RUBY-2909 Snapshot Query Examples for the Manual (#2419)

* RUBY-2909 Snapshot Query Examples for the Manual

* RUBY-2909 add sharded, min mongodb version

* RUBY-2909 add w=majority

* RUBY-2909 don't leak clients

* RUBY-2909 require stress

* RUBY-2909 remove require stress

* RUBY-2909 readd require stress

* RUBY-2909 move file to stress

* RUBY-2909 make test idiomatic to ruby

* RUBY-2909 added logging (REMOVE ME)

* RUBY-2909 add 5.0 as stress test version

* Revert "RUBY-2909 add 5.0 as stress test version"

This reverts commit 62cf07e.

* RUBY-2909 move test file back and require no auth

* RUBY-2909 require no tls

* use existing test suite clients

* RUBY-2909 run distinct on each mongos

* RUBY-2909 use attr_accessor

* RUBY-2909 add comma

* RUBY-2909 do a distinct on replica sets

* wait for snapshot reads to work

Co-authored-by: Oleg Pudeyev <[email protected]>

Author: neilshweky

spec/integration/snapshot_query_examples_spec.rb

@@ -0,0 +1,127 @@
+# frozen_string_literal: true
+# encoding: utf-8
+
+require 'spec_helper'
+
+describe 'Snapshot Query Examples' do
+  require_topology :replica_set, :sharded
+  require_no_auth
+  require_no_tls
+
+  min_server_fcv '5.0'
+
+  let(:uri_string) do
+    "mongodb://#{SpecConfig.instance.addresses.join(',')}/?w=majority"
+  end
+
+  context "Snapshot Query Example 1" do
+    before do
+      client = authorized_client.use('pets')
+      client['cats'].delete_many
+      client['dogs'].delete_many
+
+      client['cats'].insert_one(
+        name: "Whiskers",
+        color: "white",
+        age: 10,
+        adoptable: true
+      )
+
+      client['dogs'].insert_one(
+        name: "Pebbles",
+        color: "Brown",
+        age: 10,
+        adoptable: true
+      )
+      if ClusterConfig.instance.topology == :sharded
+        run_mongos_distincts "pets", "cats"
+      else
+        wait_for_snapshot(db: 'pets', collection: 'cats')
+        wait_for_snapshot(db: 'pets', collection: 'dogs')
+      end
+    end
+
+    it "returns a snapshot of the data" do
+
+      adoptable_pets_count = 0
+
+      # Start Snapshot Query Example 1
+
+      client = Mongo::Client.new(uri_string, database: "pets")
+
+      client.start_session(snapshot: true) do |session|
+        adoptable_pets_count = client['cats'].aggregate([
+          { "$match": { "adoptable": true } },
+          { "$count": "adoptable_cats_count" }
+        ], session: session).first["adoptable_cats_count"]
+
+        adoptable_pets_count += client['dogs'].aggregate([
+          { "$match": { "adoptable": true } },
+          { "$count": "adoptable_dogs_count" }
+        ], session: session).first["adoptable_dogs_count"]
+
+        puts adoptable_pets_count
+      end
+
+      # End Snapshot Query Example 1
+
+      expect(adoptable_pets_count).to eq 2
+      client.close
+    end
+  end
+
+  context "Snapshot Query Example 2" do
+    before do
+      client = authorized_client.use('retail')
+      client['sales'].delete_many
+
+      client['sales'].insert_one(
+        shoeType: "boot",
+        price: 30,
+        saleDate: Time.now
+      )
+
+      if ClusterConfig.instance.topology == :sharded
+        run_mongos_distincts "retail", "sales"
+      else
+        wait_for_snapshot(db: 'retail', collection: 'sales')
+      end
+    end
+
+    it "returns a snapshot of the data" do
+
+      total = 0
+
+      # Start Snapshot Query Example 2
+
+      client = Mongo::Client.new(uri_string, database: "retail")
+
+      client.start_session(snapshot: true) do |session|
+        total = client['sales'].aggregate([
+          {
+             "$match": {
+                "$expr": {
+                   "$gt": [
+                      "$saleDate",
+                      {
+                         "$dateSubtract": {
+                            startDate: "$$NOW",
+                            unit: "day",
+                            amount: 1
+                         }
+                      }
+                   ]
+                 }
+              }
+          },
+          { "$count": "total_daily_sales" }
+        ], session: session).first["total_daily_sales"]
+      end
+
+      # End Snapshot Query Example 2
+
+      expect(total).to eq 1
+      client.close
+    end
+  end
+end

spec/lite_spec_helper.rb

@@ -81,6 +81,7 @@ module Mrss
 require 'support/common_shortcuts'
 require 'support/client_registry'
 require 'support/client_registry_macros'
+require 'support/mongos_macros'
 require 'support/crypt'
 require 'support/json_ext_formatter'
 require 'support/sdam_formatter_integration'
@@ -102,6 +103,7 @@ class ExampleTimeout < StandardError; end
   config.include(CommonShortcuts::InstanceMethods)
   config.extend(Mrss::LiteConstraints)
   config.include(ClientRegistryMacros)
+  config.include(MongosMacros)
 
   if SpecConfig.instance.ci?
     SdamFormatterIntegration.subscribe

spec/runners/transactions/test.rb

@@ -22,6 +22,7 @@ module Transactions
     #
     # @since 2.6.0
     class TransactionsTest < CRUD::CRUDTestBase
+      include MongosMacros
 
       attr_reader :expected_results
       attr_reader :skip_reason
@@ -262,11 +263,8 @@ def setup_test
 
         coll.insert_many(@data) unless @data.empty?
 
-        $distinct_ran ||= {}
-        $distinct_ran[@spec.database_name] ||= if description =~ /distinct/ || @operations.any? { |op| op.name == 'distinct' }
-          ::Utils.mongos_each_direct_client do |direct_client|
-            direct_client.use(@spec.database_name)['test'].distinct('foo').to_a
-          end
+        if description =~ /distinct/ || @operations.any? { |op| op.name == 'distinct' }
+          run_mongos_distincts(@spec.database_name, 'test')
         end
 
         admin_support_client.command(@fail_point_command) if @fail_point_command

spec/support/common_shortcuts.rb

@@ -370,5 +370,27 @@ def mock_dns(config)
         thread.join
       end
     end
+
+    # Wait for snapshot reads to become available to prevent this error:
+    # [246:SnapshotUnavailable]: Unable to read from a snapshot due to pending collection catalog changes; please retry the operation. Snapshot timestamp is Timestamp(1646666892, 4). Collection minimum is Timestamp(1646666892, 5) (on localhost:27017, modern retry, attempt 1)
+    def wait_for_snapshot(db: nil, collection: nil, client: nil)
+      client ||= authorized_client
+      client = client.use(db) if db
+      collection ||= 'any'
+      start_time = Mongo::Utils.monotonic_time
+      begin
+        client.start_session(snapshot: true) do |session|
+          client[collection].aggregate([{'$match': {any: true}}], session: session).to_a
+        end
+      rescue Mongo::Error::OperationFailure => e
+        # Retry them as the server demands...
+        if e.code == 246 # SnapshotUnavailable
+          if Mongo::Utils.monotonic_time < start_time + 10
+            retry
+          end
+        end
+        raise
+      end
+    end
   end
 end

spec/support/mongos_macros.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module MongosMacros
+
+  class << self
+    attr_accessor :distinct_ran
+  end
+  self.distinct_ran = {}
+
+  # Work around for SERVER-39704 when seeing a Mongo::Error::OperationFailure
+  # SnapshotUnavailable error -- run the distinct command on each mongos.
+  def run_mongos_distincts(db_name, collection='test')
+    MongosMacros.distinct_ran[db_name] ||= ::Utils.mongos_each_direct_client do |direct_client|
+      direct_client.use(db_name)[collection].distinct('foo').to_a
+    end
+  end
+end