Merge pull request #7629 from betagouv/fix-exports

Fix: purge stuck exports & archives

app/dashboards/archive_dashboard.rb

@@ -11,7 +11,7 @@ class ArchiveDashboard < Administrate::BaseDashboard
     id: Field::Number,
     created_at: Field::DateTime,
     updated_at: Field::DateTime,
-    status: Field::String,
+    job_status: Field::String,
     file: AttachmentField
   }.freeze
 
@@ -24,7 +24,7 @@ class ArchiveDashboard < Administrate::BaseDashboard
     :id,
     :created_at,
     :updated_at,
-    :status,
+    :job_status,
     :file
   ].freeze
 
@@ -34,6 +34,6 @@ class ArchiveDashboard < Administrate::BaseDashboard
     :id,
     :created_at,
     :updated_at,
-    :status
+    :job_status
   ].freeze
 end

app/jobs/cron/purge_stale_archives_job.rb

@@ -3,5 +3,6 @@ class Cron::PurgeStaleArchivesJob < Cron::CronJob
 
   def perform
     Archive.stale(Archive::RETENTION_DURATION).destroy_all
+    Archive.stuck(Archive::MAX_DUREE_GENERATION).destroy_all
   end
 end

app/jobs/cron/purge_stale_exports_job.rb

@@ -3,5 +3,6 @@ class Cron::PurgeStaleExportsJob < Cron::CronJob
 
   def perform
     Export.stale(Export::MAX_DUREE_CONSERVATION_EXPORT).destroy_all
+    Export.stuck(Export::MAX_DUREE_GENERATION).destroy_all
   end
 end

app/models/archive.rb

@@ -14,6 +14,7 @@ class Archive < ApplicationRecord
   include TransientModelsWithPurgeableJobConcern
 
   RETENTION_DURATION = 4.days
+  MAX_DUREE_GENERATION = 24.hours
   MAX_SIZE = 100.gigabytes
 
   has_and_belongs_to_many :groupe_instructeurs

app/models/concerns/transient_models_with_purgeable_job_concern.rb

@@ -35,6 +35,11 @@ module TransientModelsWithPurgeableJobConcern
         .where('updated_at < ?', (Time.zone.now - duration))
     }
 
+    scope :stuck, lambda { |duration|
+      where(job_status: [job_statuses.fetch(:pending)])
+        .where('updated_at < ?', (Time.zone.now - duration))
+    }
+
     def available?
       generated?
     end

app/models/export.rb

@@ -17,6 +17,7 @@ class Export < ApplicationRecord
   include TransientModelsWithPurgeableJobConcern
 
   MAX_DUREE_CONSERVATION_EXPORT = 16.hours
+  MAX_DUREE_GENERATION = 12.hours
 
   enum format: {
     csv: 'csv',

spec/models/archive_spec.rb

@@ -4,26 +4,34 @@ describe Archive do
   before { Timecop.freeze(Time.zone.now) }
   after { Timecop.return }
 
-  let(:archive) { create(:archive, job_status: :pending) }
+  let!(:archive) { create(:archive, job_status: :pending) }
 
   describe 'scopes' do
     describe 'staled' do
-      let(:recent_archive) { create(:archive, job_status: :pending) }
+      let!(:recent_archive) { create(:archive, job_status: :pending) }
-      let(:staled_archive_still_pending) { create(:archive, job_status: :pending, updated_at: (Archive::RETENTION_DURATION + 2).days.ago) }
+      let!(:staled_archive_still_pending) { create(:archive, job_status: :pending, updated_at: (Archive::RETENTION_DURATION + 2).days.ago) }
-      let(:staled_archive_still_failed) { create(:archive, job_status: :failed, updated_at: (Archive::RETENTION_DURATION + 2).days.ago) }
+      let!(:staled_archive_still_failed) { create(:archive, job_status: :failed, updated_at: (Archive::RETENTION_DURATION + 2).days.ago) }
-      let(:staled_archive_still_generated) { create(:archive, job_status: :generated, updated_at: (Archive::RETENTION_DURATION + 2).days.ago) }
+      let!(:staled_archive_still_generated) { create(:archive, job_status: :generated, updated_at: (Archive::RETENTION_DURATION + 2).days.ago) }
 
       subject do
-        archive
-        recent_archive
-        staled_archive_still_pending
-        staled_archive_still_failed
-        staled_archive_still_generated
         Archive.stale(Archive::RETENTION_DURATION)
       end
 
       it { is_expected.to match_array([staled_archive_still_failed, staled_archive_still_generated]) }
     end
+
+    describe 'stuck' do
+      let!(:recent_archive) { create(:archive, job_status: :pending) }
+      let!(:staled_archive_still_pending) { create(:archive, job_status: :pending, updated_at: (Archive::MAX_DUREE_GENERATION + 2).days.ago) }
+      let!(:staled_archive_still_failed) { create(:archive, job_status: :failed, updated_at: (Archive::MAX_DUREE_GENERATION + 2).days.ago) }
+      let!(:staled_archive_still_generated) { create(:archive, job_status: :generated, updated_at: (Archive::MAX_DUREE_GENERATION + 2).days.ago) }
+
+      subject do
+        Archive.stuck(Archive::MAX_DUREE_GENERATION)
+      end
+
+      it { is_expected.to match_array([staled_archive_still_pending]) }
+    end
   end
 
   describe '.job_status' do

spec/models/export_spec.rb

@@ -31,6 +31,16 @@ RSpec.describe Export, type: :model do
     it { expect(Export.stale(Export::MAX_DUREE_CONSERVATION_EXPORT)).to match_array([stale_export_generated, stale_export_failed]) }
   end
 
+  describe '.stuck' do
+    let!(:export) { create(:export) }
+    let(:stuck_date) { Time.zone.now() - (Export::MAX_DUREE_GENERATION + 1.minute) }
+    let!(:stale_export_generated) { create(:export, :generated, updated_at: stuck_date) }
+    let!(:stale_export_failed) { create(:export, :failed, updated_at: stuck_date) }
+    let!(:stale_export_pending) { create(:export, :pending, updated_at: stuck_date) }
+
+    it { expect(Export.stuck(Export::MAX_DUREE_GENERATION)).to match_array([stale_export_pending]) }
+  end
+
   describe '.destroy' do
     let!(:groupe_instructeur) { create(:groupe_instructeur) }
     let!(:export) { create(:export, groupe_instructeurs: [groupe_instructeur]) }