Merge pull request #4935 from coopdevs/data-archiving

Allow data archiving using the :truncate_data task
This commit is contained in:
Luis Ramos
2020-05-03 12:58:31 +01:00
committed by GitHub
4 changed files with 305 additions and 68 deletions

View File

@@ -1,81 +1,45 @@
# frozen_string_literal: true
require 'highline'
require 'tasks/data/truncate_data'
# This task can be used to significantly reduce the size of a database
# This is used for example when loading live data into a staging server
# This way the staging server is not overloaded with too much data
#
# This is also aimed at implementing data archiving. We assume data older than
# 2 years can be safely removed and restored from a backup. This gives room for
# hubs to do their tax declaration.
#
# It's a must to perform a backup right before executing this. Then, to allow
# for a later data recovery we need to keep track of the exact moment this rake
# task was executed.
#
# Execute this in production only when the instance users are sleeping to avoid any trouble.
#
# Example:
#
# $ bundle exec rake "ofn:data:truncate[24]"
#
# This will remove data older than 2 years (24 months).
namespace :ofn do
namespace :data do
desc 'Truncate data'
task truncate: :environment do
guard_and_warn
task :truncate, [:months_to_keep] => :environment do |_task, args|
warn_with_confirmation
sql_delete_from "
spree_inventory_units #{where_order_id_in_orders_to_delete}"
sql_delete_from "
spree_inventory_units
where shipment_id in (select id from spree_shipments #{where_order_id_in_orders_to_delete})"
truncate_adjustments
sql_delete_from "spree_line_items #{where_order_id_in_orders_to_delete}"
sql_delete_from "spree_payments #{where_order_id_in_orders_to_delete}"
sql_delete_from "spree_shipments #{where_order_id_in_orders_to_delete}"
Spree::ReturnAuthorization.delete_all
truncate_order_cycle_data
sql_delete_from "proxy_orders #{where_oc_id_in_ocs_to_delete}"
sql_delete_from "spree_orders #{where_oc_id_in_ocs_to_delete}"
sql_delete_from "order_cycle_schedules #{where_oc_id_in_ocs_to_delete}"
sql_delete_from "order_cycles #{where_ocs_to_delete}"
Spree::TokenizedPermission.where("created_at < '#{date}'").delete_all
Spree::StateChange.delete_all
Spree::LogEntry.delete_all
sql_delete_from "sessions"
TruncateData.new(args.months_to_keep).call
end
def sql_delete_from(sql)
ActiveRecord::Base.connection.execute("delete from #{sql}")
end
def warn_with_confirmation
message = <<-MSG.strip_heredoc
\n
<% highlighted_message = "This will permanently change DB contents. This is not meant to be run in production as it needs more thorough testing." %>
<%= color(highlighted_message, :blink, :on_red) %>
Are you sure you want to proceed? (y/N)
MSG
private
def date
3.months.ago
end
def where_ocs_to_delete
"where orders_close_at < '#{date}'"
end
def where_oc_id_in_ocs_to_delete
"where order_cycle_id in (select id from order_cycles #{where_ocs_to_delete} )"
end
def where_order_id_in_orders_to_delete
"where order_id in (select id from spree_orders #{where_oc_id_in_ocs_to_delete})"
end
def truncate_adjustments
sql_delete_from "spree_adjustments where source_type = 'Spree::Order'
and source_id in (select id from spree_orders #{where_oc_id_in_ocs_to_delete})"
sql_delete_from "spree_adjustments where source_type = 'Spree::Shipment'
and source_id in (select id from spree_shipments #{where_order_id_in_orders_to_delete})"
sql_delete_from "spree_adjustments where source_type = 'Spree::Payment'
and source_id in (select id from spree_payments #{where_order_id_in_orders_to_delete})"
sql_delete_from "spree_adjustments where source_type = 'Spree::LineItem'
and source_id in (select id from spree_line_items #{where_order_id_in_orders_to_delete})"
end
def truncate_order_cycle_data
sql_delete_from "coordinator_fees #{where_oc_id_in_ocs_to_delete}"
sql_delete_from "
exchange_variants where exchange_id
in (select id from exchanges #{where_oc_id_in_ocs_to_delete})"
sql_delete_from "
exchange_fees where exchange_id
in (select id from exchanges #{where_oc_id_in_ocs_to_delete})"
sql_delete_from "exchanges #{where_oc_id_in_ocs_to_delete}"
exit unless HighLine.new.agree(message) { |question| question.default = "N" }
end
end
end

View File

@@ -0,0 +1,109 @@
# frozen_string_literal: true
class TruncateData
# This model lets us operate on the sessions DB table using ActiveRecord's
# methods within the scope of this service. This relies on the AR's
# convention where a Session model maps to a sessions table.
class Session < ActiveRecord::Base
end
def initialize(months_to_keep = nil)
@date = (months_to_keep || 24).to_i.months.ago
end
def call
logging do
truncate_inventory
truncate_adjustments
truncate_order_associations
truncate_order_cycle_data
sql_delete_from "spree_orders #{where_oc_id_in_ocs_to_delete}"
truncate_subscriptions
sql_delete_from "order_cycles #{where_ocs_to_delete}"
Spree::TokenizedPermission.where("created_at < '#{date}'").delete_all
remove_transient_data
end
end
private
attr_reader :date
def logging
Rails.logger.info("TruncateData started with truncation date #{date}")
yield
Rails.logger.info("TruncateData finished")
end
def truncate_order_associations
sql_delete_from "spree_line_items #{where_order_id_in_orders_to_delete}"
sql_delete_from "spree_payments #{where_order_id_in_orders_to_delete}"
sql_delete_from "spree_shipments #{where_order_id_in_orders_to_delete}"
sql_delete_from "spree_return_authorizations #{where_order_id_in_orders_to_delete}"
end
def remove_transient_data
Spree::StateChange.delete_all("created_at < '#{1.month.ago.to_date}'")
Spree::LogEntry.delete_all("created_at < '#{1.month.ago.to_date}'")
Session.delete_all("created_at < '#{2.weeks.ago.to_date}'")
end
def truncate_subscriptions
sql_delete_from "order_cycle_schedules #{where_oc_id_in_ocs_to_delete}"
sql_delete_from "proxy_orders #{where_oc_id_in_ocs_to_delete}"
end
def truncate_inventory
sql_delete_from "
spree_inventory_units #{where_order_id_in_orders_to_delete}"
sql_delete_from "
spree_inventory_units
where shipment_id in (select id from spree_shipments #{where_order_id_in_orders_to_delete})"
end
def sql_delete_from(sql)
ActiveRecord::Base.connection.execute("DELETE FROM #{sql}")
end
def where_order_id_in_orders_to_delete
"where order_id in (select id from spree_orders #{where_oc_id_in_ocs_to_delete})"
end
def where_oc_id_in_ocs_to_delete
"where order_cycle_id in (select id from order_cycles #{where_ocs_to_delete} )"
end
def where_ocs_to_delete
"where orders_close_at < '#{date}'"
end
def truncate_adjustments
sql_delete_from "spree_adjustments where source_type = 'Spree::Order'
and source_id in (select id from spree_orders #{where_oc_id_in_ocs_to_delete})"
sql_delete_from "spree_adjustments where source_type = 'Spree::Shipment'
and source_id in (select id from spree_shipments #{where_order_id_in_orders_to_delete})"
sql_delete_from "spree_adjustments where source_type = 'Spree::Payment'
and source_id in (select id from spree_payments #{where_order_id_in_orders_to_delete})"
sql_delete_from "spree_adjustments where source_type = 'Spree::LineItem'
and source_id in (select id from spree_line_items #{where_order_id_in_orders_to_delete})"
end
def truncate_order_cycle_data
sql_delete_from "coordinator_fees #{where_oc_id_in_ocs_to_delete}"
sql_delete_from "
exchange_variants where exchange_id
in (select id from exchanges #{where_oc_id_in_ocs_to_delete})"
sql_delete_from "
exchange_fees where exchange_id
in (select id from exchanges #{where_oc_id_in_ocs_to_delete})"
sql_delete_from "exchanges #{where_oc_id_in_ocs_to_delete}"
end
end

View File

@@ -0,0 +1,34 @@
require 'spec_helper'
require 'rake'
describe 'truncate_data.rake' do
describe ':truncate' do
context 'when months_to_keep is specified' do
it 'truncates order cycles closed earlier than months_to_keep months ago' do
Rake.application.rake_require 'tasks/data/truncate_data'
Rake::Task.define_task(:environment)
highline = instance_double(HighLine, agree: true)
allow(HighLine).to receive(:new).and_return(highline)
old_order_cycle = create(
:order_cycle,
orders_open_at: 7.months.ago,
orders_close_at: 7.months.ago + 1.day,
)
create(:order, order_cycle: old_order_cycle)
recent_order_cycle = create(
:order_cycle,
orders_open_at: 1.months.ago,
orders_close_at: 1.months.ago + 1.day,
)
create(:order, order_cycle: recent_order_cycle)
months_to_keep = 6
Rake.application.invoke_task "ofn:data:truncate[#{months_to_keep}]"
expect(OrderCycle.all).to contain_exactly(recent_order_cycle)
end
end
end
end

View File

@@ -0,0 +1,130 @@
require 'spec_helper'
require 'tasks/data/truncate_data'
describe TruncateData do
describe '#call' do
before do
allow(Spree::ReturnAuthorization).to receive(:delete_all)
allow(Spree::StateChange).to receive(:delete_all)
allow(Spree::LogEntry).to receive(:delete_all)
allow(TruncateData::Session).to receive(:delete_all)
allow(Rails.logger).to receive(:info)
end
context 'when months_to_keep is not specified' do
it 'truncates order cycles closed earlier than 2 years ago' do
order_cycle = create(
:order_cycle, orders_open_at: 25.months.ago, orders_close_at: 25.months.ago + 1.day
)
create(:order, order_cycle: order_cycle)
TruncateData.new.call
expect(OrderCycle.all).to be_empty
end
it 'deletes state changes older than a month' do
TruncateData.new.call
expect(Spree::StateChange)
.to have_received(:delete_all)
.with("created_at < '#{1.month.ago.to_date}'")
end
it 'deletes log entries older than a month' do
TruncateData.new.call
expect(Spree::LogEntry)
.to have_received(:delete_all)
.with("created_at < '#{1.month.ago.to_date}'")
end
it 'deletes sessions older than two weeks' do
TruncateData.new.call
expect(TruncateData::Session)
.to have_received(:delete_all)
.with("created_at < '#{2.weeks.ago.to_date}'")
end
end
context 'when months_to_keep is nil' do
it 'truncates order cycles closed earlier than 2 years ago' do
order_cycle = create(
:order_cycle, orders_open_at: 25.months.ago, orders_close_at: 25.months.ago + 1.day
)
create(:order, order_cycle: order_cycle)
TruncateData.new(nil).call
expect(OrderCycle.all).to be_empty
end
it 'deletes state changes older than a month' do
TruncateData.new.call
expect(Spree::StateChange)
.to have_received(:delete_all)
.with("created_at < '#{1.month.ago.to_date}'")
end
it 'deletes log entries older than a month' do
TruncateData.new.call
expect(Spree::LogEntry)
.to have_received(:delete_all)
.with("created_at < '#{1.month.ago.to_date}'")
end
it 'deletes sessions older than two weeks' do
TruncateData.new.call
expect(TruncateData::Session)
.to have_received(:delete_all)
.with("created_at < '#{2.weeks.ago.to_date}'")
end
end
context 'when months_to_keep is specified' do
it 'truncates order cycles closed earlier than months_to_keep months ago' do
old_order_cycle = create(
:order_cycle, orders_open_at: 7.months.ago, orders_close_at: 7.months.ago + 1.day
)
create(:order, order_cycle: old_order_cycle)
recent_order_cycle = create(
:order_cycle, orders_open_at: 1.months.ago, orders_close_at: 1.months.ago + 1.day
)
create(:order, order_cycle: recent_order_cycle)
TruncateData.new(6).call
expect(OrderCycle.all).to contain_exactly(recent_order_cycle)
end
it 'deletes state changes older than a month' do
TruncateData.new.call
expect(Spree::StateChange)
.to have_received(:delete_all)
.with("created_at < '#{1.month.ago.to_date}'")
end
it 'deletes log entries older than a month' do
TruncateData.new.call
expect(Spree::LogEntry)
.to have_received(:delete_all)
.with("created_at < '#{1.month.ago.to_date}'")
end
it 'deletes sessions older than two weeks' do
TruncateData.new.call
expect(TruncateData::Session)
.to have_received(:delete_all)
.with("created_at < '#{2.weeks.ago.to_date}'")
end
end
end
end