Class | Net::SSH::Connection::Session |
In: |
lib/net/ssh/connection/session.rb
lib/net/ssh/connection/session.rb |
Parent: | Object |
A session class representing the connection service running on top of the SSH transport layer. It manages the creation of channels (see open_channel), and the dispatching of messages to the various channels. It also encapsulates the SSH event loop (via loop and process), and serves as a central point-of-reference for all SSH-related services (e.g. port forwarding, SFTP, SCP, etc.).
You will rarely (if ever) need to instantiate this class directly; rather, you‘ll almost always use Net::SSH.start to initialize a new network connection, authenticate a user, and return a new connection session, all in one call.
Net::SSH.start("localhost", "user") do |ssh| # 'ssh' is an instance of Net::SSH::Connection::Session ssh.exec! "/etc/init.d/some_process start" end
MAP | = | Constants.constants.inject({}) do |memo, name| value = const_get(name) |
MAP | = | Constants.constants.inject({}) do |memo, name| value = const_get(name) |
loop | -> | loop_forever |
preserve a reference to Kernel#loop |
options | [R] | The map of options that were used to initialize this instance. |
options | [R] | The map of options that were used to initialize this instance. |
properties | [R] | The collection of custom properties for this instance. (See #[] and #[]=). |
properties | [R] | The collection of custom properties for this instance. (See #[] and #[]=). |
transport | [R] | The underlying transport layer abstraction (see Net::SSH::Transport::Session). |
transport | [R] | The underlying transport layer abstraction (see Net::SSH::Transport::Session). |
Create a new connection service instance atop the given transport layer. Initializes the listeners to be only the underlying socket object.
# File lib/net/ssh/connection/session.rb, line 61 61: def initialize(transport, options={}) 62: self.logger = transport.logger 63: 64: @transport = transport 65: @options = options 66: 67: @channel_id_counter = -1 68: @channels = Hash.new(NilChannel.new(self)) 69: @listeners = { transport.socket => nil } 70: @pending_requests = [] 71: @channel_open_handlers = {} 72: @on_global_request = {} 73: @properties = (options[:properties] || {}).dup 74: end
Create a new connection service instance atop the given transport layer. Initializes the listeners to be only the underlying socket object.
# File lib/net/ssh/connection/session.rb, line 61 61: def initialize(transport, options={}) 62: self.logger = transport.logger 63: 64: @transport = transport 65: @options = options 66: 67: @channel_id_counter = -1 68: @channels = Hash.new(NilChannel.new(self)) 69: @listeners = { transport.socket => nil } 70: @pending_requests = [] 71: @channel_open_handlers = {} 72: @on_global_request = {} 73: @properties = (options[:properties] || {}).dup 74: end
Sets a custom property for this instance.
# File lib/net/ssh/connection/session.rb, line 84 84: def []=(key, value) 85: @properties[key] = value 86: end
Sets a custom property for this instance.
# File lib/net/ssh/connection/session.rb, line 84 84: def []=(key, value) 85: @properties[key] = value 86: end
Returns true if there are any channels currently active on this session. By default, this will not include "invisible" channels (such as those created by forwarding ports and such), but if you pass a true value for include_invisible, then those will be counted.
This can be useful for determining whether the event loop should continue to be run.
ssh.loop { ssh.busy? }
# File lib/net/ssh/connection/session.rb, line 133 133: def busy?(include_invisible=false) 134: if include_invisible 135: channels.any? 136: else 137: channels.any? { |id, ch| !ch[:invisible] } 138: end 139: end
Returns true if there are any channels currently active on this session. By default, this will not include "invisible" channels (such as those created by forwarding ports and such), but if you pass a true value for include_invisible, then those will be counted.
This can be useful for determining whether the event loop should continue to be run.
ssh.loop { ssh.busy? }
# File lib/net/ssh/connection/session.rb, line 133 133: def busy?(include_invisible=false) 134: if include_invisible 135: channels.any? 136: else 137: channels.any? { |id, ch| !ch[:invisible] } 138: end 139: end
Closes the session gracefully, blocking until all channels have successfully closed, and then closes the underlying transport layer connection.
# File lib/net/ssh/connection/session.rb, line 106 106: def close 107: info { "closing remaining channels (#{channels.length} open)" } 108: channels.each { |id, channel| channel.close } 109: loop { channels.any? } 110: transport.close 111: end
Closes the session gracefully, blocking until all channels have successfully closed, and then closes the underlying transport layer connection.
# File lib/net/ssh/connection/session.rb, line 106 106: def close 107: info { "closing remaining channels (#{channels.length} open)" } 108: channels.each { |id, channel| channel.close } 109: loop { channels.any? } 110: transport.close 111: end
Returns true if the underlying transport has been closed. Note that this can be a little misleading, since if the remote server has closed the connection, the local end will still think it is open until the next operation on the socket. Nevertheless, this method can be useful if you just want to know if you have closed the connection.
# File lib/net/ssh/connection/session.rb, line 99 99: def closed? 100: transport.closed? 101: end
Returns true if the underlying transport has been closed. Note that this can be a little misleading, since if the remote server has closed the connection, the local end will still think it is open until the next operation on the socket. Nevertheless, this method can be useful if you just want to know if you have closed the connection.
# File lib/net/ssh/connection/session.rb, line 99 99: def closed? 100: transport.closed? 101: end
A convenience method for executing a command and interacting with it. If no block is given, all output is printed via $stdout and $stderr. Otherwise, the block is called for each data and extended data packet, with three arguments: the channel object, a symbol indicating the data type (:stdout or :stderr), and the data (as a string).
Note that this method returns immediately, and requires an event loop (see Session#loop) in order for the command to actually execute.
This is effectively identical to calling open_channel, and then Net::SSH::Connection::Channel#exec, and then setting up the channel callbacks. However, for most uses, this will be sufficient.
ssh.exec "grep something /some/files" do |ch, stream, data| if stream == :stderr puts "ERROR: #{data}" else puts data end end
# File lib/net/ssh/connection/session.rb, line 318 318: def exec(command, &block) 319: open_channel do |channel| 320: channel.exec(command) do |ch, success| 321: raise "could not execute command: #{command.inspect}" unless success 322: 323: channel.on_data do |ch2, data| 324: if block 325: block.call(ch2, :stdout, data) 326: else 327: $stdout.print(data) 328: end 329: end 330: 331: channel.on_extended_data do |ch2, type, data| 332: if block 333: block.call(ch2, :stderr, data) 334: else 335: $stderr.print(data) 336: end 337: end 338: end 339: end 340: end
A convenience method for executing a command and interacting with it. If no block is given, all output is printed via $stdout and $stderr. Otherwise, the block is called for each data and extended data packet, with three arguments: the channel object, a symbol indicating the data type (:stdout or :stderr), and the data (as a string).
Note that this method returns immediately, and requires an event loop (see Session#loop) in order for the command to actually execute.
This is effectively identical to calling open_channel, and then Net::SSH::Connection::Channel#exec, and then setting up the channel callbacks. However, for most uses, this will be sufficient.
ssh.exec "grep something /some/files" do |ch, stream, data| if stream == :stderr puts "ERROR: #{data}" else puts data end end
# File lib/net/ssh/connection/session.rb, line 318 318: def exec(command, &block) 319: open_channel do |channel| 320: channel.exec(command) do |ch, success| 321: raise "could not execute command: #{command.inspect}" unless success 322: 323: channel.on_data do |ch2, data| 324: if block 325: block.call(ch2, :stdout, data) 326: else 327: $stdout.print(data) 328: end 329: end 330: 331: channel.on_extended_data do |ch2, type, data| 332: if block 333: block.call(ch2, :stderr, data) 334: else 335: $stderr.print(data) 336: end 337: end 338: end 339: end 340: end
Same as exec, except this will block until the command finishes. Also, if a block is not given, this will return all output (stdout and stderr) as a single string.
matches = ssh.exec!("grep something /some/files")
# File lib/net/ssh/connection/session.rb, line 347 347: def exec!(command, &block) 348: block ||= Proc.new do |ch, type, data| 349: ch[:result] ||= "" 350: ch[:result] << data 351: end 352: 353: channel = exec(command, &block) 354: channel.wait 355: 356: return channel[:result] 357: end
Same as exec, except this will block until the command finishes. Also, if a block is not given, this will return all output (stdout and stderr) as a single string.
matches = ssh.exec!("grep something /some/files")
# File lib/net/ssh/connection/session.rb, line 347 347: def exec!(command, &block) 348: block ||= Proc.new do |ch, type, data| 349: ch[:result] ||= "" 350: ch[:result] << data 351: end 352: 353: channel = exec(command, &block) 354: channel.wait 355: 356: return channel[:result] 357: end
Returns a reference to the Net::SSH::Service::Forward service, which can be used for forwarding ports over SSH.
# File lib/net/ssh/connection/session.rb, line 416 416: def forward 417: @forward ||= Service::Forward.new(self) 418: end
Returns a reference to the Net::SSH::Service::Forward service, which can be used for forwarding ports over SSH.
# File lib/net/ssh/connection/session.rb, line 416 416: def forward 417: @forward ||= Service::Forward.new(self) 418: end
Adds an IO object for the event loop to listen to. If a callback is given, it will be invoked when the io is ready to be read, otherwise, the io will merely have its fill method invoked.
Any io value passed to this method must have mixed into it the Net::SSH::BufferedIo functionality, typically by calling extend on the object.
The following example executes a process on the remote server, opens a socket to somewhere, and then pipes data from that socket to the remote process’ stdin stream:
channel = ssh.open_channel do |ch| ch.exec "/some/process/that/wants/input" do |ch, success| abort "can't execute!" unless success io = TCPSocket.new(somewhere, port) io.extend(Net::SSH::BufferedIo) ssh.listen_to(io) ch.on_process do if io.available > 0 ch.send_data(io.read_available) end end ch.on_close do ssh.stop_listening_to(io) io.close end end end channel.wait
# File lib/net/ssh/connection/session.rb, line 404 404: def listen_to(io, &callback) 405: listeners[io] = callback 406: end
Adds an IO object for the event loop to listen to. If a callback is given, it will be invoked when the io is ready to be read, otherwise, the io will merely have its fill method invoked.
Any io value passed to this method must have mixed into it the Net::SSH::BufferedIo functionality, typically by calling extend on the object.
The following example executes a process on the remote server, opens a socket to somewhere, and then pipes data from that socket to the remote process’ stdin stream:
channel = ssh.open_channel do |ch| ch.exec "/some/process/that/wants/input" do |ch, success| abort "can't execute!" unless success io = TCPSocket.new(somewhere, port) io.extend(Net::SSH::BufferedIo) ssh.listen_to(io) ch.on_process do if io.available > 0 ch.send_data(io.read_available) end end ch.on_close do ssh.stop_listening_to(io) io.close end end end channel.wait
# File lib/net/ssh/connection/session.rb, line 404 404: def listen_to(io, &callback) 405: listeners[io] = callback 406: end
The main event loop. Calls process until process returns false. If a block is given, it is passed to process, otherwise a default proc is used that just returns true if there are any channels active (see busy?). The # wait parameter is also passed through to process (where it is interpreted as the maximum number of seconds to wait for IO.select to return).
# loop for as long as there are any channels active ssh.loop # loop for as long as there are any channels active, but make sure # the event loop runs at least once per 0.1 second ssh.loop(0.1) # loop until ctrl-C is pressed int_pressed = false trap("INT") { int_pressed = true } ssh.loop(0.1) { not int_pressed }
# File lib/net/ssh/connection/session.rb, line 158 158: def loop(wait=nil, &block) 159: running = block || Proc.new { busy? } 160: loop_forever { break unless process(wait, &running) } 161: end
The main event loop. Calls process until process returns false. If a block is given, it is passed to process, otherwise a default proc is used that just returns true if there are any channels active (see busy?). The # wait parameter is also passed through to process (where it is interpreted as the maximum number of seconds to wait for IO.select to return).
# loop for as long as there are any channels active ssh.loop # loop for as long as there are any channels active, but make sure # the event loop runs at least once per 0.1 second ssh.loop(0.1) # loop until ctrl-C is pressed int_pressed = false trap("INT") { int_pressed = true } ssh.loop(0.1) { not int_pressed }
# File lib/net/ssh/connection/session.rb, line 158 158: def loop(wait=nil, &block) 159: running = block || Proc.new { busy? } 160: loop_forever { break unless process(wait, &running) } 161: end
Registers a handler to be invoked when the server sends a global request of the given type. The callback receives the request data as the first parameter, and true/false as the second (indicating whether a response is required). If the callback sends the response, it should return :sent. Otherwise, if it returns true, REQUEST_SUCCESS will be sent, and if it returns false, REQUEST_FAILURE will be sent.
# File lib/net/ssh/connection/session.rb, line 440 440: def on_global_request(type, &block) 441: old, @on_global_request[type] = @on_global_request[type], block 442: old 443: end
Registers a handler to be invoked when the server sends a global request of the given type. The callback receives the request data as the first parameter, and true/false as the second (indicating whether a response is required). If the callback sends the response, it should return :sent. Otherwise, if it returns true, REQUEST_SUCCESS will be sent, and if it returns false, REQUEST_FAILURE will be sent.
# File lib/net/ssh/connection/session.rb, line 440 440: def on_global_request(type, &block) 441: old, @on_global_request[type] = @on_global_request[type], block 442: old 443: end
Registers a handler to be invoked when the server wants to open a channel on the client. The callback receives the connection object, the new channel object, and the packet itself as arguments, and should raise ChannelOpenFailed if it is unable to open the channel for some reason. Otherwise, the channel will be opened and a confirmation message sent to the server.
This is used by the Net::SSH::Service::Forward service to open a channel when a remote forwarded port receives a connection. However, you are welcome to register handlers for other channel types, as needed.
# File lib/net/ssh/connection/session.rb, line 430 430: def on_open_channel(type, &block) 431: channel_open_handlers[type] = block 432: end
Registers a handler to be invoked when the server wants to open a channel on the client. The callback receives the connection object, the new channel object, and the packet itself as arguments, and should raise ChannelOpenFailed if it is unable to open the channel for some reason. Otherwise, the channel will be opened and a confirmation message sent to the server.
This is used by the Net::SSH::Service::Forward service to open a channel when a remote forwarded port receives a connection. However, you are welcome to register handlers for other channel types, as needed.
# File lib/net/ssh/connection/session.rb, line 430 430: def on_open_channel(type, &block) 431: channel_open_handlers[type] = block 432: end
Requests that a new channel be opened. By default, the channel will be of type "session", but if you know what you‘re doing you can select any of the channel types supported by the SSH protocol. The extra parameters must be even in number and conform to the same format as described for Net::SSH::Buffer.from. If a callback is given, it will be invoked when the server confirms that the channel opened successfully. The sole parameter for the callback is the channel object itself.
In general, you‘ll use open_channel without any arguments; the only time you‘d want to set the channel type or pass additional initialization data is if you were implementing an SSH extension.
channel = ssh.open_channel do |ch| ch.exec "grep something /some/files" do |ch, success| ... end end channel.wait
# File lib/net/ssh/connection/session.rb, line 286 286: def open_channel(type="session", *extra, &on_confirm) 287: local_id = get_next_channel_id 288: channel = Channel.new(self, type, local_id, &on_confirm) 289: 290: msg = Buffer.from(:byte, CHANNEL_OPEN, :string, type, :long, local_id, 291: :long, channel.local_maximum_window_size, 292: :long, channel.local_maximum_packet_size, *extra) 293: send_message(msg) 294: 295: channels[local_id] = channel 296: end
Requests that a new channel be opened. By default, the channel will be of type "session", but if you know what you‘re doing you can select any of the channel types supported by the SSH protocol. The extra parameters must be even in number and conform to the same format as described for Net::SSH::Buffer.from. If a callback is given, it will be invoked when the server confirms that the channel opened successfully. The sole parameter for the callback is the channel object itself.
In general, you‘ll use open_channel without any arguments; the only time you‘d want to set the channel type or pass additional initialization data is if you were implementing an SSH extension.
channel = ssh.open_channel do |ch| ch.exec "grep something /some/files" do |ch, success| ... end end channel.wait
# File lib/net/ssh/connection/session.rb, line 286 286: def open_channel(type="session", *extra, &on_confirm) 287: local_id = get_next_channel_id 288: channel = Channel.new(self, type, local_id, &on_confirm) 289: 290: msg = Buffer.from(:byte, CHANNEL_OPEN, :string, type, :long, local_id, 291: :long, channel.local_maximum_window_size, 292: :long, channel.local_maximum_packet_size, *extra) 293: send_message(msg) 294: 295: channels[local_id] = channel 296: end
This is called internally as part of process. It loops over the given arrays of reader IO‘s and writer IO‘s, processing them as needed, and then calls Net::SSH::Transport::Session#rekey_as_needed to allow the transport layer to rekey. Then returns true.
# File lib/net/ssh/connection/session.rb, line 222 222: def postprocess(readers, writers) 223: Array(readers).each do |reader| 224: if listeners[reader] 225: listeners[reader].call(reader) 226: else 227: if reader.fill.zero? 228: reader.close 229: stop_listening_to(reader) 230: end 231: end 232: end 233: 234: Array(writers).each do |writer| 235: writer.send_pending 236: end 237: 238: transport.rekey_as_needed 239: 240: return true 241: end
This is called internally as part of process. It loops over the given arrays of reader IO‘s and writer IO‘s, processing them as needed, and then calls Net::SSH::Transport::Session#rekey_as_needed to allow the transport layer to rekey. Then returns true.
# File lib/net/ssh/connection/session.rb, line 222 222: def postprocess(readers, writers) 223: Array(readers).each do |reader| 224: if listeners[reader] 225: listeners[reader].call(reader) 226: else 227: if reader.fill.zero? 228: reader.close 229: stop_listening_to(reader) 230: end 231: end 232: end 233: 234: Array(writers).each do |writer| 235: writer.send_pending 236: end 237: 238: transport.rekey_as_needed 239: 240: return true 241: end
This is called internally as part of process. It dispatches any available incoming packets, and then runs Net::SSH::Connection::Channel#process for any active channels. If a block is given, it is invoked at the start of the method and again at the end, and if the block ever returns false, this method returns false. Otherwise, it returns true.
# File lib/net/ssh/connection/session.rb, line 210 210: def preprocess 211: return false if block_given? && !yield(self) 212: dispatch_incoming_packets 213: channels.each { |id, channel| channel.process unless channel.closing? } 214: return false if block_given? && !yield(self) 215: return true 216: end
This is called internally as part of process. It dispatches any available incoming packets, and then runs Net::SSH::Connection::Channel#process for any active channels. If a block is given, it is invoked at the start of the method and again at the end, and if the block ever returns false, this method returns false. Otherwise, it returns true.
# File lib/net/ssh/connection/session.rb, line 210 210: def preprocess 211: return false if block_given? && !yield(self) 212: dispatch_incoming_packets 213: channels.each { |id, channel| channel.process unless channel.closing? } 214: return false if block_given? && !yield(self) 215: return true 216: end
The core of the event loop. It processes a single iteration of the event loop. If a block is given, it should return false when the processing should abort, which causes process to return false. Otherwise, process returns true. The session itself is yielded to the block as its only argument.
If wait is nil (the default), this method will block until any of the monitored IO objects are ready to be read from or written to. If you want it to not block, you can pass 0, or you can pass any other numeric value to indicate that it should block for no more than that many seconds. Passing 0 is a good way to poll the connection, but if you do it too frequently it can make your CPU quite busy!
This will also cause all active channels to be processed once each (see Net::SSH::Connection::Channel#on_process).
# process multiple Net::SSH connections in parallel connections = [ Net::SSH.start("host1", ...), Net::SSH.start("host2", ...) ] connections.each do |ssh| ssh.exec "grep something /in/some/files" end condition = Proc.new { |s| s.busy? } loop do connections.delete_if { |ssh| !ssh.process(0.1, &condition) } break if connections.empty? end
# File lib/net/ssh/connection/session.rb, line 195 195: def process(wait=nil, &block) 196: return false unless preprocess(&block) 197: 198: r = listeners.keys 199: w = r.select { |w2| w2.respond_to?(:pending_write?) && w2.pending_write? } 200: readers, writers, = IO.select(r, w, nil, wait) 201: 202: postprocess(readers, writers) 203: end
The core of the event loop. It processes a single iteration of the event loop. If a block is given, it should return false when the processing should abort, which causes process to return false. Otherwise, process returns true. The session itself is yielded to the block as its only argument.
If wait is nil (the default), this method will block until any of the monitored IO objects are ready to be read from or written to. If you want it to not block, you can pass 0, or you can pass any other numeric value to indicate that it should block for no more than that many seconds. Passing 0 is a good way to poll the connection, but if you do it too frequently it can make your CPU quite busy!
This will also cause all active channels to be processed once each (see Net::SSH::Connection::Channel#on_process).
# process multiple Net::SSH connections in parallel connections = [ Net::SSH.start("host1", ...), Net::SSH.start("host2", ...) ] connections.each do |ssh| ssh.exec "grep something /in/some/files" end condition = Proc.new { |s| s.busy? } loop do connections.delete_if { |ssh| !ssh.process(0.1, &condition) } break if connections.empty? end
# File lib/net/ssh/connection/session.rb, line 195 195: def process(wait=nil, &block) 196: return false unless preprocess(&block) 197: 198: r = listeners.keys 199: w = r.select { |w2| w2.respond_to?(:pending_write?) && w2.pending_write? } 200: readers, writers, = IO.select(r, w, nil, wait) 201: 202: postprocess(readers, writers) 203: end
Send a global request of the given type. The extra parameters must be even in number, and conform to the same format as described for Net::SSH::Buffer.from. If a callback is not specified, the request will not require a response from the server, otherwise the server is required to respond and indicate whether the request was successful or not. This success or failure is indicated by the callback being invoked, with the first parameter being true or false (success, or failure), and the second being the packet itself.
Generally, Net::SSH will manage global requests that need to be sent (e.g. port forward requests and such are handled in the Net::SSH::Service::Forward class, for instance). However, there may be times when you need to send a global request that isn‘t explicitly handled by Net::SSH, and so this method is available to you.
ssh.send_global_request("keep-alive@openssh.com")
# File lib/net/ssh/connection/session.rb, line 259 259: def send_global_request(type, *extra, &callback) 260: info { "sending global request #{type}" } 261: msg = Buffer.from(:byte, GLOBAL_REQUEST, :string, type.to_s, :bool, !callback.nil?, *extra) 262: send_message(msg) 263: pending_requests << callback if callback 264: self 265: end
Send a global request of the given type. The extra parameters must be even in number, and conform to the same format as described for Net::SSH::Buffer.from. If a callback is not specified, the request will not require a response from the server, otherwise the server is required to respond and indicate whether the request was successful or not. This success or failure is indicated by the callback being invoked, with the first parameter being true or false (success, or failure), and the second being the packet itself.
Generally, Net::SSH will manage global requests that need to be sent (e.g. port forward requests and such are handled in the Net::SSH::Service::Forward class, for instance). However, there may be times when you need to send a global request that isn‘t explicitly handled by Net::SSH, and so this method is available to you.
ssh.send_global_request("keep-alive@openssh.com")
# File lib/net/ssh/connection/session.rb, line 259 259: def send_global_request(type, *extra, &callback) 260: info { "sending global request #{type}" } 261: msg = Buffer.from(:byte, GLOBAL_REQUEST, :string, type.to_s, :bool, !callback.nil?, *extra) 262: send_message(msg) 263: pending_requests << callback if callback 264: self 265: end
Enqueues a message to be sent to the server as soon as the socket is available for writing. Most programs will never need to call this, but if you are implementing an extension to the SSH protocol, or if you need to send a packet that Net::SSH does not directly support, you can use this to send it.
ssh.send_message(Buffer.from(:byte, REQUEST_SUCCESS).to_s)
# File lib/net/ssh/connection/session.rb, line 366 366: def send_message(message) 367: transport.enqueue_message(message) 368: end
Enqueues a message to be sent to the server as soon as the socket is available for writing. Most programs will never need to call this, but if you are implementing an extension to the SSH protocol, or if you need to send a packet that Net::SSH does not directly support, you can use this to send it.
ssh.send_message(Buffer.from(:byte, REQUEST_SUCCESS).to_s)
# File lib/net/ssh/connection/session.rb, line 366 366: def send_message(message) 367: transport.enqueue_message(message) 368: end
Performs a "hard" shutdown of the connection. In general, this should never be done, but it might be necessary (in a rescue clause, for instance, when the connection needs to close but you don‘t know the status of the underlying protocol‘s state).
# File lib/net/ssh/connection/session.rb, line 117 117: def shutdown! 118: transport.shutdown! 119: end
Performs a "hard" shutdown of the connection. In general, this should never be done, but it might be necessary (in a rescue clause, for instance, when the connection needs to close but you don‘t know the status of the underlying protocol‘s state).
# File lib/net/ssh/connection/session.rb, line 117 117: def shutdown! 118: transport.shutdown! 119: end