Merge pull request #2220 from internetee/optimize-verification-model-v2

decrease db load

Gemfile

@@ -98,3 +98,7 @@ group :test do
 end
 
 gem 'aws-sdk-sesv2', '~> 1.19'
+
+# profiles
+gem 'pghero'
+gem 'pg_query', '>= 0.9.0'

Gemfile.lock

@@ -233,6 +233,8 @@ GEM
       thor (>= 0.14.0, < 2)
     globalid (0.5.2)
       activesupport (>= 5.0)
+    google-protobuf (3.19.1)
+    google-protobuf (3.19.1-x86_64-linux)
     gyoku (1.3.1)
       builder (>= 2.1.2)
     haml (5.2.2)
@@ -338,6 +340,10 @@ GEM
       request_store (~> 1.1)
     pdfkit (0.8.5)
     pg (1.2.3)
+    pg_query (2.1.2)
+      google-protobuf (>= 3.17.1)
+    pghero (2.8.1)
+      activerecord (>= 5)
     pry (0.14.1)
       coderay (~> 1.1)
       method_source (~> 1.0)
@@ -555,6 +561,8 @@ DEPENDENCIES
   paper_trail (~> 12.1)
   pdfkit
   pg (= 1.2.3)
+  pg_query (>= 0.9.0)
+  pghero
   pry (= 0.14.1)
   puma
   que
@@ -579,4 +587,4 @@ DEPENDENCIES
   wkhtmltopdf-binary (~> 0.12.5.1)
 
 BUNDLED WITH
-   2.2.27
+   2.2.31

app/jobs/verify_emails_job.rb

@@ -1,12 +1,8 @@
 class VerifyEmailsJob < ApplicationJob
   discard_on StandardError
 
-  def perform(contact_id:, check_level: 'regex')
-    contact = Contact.find_by(id: contact_id)
-
-    return if check_contact_for_duplicate_mail(contact_id)
-
-    contact_not_found(contact_id) unless contact
+  def perform(contact:, check_level: 'regex')
+    contact_not_found(contact.id) unless contact
     validate_check_level(check_level)
     action = Actions::EmailCheck.new(email: contact.email,
                                      validation_eventable: contact,
@@ -19,16 +15,6 @@ def perform(contact_id:, check_level: 'regex')
 
   private
 
-  def check_contact_for_duplicate_mail(contact_id)
-    time = Time.zone.now - ValidationEvent::VALIDATION_PERIOD
-    contact = Contact.find(contact_id)
-    contact_ids = Contact.where(email: contact.email).where('created_at > ?', time).pluck(:id)
-
-    r = ValidationEvent.where(validation_eventable_id: contact_ids).order(created_at: :desc)
-
-    r.present?
-  end
-
   def contact_not_found(contact_id)
     raise StandardError, "Contact with contact_id #{contact_id} not found"
   end

app/models/validation_event.rb

@@ -9,10 +9,11 @@ class ValidationEvent < ApplicationRecord
   VALIDATION_PERIOD = 1.year.freeze
   VALID_CHECK_LEVELS = %w[regex mx smtp].freeze
   VALID_EVENTS_COUNT_THRESHOLD = 5
+  MX_CHECK = 3
 
   INVALID_EVENTS_COUNT_BY_LEVEL = {
     regex: 1,
-    mx: 3,
+    mx: MX_CHECK,
     smtp: 1,
   }.freeze
 

config/routes.rb

@@ -5,6 +5,12 @@
   get 'practice/index'
   get 'practice/contact'
   # https://github.com/internetee/epp_proxy#translation-of-epp-calls
+  #
+  # profiles
+  if Rails.env.development? || Rails.env.staging?
+    mount PgHero::Engine, at: "pghero"
+  end
+
   namespace :epp do
     constraints(EppConstraint.new(:session)) do
       get 'session/hello', to: 'sessions#hello', as: 'hello'

db/migrate/20211124071418_create_pghero_query_stats.rb

@@ -0,0 +1,15 @@
+class CreatePgheroQueryStats < ActiveRecord::Migration[6.1]
+  def change
+    create_table :pghero_query_stats do |t|
+      t.text :database
+      t.text :user
+      t.text :query
+      t.integer :query_hash, limit: 8
+      t.float :total_time
+      t.integer :calls, limit: 8
+      t.timestamp :captured_at
+    end
+
+    add_index :pghero_query_stats, [:database, :captured_at]
+  end
+end

db/structure.sql

@@ -71,7 +71,8 @@ COMMENT ON EXTENSION pgcrypto IS 'cryptographic functions';
 
 CREATE TYPE public.validation_type AS ENUM (
     'email_validation',
-    'manual_force_delete'
+    'manual_force_delete',
+    'nameserver_validation'
 );
 
 
@@ -2260,6 +2261,41 @@ CREATE SEQUENCE public.payment_orders_id_seq
 ALTER SEQUENCE public.payment_orders_id_seq OWNED BY public.payment_orders.id;
 
 
+--
+-- Name: pghero_query_stats; Type: TABLE; Schema: public; Owner: -
+--
+
+CREATE TABLE public.pghero_query_stats (
+    id bigint NOT NULL,
+    database text,
+    "user" text,
+    query text,
+    query_hash bigint,
+    total_time double precision,
+    calls bigint,
+    captured_at timestamp without time zone
+);
+
+
+--
+-- Name: pghero_query_stats_id_seq; Type: SEQUENCE; Schema: public; Owner: -
+--
+
+CREATE SEQUENCE public.pghero_query_stats_id_seq
+    START WITH 1
+    INCREMENT BY 1
+    NO MINVALUE
+    NO MAXVALUE
+    CACHE 1;
+
+
+--
+-- Name: pghero_query_stats_id_seq; Type: SEQUENCE OWNED BY; Schema: public; Owner: -
+--
+
+ALTER SEQUENCE public.pghero_query_stats_id_seq OWNED BY public.pghero_query_stats.id;
+
+
 --
 -- Name: prices; Type: TABLE; Schema: public; Owner: -
 --
@@ -2619,8 +2655,7 @@ CREATE TABLE public.validation_events (
     validation_eventable_type character varying,
     validation_eventable_id bigint,
     created_at timestamp(6) without time zone NOT NULL,
-    updated_at timestamp(6) without time zone NOT NULL,
-    event_type public.validation_type
+    updated_at timestamp(6) without time zone NOT NULL
 );
 
 
@@ -3162,6 +3197,13 @@ ALTER TABLE ONLY public.notifications ALTER COLUMN id SET DEFAULT nextval('publi
 ALTER TABLE ONLY public.payment_orders ALTER COLUMN id SET DEFAULT nextval('public.payment_orders_id_seq'::regclass);
 
 
+--
+-- Name: pghero_query_stats id; Type: DEFAULT; Schema: public; Owner: -
+--
+
+ALTER TABLE ONLY public.pghero_query_stats ALTER COLUMN id SET DEFAULT nextval('public.pghero_query_stats_id_seq'::regclass);
+
+
 --
 -- Name: prices id; Type: DEFAULT; Schema: public; Owner: -
 --
@@ -3685,6 +3727,14 @@ ALTER TABLE ONLY public.payment_orders
     ADD CONSTRAINT payment_orders_pkey PRIMARY KEY (id);
 
 
+--
+-- Name: pghero_query_stats pghero_query_stats_pkey; Type: CONSTRAINT; Schema: public; Owner: -
+--
+
+ALTER TABLE ONLY public.pghero_query_stats
+    ADD CONSTRAINT pghero_query_stats_pkey PRIMARY KEY (id);
+
+
 --
 -- Name: prices prices_pkey; Type: CONSTRAINT; Schema: public; Owner: -
 --
@@ -4049,6 +4099,13 @@ CREATE INDEX index_domain_transfers_on_domain_id ON public.domain_transfers USIN
 CREATE INDEX index_domains_on_delete_date ON public.domains USING btree (delete_date);
 
 
+--
+-- Name: index_domains_on_json_statuses_history; Type: INDEX; Schema: public; Owner: -
+--
+
+CREATE INDEX index_domains_on_json_statuses_history ON public.domains USING gin (json_statuses_history);
+
+
 --
 -- Name: index_domains_on_name; Type: INDEX; Schema: public; Owner: -
 --
@@ -4441,6 +4498,13 @@ CREATE INDEX index_notifications_on_registrar_id ON public.notifications USING b
 CREATE INDEX index_payment_orders_on_invoice_id ON public.payment_orders USING btree (invoice_id);
 
 
+--
+-- Name: index_pghero_query_stats_on_database_and_captured_at; Type: INDEX; Schema: public; Owner: -
+--
+
+CREATE INDEX index_pghero_query_stats_on_database_and_captured_at ON public.pghero_query_stats USING btree (database, captured_at);
+
+
 --
 -- Name: index_prices_on_zone_id; Type: INDEX; Schema: public; Owner: -
 --
@@ -4490,13 +4554,6 @@ CREATE INDEX index_users_on_identity_code ON public.users USING btree (identity_
 CREATE INDEX index_users_on_registrar_id ON public.users USING btree (registrar_id);
 
 
---
--- Name: index_validation_events_on_event_type; Type: INDEX; Schema: public; Owner: -
---
-
-CREATE INDEX index_validation_events_on_event_type ON public.validation_events USING btree (event_type);
-
-
 --
 -- Name: index_validation_events_on_validation_eventable; Type: INDEX; Schema: public; Owner: -
 --
@@ -5232,7 +5289,12 @@ INSERT INTO "schema_migrations" (version) VALUES
 ('20210215101019'),
 ('20210616112332'),
 ('20210629074044'),
-('20210628090353'),
 ('20210708131814'),
 ('20210729131100'),
-('20210729134625');
+('20210729134625'),
+('20211028122103'),
+('20211028125245'),
+('20211029082225'),
+('20211124071418');
+
+

lib/tasks/generate_mock_contacts.rake

@@ -0,0 +1,28 @@
+# namespace :generate_mock do
+#   task contacts: :environment do
+#     1000.times do
+#       c = Contact.new
+#       c.name = generate_random_string
+#       c.email = generate_random_string + "@" + generate_random_string + ".ee"
+#       c.registrar_id = registrar
+#       c.street = generate_random_string
+#       c.city = generate_random_string
+#       c.zip = '12323'
+#       c.country_code = 'EE'
+#       c.phone = "+372.59813318"
+#       c.ident_country_code = 'EE'
+#       c.ident_type = 'priv'
+#       c.ident = '38903110313'
+#       c.code = generate_random_string + ":" + generate_random_string
+#       c.save
+#     end
+#   end
+#
+#   def generate_random_string
+#     (0...10).map { (65 + rand(26)).chr }.join
+#   end
+#
+#     def registrar
+#       Registrar.last.id
+#     end
+# end

lib/tasks/verify_email.rake

@@ -1,6 +1,7 @@
 require 'optparse'
 require 'rake_option_parser_boilerplate'
 require 'syslog/logger'
+require 'active_record'
 
 namespace :verify_email do
   # bundle exec rake verify_email:check_all -- --domain_name=shop.test --check_level=mx --spam_protect=true
@@ -18,14 +19,16 @@ namespace :verify_email do
                                                        banner: banner,
                                                        hash: opts_hash)
 
-    contacts = prepare_contacts(options)
-    logger.info 'No contacts to check email selected' and next if contacts.blank?
+    batch_contacts = prepare_contacts(options)
+    logger.info 'No contacts to check email selected' and next if batch_contacts.blank?
 
-    contacts.each do |contact|
-      VerifyEmailsJob.set(wait_until: spam_protect_timeout(options)).perform_later(
-        contact_id: contact.id,
-        check_level: check_level(options)
-      )
+    batch_contacts.find_in_batches(batch_size: 10_000) do |contacts|
+      contacts.each do |contact|
+        VerifyEmailsJob.set(wait_until: spam_protect_timeout(options)).perform_later(
+          contact: contact,
+          check_level: check_level(options)
+        )
+      end
     end
   end
 end
@@ -55,35 +58,40 @@ def prepare_contacts(options)
     contacts_by_domain(options[:domain_name])
   else
     time = Time.zone.now - ValidationEvent::VALIDATION_PERIOD
-    validation_events_ids = ValidationEvent.where('created_at > ?', time).pluck(:validation_eventable_id)
+    validation_events_ids = ValidationEvent.where('created_at > ?', time).distinct.pluck(:validation_eventable_id)
 
-    # Contact.where.not(id: validation_events_ids) + Contact.where(id: failed_contacts)
-    Contact.where.not(id: validation_events_ids) | failed_contacts
+    contacts_ids = Contact.where.not(id: validation_events_ids).pluck(:id)
+    Contact.where(id: contacts_ids + failed_contacts)
   end
 end
 
 def failed_contacts
   failed_contacts = []
-  failed_validations_ids = ValidationEvent.failed.pluck(:validation_eventable_id)
-  contacts = Contact.where(id: failed_validations_ids)
-  contacts.each do |contact|
+  failed_validations_ids = ValidationEvent.failed.distinct.pluck(:validation_eventable_id)
+  contacts = Contact.where(id: failed_validations_ids).includes(:validation_events)
+  contacts.find_each(batch_size: 10_000) do |contact|
 
-    if contact.validation_events.mx.order(created_at: :asc).present?
-      failed_contacts << contact unless contact.validation_events.mx.order(created_at: :asc).last.success
-    end
+    data = contact.validation_events.order(created_at: :asc).last
 
-    if contact.validation_events.regex.order(created_at: :asc).present?
-      failed_contacts << contact unless contact.validation_events.regex.order(created_at: :asc).last.success
-    end
+    if data.failed?
+      next if data.event_data['check_level'] == 'regex'
+
+      next if data.event_data['check_level'] == 'smtp'
 
-    if contact.validation_events.smtp.order(created_at: :asc).present?
-      failed_contacts << contact unless contact.validation_events.mx.order(created_at: :asc).last.success
+      next if check_mx_contact_validation(contact)
+
+      failed_contacts << contact.id
     end
   end
 
   failed_contacts.uniq
 end
 
+def check_mx_contact_validation(contact)
+  data = contact.validation_events.mx.order(created_at: :asc).last(ValidationEvent::MX_CHECK)
+  data.all? { |d| d.failed? }
+end
+
 def contacts_by_domain(domain_name)
   domain = ::Domain.find_by(name: domain_name)
   return unless domain