cluster.rb in mongo-2.16.0.alpha1

- old
+ new

@@ -157,10 +157,14 @@
       # @sdam_flow_lock covers just the sdam flow. Note it does not apply
       # to @topology replacements which are done under @update_lock.
       @sdam_flow_lock = Mutex.new
       Session::SessionPool.create(self)
 
+      if seeds.empty? && load_balanced?
+        raise ArgumentError, 'Load-balanced clusters with no seeds are prohibited'
+      end
+
       # The opening topology is always unknown with no servers.
       # https://github.com/mongodb/specifications/pull/388
       opening_topology = Topology::Unknown.new(options, monitoring, self)
 
       publish_sdam_event(
@@ -169,25 +173,34 @@
       )
 
       @seeds = seeds = seeds.uniq
       servers = seeds.map do |seed|
         # Server opening events must be sent after topology change events.
-        # Therefore separate server addition, done here before topoolgy change
+        # Therefore separate server addition, done here before topology change
         # event is published, from starting to monitor the server which is
         # done later.
         add(seed, monitor: false)
       end
 
       if seeds.size >= 1
         # Recreate the topology to get the current server list into it
-        @topology = topology.class.new(topology.options, topology.monitoring, self)
-        publish_sdam_event(
-          Monitoring::TOPOLOGY_CHANGED,
-          Monitoring::Event::TopologyChanged.new(opening_topology, @topology)
-        )
+        recreate_topology(topology, opening_topology)
       end
 
+      if load_balanced?
+        # We are required by the specifications to produce certain SDAM events
+        # when in load-balanced topology.
+        # These events don't make a lot of sense from the standpoint of the
+        # driver's SDAM implementation, nor from the standpoint of the
+        # driver's load balancer implementation.
+        # They are just required boilerplate.
+        #
+        # Note that this call must be done above the monitoring_io check
+        # because that short-circuits the rest of the constructor.
+        fabricate_lb_sdam_events_and_set_server_type
+      end
+
       if options[:monitoring_io] == false
         # Omit periodic executor construction, because without servers
         # no commands can be sent to the cluster and there shouldn't ever
         # be anything that needs to be cleaned up.
         #
@@ -201,67 +214,69 @@
       # Update instance variables prior to starting monitoring threads.
       @connecting = false
       @connected = true
 
       if options[:cleanup] != false
-        @cursor_reaper = CursorReaper.new
+        @cursor_reaper = CursorReaper.new(self)
         @socket_reaper = SocketReaper.new(self)
         @periodic_executor = PeriodicExecutor.new([
           @cursor_reaper, @socket_reaper,
         ], options)
 
         @periodic_executor.run!
       end
 
-      # Need to record start time prior to starting monitoring
-      start_monotime = Utils.monotonic_time
+      unless load_balanced?
+        # Need to record start time prior to starting monitoring
+        start_monotime = Utils.monotonic_time
 
-      servers.each do |server|
-        server.start_monitoring
-      end
-
-      if options[:scan] != false
-        server_selection_timeout = options[:server_selection_timeout] || ServerSelector::SERVER_SELECTION_TIMEOUT
-        # The server selection timeout can be very short especially in
-        # tests, when the client waits for a synchronous scan before
-        # starting server selection. Limiting the scan to server selection time
-        # then aborts the scan before it can process even local servers.
-        # Therefore, allow at least 3 seconds for the scan here.
-        if server_selection_timeout < 3
-          server_selection_timeout = 3
+        servers.each do |server|
+          server.start_monitoring
         end
-        deadline = start_monotime + server_selection_timeout
-        # Wait for the first scan of each server to complete, for
-        # backwards compatibility.
-        # If any servers are discovered during this SDAM round we are going to
-        # wait for these servers to also be queried, and so on, up to the
-        # server selection timeout or the 3 second minimum.
-        loop do
-          # Ensure we do not try to read the servers list while SDAM is running
-          servers = @sdam_flow_lock.synchronize do
-            servers_list.dup
+
+        if options[:scan] != false
+          server_selection_timeout = options[:server_selection_timeout] || ServerSelector::SERVER_SELECTION_TIMEOUT
+          # The server selection timeout can be very short especially in
+          # tests, when the client waits for a synchronous scan before
+          # starting server selection. Limiting the scan to server selection time
+          # then aborts the scan before it can process even local servers.
+          # Therefore, allow at least 3 seconds for the scan here.
+          if server_selection_timeout < 3
+            server_selection_timeout = 3
           end
-          if servers.all? { |server| server.last_scan_monotime && server.last_scan_monotime >= start_monotime }
-            break
+          deadline = start_monotime + server_selection_timeout
+          # Wait for the first scan of each server to complete, for
+          # backwards compatibility.
+          # If any servers are discovered during this SDAM round we are going to
+          # wait for these servers to also be queried, and so on, up to the
+          # server selection timeout or the 3 second minimum.
+          loop do
+            # Ensure we do not try to read the servers list while SDAM is running
+            servers = @sdam_flow_lock.synchronize do
+              servers_list.dup
+            end
+            if servers.all? { |server| server.last_scan_monotime && server.last_scan_monotime >= start_monotime }
+              break
+            end
+            if (time_remaining = deadline - Utils.monotonic_time) <= 0
+              break
+            end
+            log_debug("Waiting for up to #{'%.2f' % time_remaining} seconds for servers to be scanned: #{summary}")
+            # Since the semaphore may have been signaled between us checking
+            # the servers list above and the wait call below, we should not
+            # wait for the full remaining time - wait for up to 1 second, then
+            # recheck the state.
+            begin
+              server_selection_semaphore.wait([time_remaining, 1].min)
+            rescue ::Timeout::Error
+              # nothing
+            end
           end
-          if (time_remaining = deadline - Utils.monotonic_time) <= 0
-            break
-          end
-          log_debug("Waiting for up to #{'%.2f' % time_remaining} seconds for servers to be scanned: #{summary}")
-          # Since the semaphore may have been signaled between us checking
-          # the servers list above and the wait call below, we should not
-          # wait for the full remaining time - wait for up to 1 second, then
-          # recheck the state.
-          begin
-            server_selection_semaphore.wait([time_remaining, 1].min)
-          rescue ::Timeout::Error
-            # nothing
-          end
         end
-      end
 
-      start_stop_srv_monitor
+        start_stop_srv_monitor
+      end
     end
 
     # Create a cluster for the provided client, for use when we don't want the
     # client's original cluster instance to be the same.
     #
@@ -319,10 +334,18 @@
     attr_reader :session_pool
 
     def_delegators :topology, :replica_set?, :replica_set_name, :sharded?,
                    :single?, :unknown?
 
+    # Returns whether the cluster is configured to be in the load-balanced
+    # topology.
+    #
+    # @return [ true | false ] Whether the topology is load-balanced.
+    def load_balanced?
+      topology.is_a?(Topology::LoadBalanced)
+    end
+
     [:register_cursor, :schedule_kill_cursor, :unregister_cursor].each do |m|
       define_method(m) do |*args|
         if options[:cleanup] != false
           @cursor_reaper.send(m, *args)
         end
@@ -597,13 +620,29 @@
     #   respective server is cleared. Set this option to true to keep the
     #   existing connection pool (required when handling not master errors
     #   on 4.2+ servers).
     # @option aptions [ true | false ] :awaited Whether the updated description
     #   was a result of processing an awaited hello.
+    # @option options [ Object ] :service_id Change state for the specified
+    #   service id only.
     #
     # @api private
     def run_sdam_flow(previous_desc, updated_desc, options = {})
+      if load_balanced?
+        if updated_desc.config.empty?
+          unless options[:keep_connection_pool]
+            servers_list.each do |server|
+              # TODO should service id be taken out of updated_desc?
+              # We could also assert that
+              # options[:service_id] == updated_desc.service_id
+              server.clear_connection_pool(service_id: options[:service_id])
+            end
+          end
+        end
+        return
+      end
+
       @sdam_flow_lock.synchronize do
         flow = SdamFlow.new(self, previous_desc, updated_desc,
           awaited: options[:awaited])
         flow.server_description_changed
 
@@ -780,12 +819,19 @@
     #
     # @since 2.0.0
     def add(host, add_options=nil)
       address = Address.new(host, options)
       if !addresses.include?(address)
-        server = Server.new(address, self, @monitoring, event_listeners, options.merge(
-          monitor: false))
+        opts = options.merge(monitor: false)
+        # Note that in a load-balanced topology, every server must be a
+        # load balancer (load_balancer: true is specified in the options)
+        # but this option isn't set here because we are required by the
+        # specifications to pretent the server started out as an unknown one
+        # and publish server description change event into the load balancer
+        # one. The actual correct description for this server will be set
+        # by the fabricate_lb_sdam_events_and_set_server_type method.
+        server = Server.new(address, self, @monitoring, event_listeners, opts)
         @update_lock.synchronize do
           # Need to recheck whether server is present in @servers, because
           # the previous check was not under a lock.
           # Since we are under the update lock here, we cannot call servers_list.
           return if @servers.map(&:address).include?(address)
@@ -897,10 +943,14 @@
     # least one server. If the client has never connected to any servers,
     # the deployment is considered to not support sessions.
     #
     # @api private
     def validate_session_support!
+      if topology.is_a?(Topology::LoadBalanced)
+        return
+      end
+
       @state_change_lock.synchronize do
         @sdam_flow_lock.synchronize do
           if topology.data_bearing_servers?
             unless topology.logical_session_timeout
               raise_sessions_not_supported
@@ -979,9 +1029,44 @@
       else
         "The following servers have null logical session timeout: #{offending_servers.map(&:address).map(&:seed).join(', ')}"
       end
       msg = "The deployment that the driver is connected to does not support sessions: #{reason}"
       raise Error::SessionsNotSupported, msg
+    end
+
+    def fabricate_lb_sdam_events_and_set_server_type
+      # Although there is no monitoring connection in load balanced mode,
+      # we must emit the following series of SDAM events.
+      server = @servers.first
+      # We are guaranteed to have the server here.
+      server.publish_opening_event
+      server_desc = server.description
+      # This is where a load balancer actually gets its correct server
+      # description.
+      server.update_description(
+        Server::Description.new(server.address, {},
+          load_balancer: true,
+          force_load_balancer: options[:connect] == :load_balanced,
+        )
+      )
+      publish_sdam_event(
+        Monitoring::SERVER_DESCRIPTION_CHANGED,
+        Monitoring::Event::ServerDescriptionChanged.new(
+          server.address,
+          topology,
+          server_desc,
+          server.description
+        )
+      )
+      recreate_topology(topology, topology)
+    end
+
+    def recreate_topology(new_topology_template, previous_topology)
+      @topology = topology.class.new(new_topology_template.options, new_topology_template.monitoring, self)
+      publish_sdam_event(
+        Monitoring::TOPOLOGY_CHANGED,
+        Monitoring::Event::TopologyChanged.new(previous_topology, @topology)
+      )
     end
   end
 end
 
 require 'mongo/cluster/sdam_flow'