From 5333447be0c0e278d6f591bb6004fdee903a08f7 Mon Sep 17 00:00:00 2001
From: Claire <claire.github-309c@sitedethib.com>
Date: Wed, 9 Nov 2022 14:08:19 +0100
Subject: [PATCH] Change account deletion requests to spread out over time
 (#20222)

---
 app/workers/admin/account_deletion_worker.rb  |  2 +-
 .../suspended_user_cleanup_scheduler.rb       | 38 +++++++++++++++++++
 .../scheduler/user_cleanup_scheduler.rb       |  7 ----
 config/sidekiq.yml                            |  4 ++
 4 files changed, 43 insertions(+), 8 deletions(-)
 create mode 100644 app/workers/scheduler/suspended_user_cleanup_scheduler.rb

diff --git a/app/workers/admin/account_deletion_worker.rb b/app/workers/admin/account_deletion_worker.rb
index 82f269ad6f..6e0eb331be 100644
--- a/app/workers/admin/account_deletion_worker.rb
+++ b/app/workers/admin/account_deletion_worker.rb
@@ -3,7 +3,7 @@
 class Admin::AccountDeletionWorker
   include Sidekiq::Worker
 
-  sidekiq_options queue: 'pull'
+  sidekiq_options queue: 'pull', lock: :until_executed
 
   def perform(account_id)
     DeleteAccountService.new.call(Account.find(account_id), reserve_username: true, reserve_email: true)
diff --git a/app/workers/scheduler/suspended_user_cleanup_scheduler.rb b/app/workers/scheduler/suspended_user_cleanup_scheduler.rb
new file mode 100644
index 0000000000..50768f83cc
--- /dev/null
+++ b/app/workers/scheduler/suspended_user_cleanup_scheduler.rb
@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+class Scheduler::SuspendedUserCleanupScheduler
+  include Sidekiq::Worker
+
+  # Each processed deletion request may enqueue an enormous
+  # amount of jobs in the `pull` queue, so only enqueue when
+  # the queue is empty or close to being so.
+  MAX_PULL_SIZE = 50
+
+  # Since account deletion is very expensive, we want to avoid
+  # overloading the server by queing too much at once.
+  # This job runs approximately once per 2 minutes, so with a
+  # value of `MAX_DELETIONS_PER_JOB` of 10, a server can
+  # handle the deletion of 7200 accounts per day, provided it
+  # has the capacity for it.
+  MAX_DELETIONS_PER_JOB = 10
+
+  sidekiq_options retry: 0
+
+  def perform
+    return if Sidekiq::Queue.new('pull').size > MAX_PULL_SIZE
+
+    clean_suspended_accounts!
+  end
+
+  private
+
+  def clean_suspended_accounts!
+    # This should be fine because we only process a small amount of deletion requests at once and
+    # `id` and `created_at` should follow the same order.
+    AccountDeletionRequest.reorder(id: :asc).take(MAX_DELETIONS_PER_JOB).each do |deletion_request|
+      next unless deletion_request.created_at < AccountDeletionRequest::DELAY_TO_DELETION.ago
+
+      Admin::AccountDeletionWorker.perform_async(deletion_request.account_id)
+    end
+  end
+end
diff --git a/app/workers/scheduler/user_cleanup_scheduler.rb b/app/workers/scheduler/user_cleanup_scheduler.rb
index 7a6995a1ff..63f9ed78cc 100644
--- a/app/workers/scheduler/user_cleanup_scheduler.rb
+++ b/app/workers/scheduler/user_cleanup_scheduler.rb
@@ -7,7 +7,6 @@ class Scheduler::UserCleanupScheduler
 
   def perform
     clean_unconfirmed_accounts!
-    clean_suspended_accounts!
     clean_discarded_statuses!
   end
 
@@ -22,12 +21,6 @@ class Scheduler::UserCleanupScheduler
     end
   end
 
-  def clean_suspended_accounts!
-    AccountDeletionRequest.where('created_at <= ?', AccountDeletionRequest::DELAY_TO_DELETION.ago).reorder(nil).find_each do |deletion_request|
-      Admin::AccountDeletionWorker.perform_async(deletion_request.account_id)
-    end
-  end
-
   def clean_discarded_statuses!
     Status.unscoped.discarded.where('deleted_at <= ?', 30.days.ago).find_in_batches do |statuses|
       RemovalWorker.push_bulk(statuses) do |status|
diff --git a/config/sidekiq.yml b/config/sidekiq.yml
index e3156aa346..71e7cb33d8 100644
--- a/config/sidekiq.yml
+++ b/config/sidekiq.yml
@@ -53,3 +53,7 @@
     interval: 1 minute
     class: Scheduler::AccountsStatusesCleanupScheduler
     queue: scheduler
+  suspended_user_cleanup_scheduler:
+    interval: 1 minute
+    class: Scheduler::SuspendedUserCleanupScheduler
+    queue: scheduler