Merge branch 'increase-event-retention-prune-old-events-worker' into 'master'

Prune events older than 2 years instead of 12 months

See merge request gitlab-org/gitlab-ce!22145

4 files changed, 22 insertions, 12 deletions

diff --git a/app/workers/prune_old_events_worker.rb b/app/workers/prune_old_events_worker.rb
index d44ad0d8030..dc4b7670131 100644
--- a/app/workers/prune_old_events_worker.rb
+++ b/app/workers/prune_old_events_worker.rb
@@ -6,14 +6,13 @@ class PruneOldEventsWorker
 
   # rubocop: disable CodeReuse/ActiveRecord
   def perform
-    # Contribution calendar shows maximum 12 months of events.
-    # Double nested query is used because MySQL doesn't allow DELETE subqueries
-    # on the same table.
+    # Contribution calendar shows maximum 12 months of events, we retain 2 years for data integrity.
+    # Double nested query is used because MySQL doesn't allow DELETE subqueries on the same table.
     Event.unscoped.where(
       '(id IN (SELECT id FROM (?) ids_to_remove))',
       Event.unscoped.where(
         'created_at < ?',
-        (12.months + 1.day).ago)
+        (2.years + 1.day).ago)
       .select(:id)
       .limit(10_000))
     .delete_all
diff --git a/changelogs/unreleased/copy-changes-for-abuse-clarity.yml b/changelogs/unreleased/copy-changes-for-abuse-clarity.yml
new file mode 100644
index 00000000000..00d9fec5e42
--- /dev/null
+++ b/changelogs/unreleased/copy-changes-for-abuse-clarity.yml
@@ -0,0 +1,5 @@
+---
+title: Increased retained event data by extending events pruner timeframe to 2 years
+merge_request: 22145
+author:
+type: changed
diff --git a/doc/api/events.md b/doc/api/events.md
index fb5ebb71a86..cd84b32029e 100644
--- a/doc/api/events.md
+++ b/doc/api/events.md
@@ -44,7 +44,7 @@ YYYY-MM-DD
 
 ### Event Time Period Limit
 
-GitLab removes events older than 1 year from the events table for performance reasons. The range of 1 year was chosen because user contribution calendars only show contributions of the past year.
+GitLab removes events older than 2 years from the events table for performance reasons.
 
 ## List currently authenticated user's events
 
diff --git a/spec/workers/prune_old_events_worker_spec.rb b/spec/workers/prune_old_events_worker_spec.rb
index ea974355050..b999a6fd5b6 100644
--- a/spec/workers/prune_old_events_worker_spec.rb
+++ b/spec/workers/prune_old_events_worker_spec.rb
@@ -4,23 +4,29 @@ describe PruneOldEventsWorker do
   describe '#perform' do
     let(:user) { create(:user) }
 
-    let!(:expired_event) { create(:event, :closed, author: user, created_at: 13.months.ago) }
-    let!(:not_expired_event) { create(:event, :closed, author: user,  created_at: 1.day.ago) }
-    let!(:exactly_12_months_event) { create(:event, :closed, author: user, created_at: 12.months.ago) }
+    let!(:expired_event) { create(:event, :closed, author: user, created_at: 25.months.ago) }
+    let!(:not_expired_1_day_event) { create(:event, :closed, author: user,  created_at: 1.day.ago) }
+    let!(:not_expired_13_month_event) { create(:event, :closed, author: user,  created_at: 13.months.ago) }
+    let!(:not_expired_2_years_event) { create(:event, :closed, author: user, created_at: 2.years.ago) }
 
-    it 'prunes events older than 12 months' do
+    it 'prunes events older than 2 years' do
       expect { subject.perform }.to change { Event.count }.by(-1)
       expect(Event.find_by(id: expired_event.id)).to be_nil
     end
 
     it 'leaves fresh events' do
       subject.perform
-      expect(not_expired_event.reload).to be_present
+      expect(not_expired_1_day_event.reload).to be_present
     end
 
-    it 'leaves events from exactly 12 months ago' do
+    it 'leaves events from 13 months ago' do
       subject.perform
-      expect(exactly_12_months_event).to be_present
+      expect(not_expired_13_month_event.reload).to be_present
+    end
+
+    it 'leaves events from 2 years ago' do
+      subject.perform
+      expect(not_expired_2_years_event).to be_present
     end
   end
 end