option :verbose, type: :boolean, aliases: [:v]
option :dry_run, type: :boolean, default: false option :dry_run, type: :boolean, default: false
option :verbose, type: :boolean, default: false
desc 'build [USERNAME]', 'Build home and list feeds for one or all users' desc 'build [USERNAME]', 'Build home and list feeds for one or all users'
long_desc <<-LONG_DESC long_desc <<-LONG_DESC
Build home and list feeds that are stored in Redis from the database. Build home and list feeds that are stored in Redis from the database.
With the --all option, all active users will be processed. With the --all option, all active users will be processed.
Otherwise, a single user specified by USERNAME. Otherwise, a single user specified by USERNAME.
With the --background option, regeneration will be queued into Sidekiq,
and the command will exit as soon as possible.
With the --dry-run option, no work will be done.
With the --verbose option, when accounts are processed sequentially in the
foreground, the IDs of the accounts will be printed.
LONG_DESC LONG_DESC
def build(username = nil) def build(username = nil)
dry_run = options[:dry_run] ? '(DRY RUN)' : '' dry_run = options[:dry_run] ? '(DRY RUN)' : ''
if options[:all] || username.nil? if options[:all] || username.nil?
processed = 0
queued = 0
User.active.select(:id, :account_id).reorder(nil).find_in_batches do |users| processed, = parallelize_with_progress(Account.joins(:user).merge(User.active)) do |account|
if options[:background] PrecomputeFeedService.new.call(account) unless options[:dry_run]
RegenerationWorker.push_bulk(users.map(&:account_id)) unless options[:dry_run]
queued += users.size
else
users.each do |user|
RegenerationWorker.new.perform(user.account_id) unless options[:dry_run]
options[:verbose] ? say(user.account_id) : say('.', :green, false)
processed += 1
end
end
end end
if options[:background] say("Regenerated feeds for #{processed} accounts #{dry_run}", :green, true)
say("Scheduled feed regeneration for #{queued} accounts #{dry_run}", :green, true)
else
say
say("Regenerated feeds for #{processed} accounts #{dry_run}", :green, true)
end
elsif username.present? elsif username.present?
account = Account.find_local(username) account = Account.find_local(username)
@ -63,11 +41,7 @@ module Mastodon
exit(1) exit(1)
end end
if options[:background] PrecomputeFeedService.new.call(account) unless options[:dry_run]
RegenerationWorker.perform_async(account.id) unless options[:dry_run]
else
RegenerationWorker.new.perform(account.id) unless options[:dry_run]
end
say("OK #{dry_run}", :green, true) say("OK #{dry_run}", :green, true)
else else

View file

@ -7,14 +7,15 @@ require_relative 'cli_helper'
module Mastodon module Mastodon
class MediaCLI < Thor class MediaCLI < Thor
include ActionView::Helpers::NumberHelper include ActionView::Helpers::NumberHelper
include CLIHelper
def self.exit_on_failure? def self.exit_on_failure?
true true
end end
option :days, type: :numeric, default: 7 option :days, type: :numeric, default: 7, aliases: [:d]
option :background, type: :boolean, default: false option :concurrency, type: :numeric, default: 5, aliases: [:c]
option :verbose, type: :boolean, default: false option :verbose, type: :boolean, default: false, aliases: [:v]
option :dry_run, type: :boolean, default: false option :dry_run, type: :boolean, default: false
desc 'remove', 'Remove remote media files' desc 'remove', 'Remove remote media files'
long_desc <<-DESC long_desc <<-DESC
@ -22,49 +23,25 @@ module Mastodon
The --days option specifies how old media attachments have to be before The --days option specifies how old media attachments have to be before
they are removed. It defaults to 7 days. they are removed. It defaults to 7 days.
With the --background option, instead of deleting the files sequentially,
they will be queued into Sidekiq and the command will exit as soon as
possible. In Sidekiq they will be processed with higher concurrency, but
it may impact other operations of the Mastodon server, and it may overload
the underlying file storage.
With the --dry-run option, no work will be done.
With the --verbose option, when media attachments are processed sequentially in the
foreground, the IDs of the media attachments will be printed.
DESC DESC
def remove def remove
time_ago = options[:days].days.ago time_ago = options[:days].days.ago
queued = 0 dry_run = options[:dry_run] ? '(DRY RUN)' : ''
processed = 0
size = 0
dry_run = options[:dry_run] ? '(DRY RUN)' : ''
if options[:background] processed, aggregate = parallelize_with_progress(MediaAttachment.cached.where.not(remote_url: '').where('created_at < ?', time_ago)) do |media_attachment|
MediaAttachment.where.not(remote_url: '').where.not(file_file_name: nil).where('created_at < ?', time_ago).select(:id, :file_file_size).reorder(nil).find_in_batches do |media_attachments| next if media_attachment.file.blank?
queued += media_attachments.size
size += media_attachments.reduce(0) { |sum, m| sum + (m.file_file_size || 0) } size = media_attachment.file_file_size
Maintenance::UncacheMediaWorker.push_bulk(media_attachments.map(&:id)) unless options[:dry_run]
end unless options[:dry_run]
else media_attachment.file.destroy
MediaAttachment.where.not(remote_url: '').where.not(file_file_name: nil).where('created_at < ?', time_ago).reorder(nil).find_in_batches do |media_attachments| media_attachment.save
media_attachments.each do |m|
size += m.file_file_size || 0
Maintenance::UncacheMediaWorker.new.perform(m) unless options[:dry_run]
options[:verbose] ? say(m.id) : say('.', :green, false)
processed += 1
end
end end
size
end end
say say("Removed #{processed} media attachments (approx. #{number_to_human_size(aggregate)}) #{dry_run}", :green, true)
if options[:background]
say("Scheduled the deletion of #{queued} media attachments (approx. #{number_to_human_size(size)}) #{dry_run}", :green, true)
else
say("Removed #{processed} media attachments (approx. #{number_to_human_size(size)}) #{dry_run}", :green, true)
end
end end
end end
end end

View file

@ -8,87 +8,52 @@ require_relative 'cli_helper'
module Mastodon module Mastodon
class PreviewCardsCLI < Thor class PreviewCardsCLI < Thor
include ActionView::Helpers::NumberHelper include ActionView::Helpers::NumberHelper
include CLIHelper
def self.exit_on_failure? def self.exit_on_failure?
true true
end end
option :days, type: :numeric, default: 180 option :days, type: :numeric, default: 180
option :background, type: :boolean, default: false option :concurrency, type: :numeric, default: 5, aliases: [:c]
option :verbose, type: :boolean, default: false option :verbose, type: :boolean, aliases: [:v]
option :dry_run, type: :boolean, default: false option :dry_run, type: :boolean, default: false
option :link, type: :boolean, default: false option :link, type: :boolean, default: false
desc 'remove', 'Remove preview cards' desc 'remove', 'Remove preview cards'
long_desc <<-DESC long_desc <<-DESC
Removes locally thumbnails for previews. Removes local thumbnails for preview cards.
The --days option specifies how old preview cards have to be before The --days option specifies how old preview cards have to be before
they are removed. It defaults to 180 days. they are removed. It defaults to 180 days. Since preview cards will
not be re-fetched unless the link is re-posted after 2 weeks from
last time, it is not recommended to delete preview cards within the
last 14 days.
With the --background option, instead of deleting the files sequentially, With the --link option, only link-type preview cards will be deleted,
they will be queued into Sidekiq and the command will exit as soon as leaving video and photo cards untouched.
possible. In Sidekiq they will be processed with higher concurrency, but
it may impact other operations of the Mastodon server, and it may overload
the underlying file storage.
With the --dry-run option, no work will be done.
With the --verbose option, when preview cards are processed sequentially in the
foreground, the IDs of the preview cards will be printed.
With the --link option, delete only link-type preview cards.
DESC DESC
def remove def remove
prompt = TTY::Prompt.new time_ago = options[:days].days.ago
time_ago = options[:days].days.ago dry_run = options[:dry_run] ? ' (DRY RUN)' : ''
queued = 0 link = options[:link] ? 'link-type ' : ''
processed = 0 scope = PreviewCard.cached
size = 0 scope = scope.where(type: :link) if options[:link]
dry_run = options[:dry_run] ? '(DRY RUN)' : '' scope = scope.where('updated_at < ?', time_ago)
link = options[:link] ? 'link-type ' : ''
scope = PreviewCard.where.not(image_file_name: nil)
scope = scope.where.not(image_file_name: '')
scope = scope.where(type: :link) if options[:link]
scope = scope.where('updated_at < ?', time_ago)
if time_ago > 2.weeks.ago processed, aggregate = parallelize_with_progress(scope) do |preview_card|
prompt.say "\n" next if preview_card.image.blank?
prompt.say('The preview cards less than the past two weeks will not be re-acquired even when needed.')
prompt.say "\n"
unless prompt.yes?('Are you sure you want to delete the preview cards?', default: false) size = preview_card.image_file_size
prompt.say "\n"
prompt.warn 'Nothing execute. Bye!'
prompt.say "\n"
exit(1)
end
end
if options[:background] unless options[:dry_run]
scope.select(:id, :image_file_size).reorder(nil).find_in_batches do |preview_cards| preview_card.image.destroy
queued += preview_cards.size preview_card.save
size += preview_cards.reduce(0) { |sum, p| sum + (p.image_file_size || 0) }
Maintenance::UncachePreviewWorker.push_bulk(preview_cards.map(&:id)) unless options[:dry_run]
end end
else size
scope.select(:id, :image_file_size).reorder(nil).find_in_batches do |preview_cards|
preview_cards.each do |p|
size += p.image_file_size || 0
Maintenance::UncachePreviewWorker.new.perform(p.id) unless options[:dry_run]
options[:verbose] ? say(p.id) : say('.', :green, false)
processed += 1
end
end
end end
say say("Removed #{processed} #{link}preview cards (approx. #{number_to_human_size(aggregate)})#{dry_run}", :green, true)
if options[:background]
say("Scheduled the deletion of #{queued} #{link}preview cards (approx. #{number_to_human_size(size)}) #{dry_run}", :green, true)
else
say("Removed #{processed} #{link}preview cards (approx. #{number_to_human_size(size)}) #{dry_run}", :green, true)
end
end end
end end
end end