From f4b8d39c498c5c817aae3f91ae446b96a46dcf92 Mon Sep 17 00:00:00 2001
From: Toby Privett <toby.privett@digital.justice.gov.uk>
Date: Fri, 15 Nov 2024 12:19:02 +0000
Subject: [PATCH] MAP-1781 Cleanup Access Logs more than 3 months old (#2408)

* Cleanup Access Logs more than 3 months old

* Fix rubocop issues

* Prefer a cutoff date of 6 months

* Add a spec to test the access_logs:cleanup task

* Prefer cutoff date of 1 year

* `reenable` lets us run a rake task multiple times in a spec
---
 helm_deploy/values-preprod.yaml    |  3 +++
 helm_deploy/values-production.yaml |  3 +++
 helm_deploy/values-staging.yaml    |  3 +++
 helm_deploy/values-uat.yaml        |  3 +++
 lib/tasks/access_logs.rake         | 14 ++++++++++++++
 spec/factories/access_logs.rb      |  1 -
 spec/lib/tasks/access_logs_spec.rb | 31 ++++++++++++++++++++++++++++++
 7 files changed, 57 insertions(+), 1 deletion(-)
 create mode 100644 lib/tasks/access_logs.rake
 create mode 100644 spec/lib/tasks/access_logs_spec.rb

diff --git a/helm_deploy/values-preprod.yaml b/helm_deploy/values-preprod.yaml
index 93839a684..2dec95c7e 100644
--- a/helm_deploy/values-preprod.yaml
+++ b/helm_deploy/values-preprod.yaml
@@ -49,6 +49,9 @@ generic-service:
       REDIS_URL: url
 
 cronJobs:
+  - name: access-log-cleanup
+    schedule: "0 2 * * *"
+    command: ["bundle", "exec", "rake", "access_logs:cleanup"]
   - name: token-cleanup
     schedule: "0 1 * * 0"
     command: ["bundle", "exec", "rake", "doorkeeper:db:cleanup"]
diff --git a/helm_deploy/values-production.yaml b/helm_deploy/values-production.yaml
index beacd90e7..f0ce86a8b 100644
--- a/helm_deploy/values-production.yaml
+++ b/helm_deploy/values-production.yaml
@@ -77,6 +77,9 @@ sidekiq:
         GOVUK_NOTIFY_API_KEY: govuk_notify_api_key
 
 cronJobs:
+  - name: access-log-cleanup
+    schedule: "0 2 * * *"
+    command: ["bundle", "exec", "rake", "access_logs:cleanup"]
   - name: token-cleanup
     schedule: "0 1 * * 0"
     command: ["bundle", "exec", "rake", "doorkeeper:db:cleanup"]
diff --git a/helm_deploy/values-staging.yaml b/helm_deploy/values-staging.yaml
index c72b9e94b..f4944e855 100644
--- a/helm_deploy/values-staging.yaml
+++ b/helm_deploy/values-staging.yaml
@@ -63,6 +63,9 @@ sidekiq:
         GOVUK_NOTIFY_API_KEY: govuk_notify_api_key
 
 cronJobs:
+  - name: access-log-cleanup
+    schedule: "0 2 * * *"
+    command: ["bundle", "exec", "rake", "access_logs:cleanup"]
   - name: token-cleanup
     schedule: "0 1 * * 0"
     command: ["bundle", "exec", "rake", "doorkeeper:db:cleanup"]
diff --git a/helm_deploy/values-uat.yaml b/helm_deploy/values-uat.yaml
index 94d0b6429..0431e9255 100644
--- a/helm_deploy/values-uat.yaml
+++ b/helm_deploy/values-uat.yaml
@@ -55,6 +55,9 @@ sidekiq:
         GOVUK_NOTIFY_API_KEY: govuk_notify_api_key
 
 cronJobs:
+  - name: access-log-cleanup
+    schedule: "0 2 * * *"
+    command: ["bundle", "exec", "rake", "access_logs:cleanup"]
   - name: token-cleanup
     schedule: "0 1 * * 0"
     command: ["bundle", "exec", "rake", "doorkeeper:db:cleanup"]
diff --git a/lib/tasks/access_logs.rake b/lib/tasks/access_logs.rake
new file mode 100644
index 000000000..87b375142
--- /dev/null
+++ b/lib/tasks/access_logs.rake
@@ -0,0 +1,14 @@
+namespace :access_logs do
+  desc 'Cleanup access_logs older than 1 year'
+  task cleanup: [:environment] do
+    cutoff_date = 1.year.ago
+    access_logs_count = AccessLog.where('timestamp < ?', cutoff_date).count
+    number_of_iterations = (access_logs_count.to_f / 1000).ceil
+
+    puts "Cleaning up #{access_logs_count} access_logs in #{number_of_iterations} iterations"
+
+    1.upto(number_of_iterations).each do
+      AccessLog.where('timestamp < ?', cutoff_date).limit(1000).delete_all
+    end
+  end
+end
diff --git a/spec/factories/access_logs.rb b/spec/factories/access_logs.rb
index 26b36920a..055491584 100644
--- a/spec/factories/access_logs.rb
+++ b/spec/factories/access_logs.rb
@@ -1,6 +1,5 @@
 FactoryBot.define do
   factory :access_log do
-    id { '35cc6a19-0d88-453a-a0cb-970e161b5cbb' }
     request_id { 'b68c0883-540c-426a-a9a4-daf586eb5c78' }
     timestamp { Time.zone.now }
     whodunnit { 'AUSER01' }
diff --git a/spec/lib/tasks/access_logs_spec.rb b/spec/lib/tasks/access_logs_spec.rb
new file mode 100644
index 000000000..e27abf4e1
--- /dev/null
+++ b/spec/lib/tasks/access_logs_spec.rb
@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+require 'rails_helper'
+
+RSpec.describe Rake::Task['access_logs:cleanup'] do
+  before do
+    # 2 access_logs to retain
+    create(:access_log, timestamp: 1.month.ago)
+    create(:access_log, timestamp: 7.months.ago)
+
+    # 3 access_logs to delete
+    create(:access_log, timestamp: 13.months.ago)
+    create(:access_log, timestamp: 14.months.ago)
+    create(:access_log, timestamp: 15.months.ago)
+
+    allow($stdout).to receive(:puts)
+
+    described_class.reenable
+    described_class.invoke
+  end
+
+  it 'cleans up the access_logs' do
+    expect(AccessLog.count).to eq(2)
+  end
+
+  it 'writes to stdout' do
+    expect($stdout)
+      .to have_received(:puts)
+      .with('Cleaning up 3 access_logs in 1 iterations')
+  end
+end