Connection hangs with debug logging enabled, causing socket.timeout errors

[awlodge]

Running RPyC tip code, the following code doesn't work:

import logging
from rpyc.utils.zerodeploy import DeployedServer
from plumbum.machines.paramiko_machine import ParamikoMachine
from paramiko import AutoAddPolicy

logging.basicConfig(level=logging.DEBUG)

p = ParamikoMachine("localhost", missing_host_policy=AutoAddPolicy())
s = DeployedServer(p)
r = s.classic_connect()

# Configure logging, so that remote logging events are correctly
# passed to the local logger to be handled.
rlogger = r.modules.logging.getLogger()
rlogger.parent=logging.getLogger()

# If this is set to a log level higher than DEBUG, the bug doesn't reproduce.
rlogger.setLevel(logging.DEBUG)
rlogger.addHandler(logging.NullHandler)

It hits a socket.timeout error after 30 seconds:

Traceback (most recent call last):
  File "rpyc_hang_repro.py", line 14, in <module>
    rlogger.addHandler(logging.NullHandler)
  File "/home/rpyc/rpyc/core/netref.py", line 148, in __getattribute__
    return syncreq(self, consts.HANDLE_GETATTR, name)
  File "/home/rpyc/rpyc/core/netref.py", line 71, in syncreq
    return conn.sync_request(handler, oid, *args)
  File "/home/rpyc/rpyc/core/protocol.py", line 467, in sync_request
    raise socket.timeout
socket.timeout

It seems to hang indefinitely, even manually setting SYNC_REQUEST_TIMEOUT in rpyc/core/protocol.py to five minutes hits the same error (after five minutes).

Digging into it, it looks like this is caused by 8c8481a - reverting this commit causes the bug to no longer reproduce.