Module: Paraspec::DrbHelpers

Included in:
Supervisor, Worker
Defined in:
lib/paraspec/drb_helpers.rb

Constant Summary collapse

WAIT_TIME = 
500

Instance Method Summary collapse

Instance Method Details

#master_client ⇒ Object

class TimeoutWrapper < BasicObject

  def initialize(target, timeout)
    @target, @timeout = target, timeout
  end

  def method_missing(m, *args)
    ::Timeout.timeout(@timeout) do
      @target.send(m, *args)
    end
  end

  def respond_to?(m, *args)
    super(m, *args) || @target.respond_to?(m, *args)
  end
end

# Connects to a DRb service and waits for the connection to start working.
#
# Interestingly, even when the remote end is up and running
# talking to it may fail the first time (or few times?)?
# Supervisor is able to invoke methods on master and subsequently
# when a worker connects to the master the DRb calls from worker fail.
# No idea why this is.
# Work around this by pinging and retrying each DRb connection
# prior to using it for real work.
#
# It appears that any DRb operation can also hang while producing
# no exceptions or output of any sort.
private def drb_connect(uri, timeout: true)
  start_time = Time.now
  Paraspec.logger.debug("#{ident} Connecting to DRb")
  remote = TimeoutWrapper.new(DRbObject.new_with_uri(uri), 2)
  Paraspec.logger.debug("#{ident} Waiting for DRb")
  begin
    # Assumes remote has a ping method
    remote.ping
  rescue DRb::DRbConnError, TypeError
    raise if timeout && Time.now - start_time > WAIT_TIME
    sleep 0.5
    Paraspec.logger.debug("#{ident} Retrying DRb ping")
    retry
  rescue Timeout::Error
    raise if timeout && Time.now - start_time > WAIT_TIME
    Paraspec.logger.debug("#{ident} Reconnecting to DRb")
    remote = TimeoutWrapper.new(DRbObject.new_with_uri(uri), 2)
    retry
  end
  remote
end



59
60
61
62
 
# File 'lib/paraspec/drb_helpers.rb', line 59

def master_client
  # TODO pass terminal option
  @master_client ||= MsgpackClient.new
end